I havent used an Iphone since the 3G, but Im pretty sure everybody I know, especially normies/non techies who use Iphones will use iphone/mac products for the remainder of their lives. Not because they love the product per se, but its the only thing they know how to use and have zero interest in learning to use anything else.
Who's going to displace Apple - who's even trying? The smart phone market is just a battle/division of two ecosystems: Apple vs Android.
Apple is as much a lifestyle/fashion brand as a technology one. It would take a lot more than a theoretical cool AI-centric pocket device from some wannabe competitor to displace them.
Apple is a luxury in some segments. But don't forget Macs are also computers of choice for majority of creative class and also lawyers. It's the same type of people who prefer or even legally require strong privacy. So buying top tier apple mac for running good enough private local model is exactly reason why would these people invest in such machine.
I don't think it's a winner-take-all market like search is. It's more like cars or computer chips. There will be a handful of more dominant players but plenty of competition and a multitude of large and small applications.
On the "echoes of dotcom" I think there is no way that Anthropic is worth $1.5T, same for SpaceX. Same for OpenAI. I agree that there will definitely be a Dotcom sized reckoning.
You cannot recreate mass production of cars quickly, it requires complex supply chain, many people, and large investments - look at Tesla as example. But you can absolutely create your own models, scale them up and offer services - capital you need for that is incomparably smaller, especially now as you can do that with AI assistance. I don't see how large software and AI companies can stay profitable in such environment.
I don't really know anything about IPOs, or the stock market, but it seems to me like there's an advantage to being the first to IPO. If this goes poorly, at least you're on the market. If this goes poorly for Anthropic, isn't OpenAI going to be in a bit of a pickle? It sets a bar for how highly OpenAI could be priced, and attracting new investors must become harder.
Nowadays there is no advantage to IPOing since private lending is so big, so the only reason is to cash out on retail investors when your company is over-hyped.
Well OpenAI just got a bit smaller private round than they wanted to, and there are plenty of news about frontier labs' bonds being junk rated.
So the point remain the same: maybe the IPO has to happen, and Anthropics' IPO performance matter for OpenAI, because all willing capital is tapped out.
An IPO is not a universally good thing. You sacrifice a significant amount of control once you go public. It's like a special form of bankruptcy if you look at it the right way.
Racing toward an IPO is more likely an indication of a liquidity crisis than about having superior product, prospects and roadmap. If things are going so well, why would you bring this amount of disruption to the company?
Pets.com generated $5.8 million in annual revenue while losing $61.8 million. Anthropic has already generated $33.2 billion in revenue in the first nine months of 2026, with thousands of large paying business customers. The dot-com bubble was heavily driven by speculation about future profits, whereas today's AI boom is backed by enormous real demand, established profitable technology companies, and measurable earnings growth. It's growing faster than any other industry or sector in history.
To be clear, I'm not bullish on the AI companies themselves. I think Chinese companies will continue to distill these models and once they achieve RSI (which is imminent), intelligence will be abundant and cheap. The real profit will be in the other layers of the AI ecosystem: electricity generation, chip fabrication (memory, storage, CPU/GPU), infrastructure (datacentres), and probably most notably, the people and companies which produce compelling software and services using this technology.
Claude got really good at ML Engineering recently. If they can't get the current models to RSI, they aren't trying.
How much juice is left in that tank after 5 years of intense focus on optimizing LLMs is another matter. If we are on top of a local architecture optimum, speeding up the hill-climbing will deliver underwhelming results until someone discovers the next hill.
That includes paper losses due to the increased value of their equity. Those are non-cash losses and not really indicative of how healthy the business is.
The weird thing is that 61.8 / 5.8 is around 10x the loss vs revenue, whereas 80 / 33.2 is only 2.4x more loss than revenue.
So there's actually a coherent argument that Anthropic isn't hemorrhaging money quite as badly as pets.com. But yes, I'm glad I'm not the only one who thought omitting Anthropic's loss this was a blazingly odd way to open an argument.
Demand is there, I just dont belive OpenAI and Anthropic can deliver in the long run. Not once token prices fall off a cliff.
Also, please ask Anthropic why their sec-1 filling is missing financial data after that quarter they made "profit". Most analysts belive that their profit was financial juggling and things went downhill really quickly after June.
Yeah, I think Anthropic would be an amazing company, were it not for the fact that, for whatever reason, all of the other LLM providers are only a few months behind them. Right now, I'm using Opus 5.5 as the main model, Longcat 2.5 for all subagents, and GLM 5.3 for adversarial security PR reviews.
Longcat is probably the worst of all Chinese models, but it's dirt cheap, and it works well enough. So my usage of Anthropic's models (and overall amount of money spent on AI) has decreased drastically.
> The dot-com bubble was heavily driven by speculation about future profits
So is Anthropic at their current valuation; they are gambling that they will become a multi trillion dollar company soon. That is speculation, as they are not profitable yet and everyone just hope that they will succeed.
> Anthropic has already generated $33.2 billion in revenue in the first nine months of 2026
Profit statement would sound nicer. But ofc, "Amazon was unprofitable when it IPOd too" and $28M losses on IPO year is comparable (not) after you adjust from inflation.
You shouldn't be downvoted. AI could be world-changing technology, like the railroads, flight, and the internet, and still leave a list of failed companies in its wake.
Airlines are a particularly interesting case since they enable so much economic value, but capture very little of it for themselves. On the AI side, the lack of moats and open models feels like this is the direction we may be trending as things settle out.
Airlines capture quite a lot, but through partnerships, loyalty programs, referral fees, and through partner networks. The same is going to happen with AI: inference is going to be cheap and plentyful, the money will have to come from the harness, the ads, the integrations... and Anthrophic doesn't have that worked out very well imho.
I’d argue Anthropic, with its push towards productization, probably has it better than the other AI companies with the integrations (with a lot of space for improvement, nevertheless)
Millions of people are using it in many different fields, and are constantly finding more uses. The only downside is that website developers must defend themselves against Anthropic.
> Millions of people are using it in many different fields, and are constantly finding more uses.
That doesn’t mean they’re positive for the world.
> The only downside is that website developers must defend themselves against Anthropic.
I’m pretty sure the parents of the teenagers who killed themselves, the communities who must deal with the noise, increased energy prices, and lack of water, the people who lost their jobs, the people who are feeling the environmental impacts (which will eventually include you, wherever you live), those who died due to (mis)targeted weapon strikes (and their families), artists all over, those with spouses who grew distant, those arrested by incorrect facial recognition, victims of deepfake nudes, and many many many more would disagree.
I don't know any of these laws since they haven't been relevant to me, but it's crazy that you can't just stop working immediately if your company stops paying you.
Don't know for Portugal explicitly but labor-employer relations have historically been quite different in Europe (and across Europe, since it's still different countries).
People "showing up" for their employers to weather storms wasn't uncommon in the 80s/90s before the stock markets and globalization created a larger disconnect between where profits went and thus also employees, them showing up could thus translate to bonuses,etc down the line (and probably that they had been getting them during "good times").
Some of it is still codified, in Sweden f.ex. there's a "Salary Guarantee" that the tax-agency will provide backpay to employees in case of bankrupcies (to combat possible fraud, the flipside of it is that even in LLC's C-level and boards of employers can become personally on the hook for such debts making them less risk-prone and more likely to file bankrupcies instead of trying to save companies).
So that's one way these different breaking points are, bankrupcy itself, but also the line where salary guarantee isn't given anymore (when you're just supposed to go since the state won't reimburse you further) and when people lose trust in any possibility of a rescue.
>before the stock markets and globalization created a larger disconnect between where profits went and thus also employees
There is a greater connection between profits and employee compensation in publicly listed companies on the stock market than there is for non publicly listed companies. All those RSUs that allow people to earn so much working at tech companies for the past 2 decades are because of the profits.
Not to mention that companies listed on the stock market have to publicize their financials, whereas employees of private companies don’t even know what the profit is.
In most countries, labor laws are generally a double-edged sword. If they are not able to leave without notice, that usually means they also can't be fired without advanced notice as well.
I had an employee in Czech Republic that didnt show up for 12 months. HR still trying to fire them. They collect full pay. I have 2 others in Slovakia with similar situations.
One company I worked for that I wont name, was so sick of its Italy site that they turned them all into project managers hoping they would quit out of frustration. Some did, many didn't and it just went down hill from there.
Neither company, my old or new, will hire out of Europe anymore. Its just not worth it.
> I had an employee in Czech Republic that didnt show up for 12 months. HR still trying to fire them. They collect full pay.
Sounds like a fantasy, or HR is protecting the employee. In Czechia, you can put employee on PIP (by delivering a written warning) and then you can proceed to termination when there is no improvement in a month or two. There might be some notice period, though (1-2 months).
BUT not showing up at work is a gross contract breach, so termination can be very quick (immediate). [§55(1)(b) - Czech Labour Code]
In my country (Kazakhstan) I think the only consequence would be a record in your employment book that would say that you violated part of your work agreement.
And then when you would search for new job, they might ask you about it
The Google of this era is a seedling in my care, but I'm just busy digging my technical moat deeper and deeper while i wait for the clowns to be sent packing.
> Anthropic and OpenAI are the AltaVista of the AI era. With first starter advantage but no real moat and very good at spending money.
Disagree. Google's search algorithm / variants were available for people to replicate. But there is only 1 google today. There are many office suites but people are still stuck to Microsoft Office -- many such examples.
While Anthropic and OpenAI have no technology moat people will become habituated to their user/api interfaces and rich features (which will take competitors a long slog to build).
Google came after AltaVista and Yahoo were already quite established. Google were quite late to the party and won on doing search much better than what came before them.
The reason why Google was #1 was they had a product that was ultrafast and without ads, so anybody who saw it switched instantly, there was no reason to look for another option. As for office suites, Microsoft won because of their aggressive tactics regarding format compatibility.
None of these apply to LLMs. I do enjoy cheap subscription to use frontier models, but if tomorrow I need to switch to the so-called open models, I won't regret it that much.
The lock-in you're missing here is that businesses use LLMs in conjunction with collaboration. They're building one-off apps that live inside the Claude GUI and sharing them with teammates.
To move that to an open model, you need to replicate the software surface (collaboration, functionality, permissions, etc.) and then provide a way to migrate the existing artifacts/data/history. Migrating developers is the easy version of the problem, migrating everyone else is potentially more difficult.
And then, it has to be trustworthy. Do I trust the provider with my sensitive business assets? If we are spending (say) $50k/yr on this, is it worth me risking my job by switching the company to a no-name brand to save $25k/yr?
The lock-in effect is stronger because so many engineers don't think it exists at all.
The moat is, as it's often been before, brand and trust. "No one ever got fired for choosing IBM" will become "No one ever got fired for choosing Claude Code"
I tend to agree that there's no moat in models, which is basically the entire business model for OAI and Ant today, but I do think there is still some money to be made there and it's likely they'll create some useful products around their models.
I don't think these companies are as bad as some dot-com companies, but I personally struggle with coming up with an investment thesis at the current price given it assumes they'll retain pricing power for models. And another problem is that this isn't like Amazon or Facebook where you build a decent product and with a few tweaks every couple of years people will continue to use them for decades – these companies need to develop and release new competitive models every month or two just to retain their market position.
I think the only thesis that makes sense is that the singularity happens, the private companies that owns the models which are produced as a result retain legal ownership of them, and that humans retain control over the models. In my mind I'd give this outcome a sub 1% probability.
They have a huge moat, unfortunately. The last models are steps above any other company out there.
The newer Chinese models get somewhat close, but OpenAI & Claude got better at stopping them from copying the model so the Chinese models will not make such big jumps anymore. And most Western companies will not use Chinese models if they are not open weight.
If they keep going like this huge parts knowledge work sector (legal, marketing, software, ads, health) economy is AI driven in 2-4 years and OpenAI & Claude will own it. Another huge shift of wealth from the Western World to Silicon Valey.
I agree, as a private user of Claude I don't find the newer models rather useful. But rather a better moat would be something like - a better deal (more credits, cheaper etc.)
Thats just silly. Yeah having a slightly worse model doesn't matter if its still capable. But the latest claude models are getting so good that they actually can run without human supervision in highly complex tasks. I haven't used them enough, but my guess is OpenAI is also getting close to that. For every test I run with Chinese models, it just randomly goes the wrong direction every now and then, making it untrustworthy. That is a huge difference.
It depends on what tasks end up producing the most revenue I suppose? Will it be these long running complex tasks, or a series of short tasks, where the difference is less apparent?
I had no idea what "Bagger" and "Cat" were at when reading the first paragraph and assumed they were hypothetical names of models meant to riff on how the names are just arbitrary words. I assume based on the later details in your comment that "Cat" is the company name (since I'm aware that all of the Caterpillar vehicles have "CAT" on them), so you've also indirectly provided some genuine commentary on the naming schemes (since at least the vehicle manufacturers have the decency to just use a numeric code after their company name instead of a bunch of arbitrary words that have no easy way to compare between reach one other than memorizing which one is which)
Having a good product doesn't mean you have a moat. Think of a car manufacturer. The only moat Ferrari really has is their branding, and maybe a few intangibles (e.g. patents).
Anthropic doesn't even have a brand worth protecting.
Is this the first time in history that every time someone finds a new valuable use for this technology, I can use that very same technology to copy the moat?
I agree with you that the two are way ahead of other companies (apart from google) with regards to model performance. but I doubt everybody will always need the smartest models - for most of the rudimentary daily work in the economic current models are already more than good enough. if openAI and anthropic can use their advantage to offer this rudimentary capability at much lower prices than other competitors then they have a real moat. but just having the smartest models by itself isn't gonna justify these valuations.
I have my own little app / ui that I work with and have all models, also round robin to test models, but GLM and Qwen just make weird mistakes & misunderstand the repo like Opus did last year when it first came out.
Where Fable 5.2 and Opus 5.5 are just always spot on. I dont need context, I write garbled prompts and it just gets me always. And it knows better and better, where it should push back & go into more detail and when it should just only do what I ask.
If Open weight models can reach Opus 5.5 level of consistency, I agree with you. If they can't I don't.
> but I doubt everybody will always need the smartest models - for most of the rudimentary daily work in the economic current models are already more than good enough
Yep. Actually, I do most of my coding work with GPT Luna. It's cheap and good enough.
I agree they have a moat. But it's not due to model outperformance as much as first-mover advantage turning into a distribution advantage.
Claude is a known brand and installed product for a decent fraction of the economy. To avoid churn, Anthropic doesn't need to stay way ahead of everyone else. They just can't fall behind. The latter is much easier than the former.
Note: I don't have a strong view on this IPO. Don't trade on it based on my comments.
My experience is that small businesses have defined computing needs and dislike tech changes. If they get something that works they don’t need progress.
On the one hand this will help the ai companies reduce their tech spending but on the other hand smaller specific models is all the customers need.
I don't know if people even need the best models anymore. Some people I've worked with have expressed a preference for older versions.
Things have gone very very fast. Two years ago, you couldn't reliably have AI write you a medium sized code project. It would spin in circles and apologize. Now, an old model can write it with only a little direction, and the programmer can have a reasonable preference against the newer model.
It seems to me open-weight models will put a floor on the pricing. They're slightly behind, but slightly behind last year meant nothing worked, now it means "just as good".
In general I think there's going to be diminishing returns on any given aspect of LLMs because not everything can be infinitely improved upon. If something can write a medium sized program near flawlessly, and I'm writing medium sized programs - then nothing else it can, or cannot, do matters.
Maybe bigger models will always just get progressively better at writing out bigger and bigger programs to the point that we're one-shotting fully functioning operating systems with wide driver support in the future, but again these features just don't matter if that's not what you want to do.
I think we're already limited by our own desires when it comes to coding.
The reason you can't one-shot something complicated is that you haven't generally considered all the important choices. You have to be asked for every choice, and complicated things have a lot of choices. But I think you can do something waterfall-esque, which used to be a big no-no in programming, if you just list out all the decisions meticulously.
I've had some success with doing that, just talking it over and thinking a lot about design decisions, making a document, and then letting it loose on the implementation. Inevitably, you find that there were contradictions or missing choices in the design, and you have to answer some questions.
So, not one-shotting, but it's my fault, not the AI's fault.
You can of course tell it to just fill in the gaps with whatever it thinks is reasonable, and then you'll have something potentially quite different from what you imagined.
The big improvement still coming down the pipe is that the models have better "taste" about filling in the things you didn't meticulously specify. A model from a year ago could make an amazing game, but you had to specify every detail precisely; now you can get similar quality results with a much more vague prompt.
I think LLMs can be improved much more but most likely significantly diminishing economic returns on the investment for that improvement. Ultimately, it becomes too expensive for a few % more improvement at increasing cost. I believe many of the LLM providers are approaching that problem. So it's time for them to commercialize viable long term business, maybe keeping a team still trying to dig up improvements without blowing the bank.
I agree - for the majority of people, there really is a hard limit on the technology. (Can it do what I want it to do in a reasonable amount of time with reasonable accuracy)
Many of my coworkers liked Opus 4.8, it was the moment they could actually use the models more reliably.
We're there, and moving forward doesn't really get us anything better. Most people are using them to format documents, spit out spreadsheets, pul together data points, and ask the models search questions.
We can already do that, it's only a matter of time before we realize we're paying 200/Mo for Jane in accounting to handle some of her work, when 5$ / Month would render the same benefit.
MOST people simply do not need frontier level intelligence.
The new models can pretty much run without human review. So yeah that flips the perspective: do we still need a human for that task, not which model can the human use.
I also expect AI to be more for machines and less for humans.
These providers are not selling chatbots that consumers can use as personal assistants anymore. Not even selling coding assistants. They want to manage the digital layer of every industry in the world.
Exactly. After horses, the Ford Model T [1] looked awesome. After a few decades, supercars started coming out. While it is nice to have a very fast Corvette, most people prefer to use a more basic SUV or sedan that is good enough for their day to day priorities.
>> They have a huge moat, unfortunately. The last models are steps above any other company out there.
A "huge moat" is not being incrementally better than one's current competition, not in tech. Anthropic's moat could be eliminated by one great model from any number of competitors, which means it isn't much of a moat. A proper "huge moat" would be a bundle of patents meaning nobody else was allowed to compete (apple/android) or a decades-old network to lock in users (windows). Anthropic has fanboys but most users would drop it in a hot second if a slightly better model appeared at a similar price.
> If they keep going like this huge parts knowledge work sector (legal, marketing, software, ads, health) economy is AI driven in 2-4 years and OpenAI & Claude will own it.
Exactly. If you haven't embedded AI in your workflows yet then it's just a matter of time or you're going to crash hard.
It's clear that it can do the work of several people, faster, cheaper and better and that will never not be the case.
They could triple the price (please don't) and I wouldn't flinch.
Again, I think people overestimate how much these models disrupt other jobs. It is good at code and technical writing, but in other fields I am not sure it is as helpful. And in those fields tripling the price would make it an absolute no.
No opposite, coding is at the forefront, everything where AI can verify if it's output is correct, and law is very easy, health also. Will all be as good as coding is now.
> They could triple the price (please don't) and I wouldn't flinch.
> They could triple the price ... and I wouldn't flinch.
> (please don't)
If you wouldn't flinch, why do you care if they take more of the froth? It's all worth it to you at 3x what it currently is, why object that they take what the market will bear?
> They have a huge moat, unfortunately. The last models are steps above any other company out there.
But "they" here is Anthropic and OAI together, and you probably need to add Google onto that list too, now that they appear to have got themselves back on track.
If it was just Anthropic, sure, this would be valid. But with 2-3 companies, we've got constant price competition bringing token costs down (as we are currently seeing) and no guarantee that Anthropic will ever come out with a clear advantage. If it's a three-way split, then that list price seems hard to justify.
And this is all assuming that the Chinese models don't work out how to get back to near-parity, which is also a massive gamble.
because remember good enough is better than "better".
my take Microsoft will be the bigger win, followed by Palantir due to Microsoft understanding how businesses small, medium & large enterprises use software. Microsoft now has its own dedicated FDE org - & for "a.i" to pay off you need a lot of FDEs - since the gains are all on the application layer. the models have become largely undifferentiated
so whether microsoft uses their own models or openai's - if their own models are good enough it doesn't matter they suck by a few percentage points against the frontier they will win.
Google is top of the benchmarks at the moment, and Meta is doing very well too, with this just being the first iteration of the Wang-era models. Grok is lagging a bit, but probably nothing money can't fix.
For many use case the models that already exist are more than good enough, if not entirely overkill to be replaced by cheaper alternatives like Jev.
> The last models are steps above any other company out there.
Google and Microsoft just released competitive models within the last month. Opus 4.8 was just released in May, and there’s probably 4+ non-Open AI/Anthropic models that have overtaken it. You can’t build a moat around a six month technical lead.
>The newer Chinese models get somewhat close, but OpenAI & Claude got better at stopping them from copying the model so the Chinese models will not make such big jumps anymore.
I missed that, do you mind giving more details about how it is harder now for Chinese model? What was put in place but more importantly which proof we have it actually is effective?
A moat would be something that would allow them to slow down. They can't do that. A moat is something that can't be replicated. The performance of their current models will be replicated at some point and sooner rather than later. They can't stop because I don't think they even know how long it will take them to get some better model.
What they have right now is enough money to to work through that. But they don't have enough revenue to keep this up forever and they can't stop either.
Their actual operating loss was ~$8B on $4.6B in revenue. Being skeptical of the valuation is totally fair, but throwing around the $42B/$102B headline figure without understanding the convertible debt accounting is disingenuous. It's the same story in 2026: ~90% ($94B) of their $104B nine-month net loss came from convertible financing charges, while revenue jumped to $33.2B.
Anthropic already turned an operating profit in Q2 2026, and it has the potential to be insanely profitable in the coming years. Calling it "entirely speculative" just isn't accurate.
$42billion-$8billion. That is $34billion in speculative capital expenditure on a business generating $4billion revenue and operating losses of $8billion.
It is overwhelmingly speculative.
Yes Q2 operating profit is great, but you cannot ignore speculative capital expenditure which is significantly larger than you'd traditionally expect even for growth stocks.
My recollection is the Q2 operating profit was due to SpaceX compute discount (which went away after Q2 and is why the IPO had to happen now, although that is unproven speculation on my part).
The bigger concern (and what I think you allude to) is the $901B in purchase and lease commitments.
These are 2025 numbers, which doesn’t take into account the ridiculous growth in revenue this year. Their net loss in 2026 was around 100B but subtract 80B from that because of the convertible notes “loss” (gaap accounting says that bc they sold shares to Google et. al. via convertible notes, and bc they rose in value, that counts as a loss to the books, even though it’s not actually money out of their accounts), and the fact they posted a 900M operating profit (!!) in q2 this year, and these numbers seem pretty reasonable? I mean why base your viewpoint on 2025 numbers when everything changed in 2026?
Operational profits don't negate capital erosion; because share price is essentially a measure of capital, those paper losses must be accounted for in the valuation.
But yes your point on 2026 revenue is a valid one, although the forward looking trend changes quite dramatically looking at their monthly ARR that they've been reporting.
You're living in a bubble. Not even my tech coworkers have a private Claude account. None of my immediate family uses anything other than Googles AI summary from searches.
Google IPOed after the custom from dot.com had mostly settled. At a massively more reasonable valuation as well. Antrophic is priced as if its guaranteed to grow at considerable rate and massively improve it's profitability in the near future. So it's not obvious there is much value remaining if you buy at the IPO price (Google was complete opposite in that sense)
Why is IPOing at a lower price better for the company? They could have fleeced a bunch of initial investors at the height of the bubble and had a huge mountain of cash to weather the downturn. Isn't that better?
ai it self is like search, good digital tool empowering bullshit jobs.
to really extracting more money, they have to be good at creating drugs and weapons, and monopoly the way of doing it. ant founders understand it, I think it won't take them too long to get there.
> Google had a better product, and then - some time later, mind - they discovered that ads work really well for them.
don't forget it was text ads too. Prior to that all ads were "punch the monkey" giant banners, images, and flashing text. Google put unobtrusive text ads on the right hand side of the search results page and users didn't mind. IMO that was the breakthrough that turned on the money hose, Google found a way to get ads in front of users in a least annoying way.
I was laid off for a horribly long time and was finally contacted by a friend at DoubleClick to come work for them -- at the time all those horrible ads you mentioned were their bread and butter and I wanted no part of it.
The same day I got an offer to work at Genentech in their IT support and chose that as it was a real job at a real company.
2 years later I had to move on because corporate politics suck and my friend was now working at Google via the acquisition. He's been retired for many years and I'm still grinding away. At least I get to grind...
> Google had a better product, and then - some time later, mind - they discovered that ads work really well for them.
> OpenAI and Anthropic have a good product and they (especially Anthropic) are increasingly good at generating revenue from it.
> What's missing, in your opinion?
Google had a product that was miles better. If you were using Altavista or Yahoo to find something, chances were you'd never find it even though it existed. With google, you almost certainly found it, sometimes even on the first go ("I'm feeling lucky").
Altavista and Yahoo tried to curate the web for users. The wanted to be the internet for users, they had guardrails! (Sound familiar?)
Google went in the opposite direction - they tried to get you what you wanted and ignored things like guardrails (Still sounding familiar to me).
I see the analogy, with ANT and OAI wanting to curate the AI experience for users, and preventing their users from doing anything that goes against the companies' opinions. The open-weight models are roughly similar to year-1 Google: curate nothing, don't get involved, and don't push your opinion on your users.
Maybe ANT and AOI will change their tune before either of them manage to lock up AI into a monopoly, but I doubt it. They aspire to be like Google: get the monopoly and then abuse the fuck out of your users for profit. Then abuse the fuck out of your paying customers for profit.
> Do you remember companies like AltaVista or maybe pets.com?
Compaq sold "83% of its AltaVista Internet search engine to CMGI Inc" for $2.3bn [1]. Four months later, ahead of its planned IPO, it disclosed $54mm in revenues [2]. 43x.
Pets.com IPO'd with $5.2mm of sales [3] and a $290mm valuation [4]. 56x.
Anthropic is going public at 40x trailing revenue (AltaVista territory) and 22x Q3 annualized revenue (not comparable).
> every time I question these crazy valuations I get down voted really quickly
You're the top comment. And this is against the guidelines: "Please don't comment about the voting on comments. It never does any good, and it makes boring reading" [5].
That’s true; public sentiment about AI is already trending downward, and I suspect we’ll reach a point where these companies start experiencing steady negative growth, once people realize the tech is only ever going to be great for certain specific use cases (e.g. coding). So Anthropic seems equal to or worse than AltaVista from that perspective.
I don’t know, every professional lawyer, doctor, accountant, nurse etc is using AI now in some way. Every person under the age of 60 is using it.
It’s going to be embedded in our professional class and we will have to see who wins. I agree there isn’t as much of a moat for any specific model here, it’s really the last mile who will capture the user/integrations.
Yes, I know a few myself. LLMs are useful but the huge distruption has to be slowed and thought out. These companies are pedal to the metal, they don’t care about anything as if laws don’t apply to them.
And we should still keep our people educated, now people are wondering if education is worthless since the LLMs took the limelight. This can’t lead us to a good place. Profit is driving everything.
the best thing antrhopic/openai could do would be to get rid of their free tier and focus on enterprise/b2b.
it'll cut down the school cheating and the laziest, highest-visiblity slop on social media that is driving so much negative sentiment. AND they dont have to worry about providing capacity for people that ndo not (and never will) sign up for a subscription.
Don't worry it's coming. This is the same play that every other start-up uses, get them hooked on it while it's cheap/free, start charging them once they can't live without it.
Yeah, I hear Google is putting the finishing touches on finally charging for search after 28 years.
EDIT since I can't reply: the whole "if you're not paying, you're the product, not the user" is of course true, but totally orthogonal to your point about getting them hooked on free and then starting to charge. And ChatGPT already has ads, so your original point really makes no sense even in this context.
Anecdote: I produce code for a living, and I use AI for that task. I don't generally use it for anything else -- I can count maybe a handful of times that I've used it for something other than my job or personal coding projects.
So when I retire, or if I were to switch careers, I don't think it's necessarily true that I would use it at all, let alone daily.
They use it like Google, just kinda there because it's cheap and easy. Not all in omg I need a max plan like some devs.
Can the AI companies be as successful as Google ended if with web? Maybe, just feels like the ad model will not work with these use cases like it did with Google 25 years ago..
I don't think this is true, or is a reflection of being in tech (assumption on my part about you, correct me if wrong).
Anecdotally, my peers (26ish, so gen z), none of them use AI except me, and that's for work as a SWE. My 68 year old step-dad uses it, but I suspect that's out of boredom more than anything.
I know students are using it for tons to essentially cheat, but I don't know anyone in that cohort using it for personal use of any kind.
> I don’t know, every professional lawyer, doctor, accountant, nurse etc is using AI now in some way. Every person under the age of 60 is using it.
This is obviously false on its face. For example, my household consists of four people all under the age of 60, and only one of them (me) uses AI at all.
And I don't believe my accountant uses AI, at least not for anything of importance. If I found out they were, I would fire them immediately.
Well, as a lawyer, for the past several years, people, especially here, insisted that I'd just be out of a job entirely because of AI. Now AI providers come to my office to beg me to sign up for their products, which apparently cannot replace me. The idea that I need cutting edge AI as opposed to a more specifically well-trained model is also just as silly. Harvey is the company that will benefit from law firms, not Anthropic and not OpenAI
Public sentiment against data centers is definitely trending downward, public sentiment against AI is a little trickier. A lot of the anti-slop crusaders are being forced to admit that AI output is getting pretty good, and it's slowly making inroads in more industries and roles. I think there's a begrudging acceptance from haters that generative AI is just how things are going to be from now on.
I think they burned more than 9 billion, they had a loss of 8B last year and 3B the year before[0], presumably they had a few more in the previous 3 years?
Still, a valid point, but one should trace the increase in burn rate next to the increase in revenue.
Look, if you find that line of reasoning convincing, then buy all the shares you can after they go public. If you’re right, and anthropic is the cash volcano that people insist it is, then a negative sentiment around its IPO does nothing but increase your ROI.
It's incredibly cash efficient, but they've spent compute credits that they've raised in previous funding rounds (that was a great deal for their investors)
The problem with growth as a metric is that you can more-or-less linearly trade cash for growth. This is the same game that Uber and WeWork played, giving away product below market for years, in a straight trade of investor cash for growth
Despite uber being focused on the physical world (moving people or food from A to B) and Anthropic being focused on the digital (mostly text and images), somehow Uber is much more capital efficient because they don’t have to own the cars, employ the drivers, insure everything, etc.
Anthropic has far more capital expenditure because of their datacenters, hardware, etc
The CMGI boards on The Motley Fool were a wild time. The bears had all the facts and logic on their side, and the bulls had "Oh yeah, well then why does my CMGI keep going up? PPPPPPPTTTHHHH!"
The 'AI company' Allbirds is 11 years old... Putting out ages of business as if it reflects the maturity of the market or business is a little silly. AI harnesses as a general term basically emerged this year (open claw was first published in nov 2025). ChatGPT only released nov 2022 claiming ages longer than that reflects any form of stability seems beyond generous.
The better point is those companies listed never had a positive revenue before (or after) their IPO while Anthropic has had at least a few profitable quarters (though you'd really like a fiscal year to rule out the more obvious shenanigans). Though the bigger questions remain around the broader market and what that means for income and depreciation (amusingly because of the supply crunch the 'wildly optimistic' deprecation assumptions around GPUs that AI companies have been making have held true (technically you could argue they're appreciating assets, though that won't be the case if data centers start getting liquidated).)
The difference is that Anthropic and OpenAI are likely to continue being useful, profitable companies even if initial investors lose out. They’re going to be operationally profitable for the foreseeable, like obviously so.
> With first starter advantage but no real moat and very good at spending money.
They do have moat. Every single company is pushing their private data into these models. Do you really believe that data isn't making into the training data one way or another?
There's a popular argument that every new technology has to be a bubble because something something pets dot com. Because it is destined. Because it is meant to be.
No, it doesn't matter that whatever the revenues are. Bubble because pets dot com. That's all the logic.
> Edit: every time I question these crazy valuations I get down voted really quickly.
You've gotta understand there's a lot of people on HN with their wealth tied to the success of AI at this point. Every downvote on a post like this is just a lowly engineer that wants to keep having steak dinners.
I remember a whole bunch of companies like pets.com.
But I also remember the clear "leaders" at that time: Amazon, eBay, Yahoo!, Cisco, Akamai.
The rest felt like coat-tail riders. Obviously, they didn't all deliver stellar returns, but, in aggregate, these 'blue-chip' dot coms did quite well (I'd say 3/5 did well).
I also remember wanting to invest in Amazon, but once they finally went public, I thought, "well, too rich for my blood!"
Anthropic "feels" like one of those.
Mind you, I'm not investing in ANT, except through my mutual funds, but drawing a parallel between Anthropic and pets.com is ridiculous.
edits: You can kind of throw Google in there, too, but they were a little bit later, but it was also clear in the early 2000s that they were a force to be reckoned with.
...Amazon was still only selling books in y2k. eBay was still small taters as well. Yahoo! hadn't been bought by Apollo yet so they were still smallish but had a good search engine and sturdy webmail
ecommerce did not take off until after 2009 and certainly got magnitudes bigger over the past 8 years
you probably are being hyperbolic about these choices as leaders back then.
Cisco, AKAM: legit and still around. Think also HP/Compaq, Motorola, VZ, Sprint (RIP), ATT.
Fiber build out was the infrastructure side. Servers and switching equip as well. Consumers increasingly wanted gaming computers, wherein NVIDIA began as a IDE & PCI video card developer, which eventually became ...well...
Amazon was selling more than books long before 2000.
I remember the stock price would skyrocket every time they added a new "tab" on their front page. At first it was books, but in the late '90s they added housewares and other stuff -- each of those categories got a new tab on their front page.
I wasn't being hyperbolic. Those (in my opinion at the time) were the big dogs.
All the other dotcoms felt like me-too companies. Do you remember uBID (trashy version of eBay)? Lots of trash, only a handful of leaders.
Ok. I stand corrected about amazon...call it sour grapes or selective memory because...i wasn't a believer. If Only...
heres an LA Times piece from that year: latimes.com/archives/la-xpm-2000-jul-27-fi-59951-story.html
Thanks for the headsup. It improves me.
That image link works for me, so maybe a server problem blocking your viewing or maybe you need to prepend a https slash slash...its a pic of NVIDIA products since 1995.
I was done with Yahoo when I started using AltaVista.
This is total revisionist history from my perspective.
From my perspective, Amazon was a massively over valued stock for an online book store and the "obvious" winner when the dust settles would be AltaVista.
I think it's more so the valuation than the actual prospects of the company. Google is not analogous to Anthropic because its valuation was massively lower so there was a lot more growth left. Anthropic is priced as if it already won and managed to solve its profitability problems.
Google IPO'ed at 7.4x P/S and was already technically profitable. It would be more similar if they initial stock cost the same as in 2012+. Also Google grew by over 100x since then. If Antrhopic grew by 100x it's market cap would be 20% higher than the entire global GDP.
Agree with everything you say, which is also why I wouldn't invest in ANT today. I mean, how much can their market cap possibly grow from here? 5x? If they're reallllly lucky?
edit: my main point was that I don't think ANT is going away any time soon. They're not pets.com
I love Claude. Best bang for your buck! For how long? Testing local LLMs again recently, same hardware worlds apart from months ago.
These Cloud AI are just itching for regulatory capture and I think it's too late, the open weights are catch up faster than they can get regulations to happen.
Whatever bubble pop happens, will be interesting. Dot con caused a lot more people to get jobs.
This was the opposite, so will the burst cause hiring sprees? Or will the burst crumble the economy?
The bubble is similar to dotcom, but the bursting consequence to me seems potentially drastically different fallouts.
Anyone else find it suspicious that they are going above their recent evaluation just as I don't know everything involving the Internet is falling apart?
Hundreds of outages, many of them critical infrastructure in the past week. Also Nvidia
I can't tell if the game is get the money for share holders and pop. Or, make some announcement about how only anthropic can be trusted and some new bills drop regulating away it's competitors. Either way I think it's bad news for the average person.
I should probably pull my 401k from the market just in case...
A useful analogy would be the IPO of the European unicorn World Online in 2000. It was oversubscribed by 20 times, didn't rise on the first day and only fell afterwards.
Later it turned out that insiders like Brink had sold their shares before the IPO. Soon after the bubble burst completely.
Anthropic is only six times oversubscribed. A we see this does not at all mean that the price is going to rise above the IPO price. My guess is that the Anthropic IPO is the new World Online, full of scams and the precursor to the bubble bursting completely (which is why they rushed the IPO).
VA Linux Systems went public on December 9, 1999, under the ticker LNUX, pricing at $30 per share and surging roughly 700% on its first day of Nasdaq trading.
They were fine machines but they were just another PC builder and got their stock fell way down and turned into Sourceforge and then got bought by Gamestop.
Eh, options will be available from Monday. Get an option or a derivative if shorting isn't your thing. Plenty of ways to bet on the value going down over whatever term you think it's going to decrease over.
I've put down a bet on ANT, and I'm not even all that sure it'll go up instead of down. I don't think it's ridiculous at all to bet even a small amount on a thing you're pretty certain of.
> Edit: every time I question these crazy valuations I get down voted really quickly.
This is a problem of hackernews. I also made critical comments in another thread and got insta-downvoted. This may even come from bots, it is quite strange. That has not happened on old.reddit.com, for instance. Something is not working well on hackernews in general here. See also the speed limitation, after 5 comments I can not comment for some hours. That's just tedious ...
given the trend you cite, i am surprised the critical content on this thread has held up as well as it has. pets.com comes up and there's a lot of mega-reverse false equivocation. like the analogy is deemed wrong because it is judged on the terms of Borges' On Exactitude in Science, but someone brings up something very wrong in defense, and time is spend arguing over that. the original point is lost in paper millions at stake, and my children need wine!
I used to think they didn't have a moat. But that was when it was just us developers using the Claude CLI to write code. I'm starting to see Claude become more and more engrained in other parts of business, like people sharing Claude docs/artifacts where they used to email out word docs. Or even finance people using the Claude plugin in Excel even though Copilot is available.
At this point, it's like saying that Microsoft has no moat because anyone can write an OS or a word processor. The moat is the friction of switching to a different platform. That alone won't guarantee success in the long term, but they won't be going anywhere any time soon. Even WordPerfect managed to stick around a for a few years after Office came along.
why would that be a moat? if software is "free" like everyone is claiming, I can make a plugin for whatever LLM in excel. I could even make software that ports what was previously done with claude.
>I am still waiting for the Google of this era to show its face
I believe Google is the Google of this era.
Going back a decade to when the hype around self-driving cars was peaking, you had several investor driven companies like Uber, GM's Cruise, and Uber all behaving like cowboys. They were burning through massive cash piles but under unrealistic amounts of pressure to deliver results to nervous and impatient investors, and so they were heavily incentivized to make it appear as though they were further ahead than they were, they behaved recklessly, and they lost.
Waymo on the other hand was effectively patronized by it's parent. They kept their heads down. They played by the rules. They didn't go out of their way to draw unnecessary scrutiny. Waymo won, it does not have any real domestic competitors. All this serves as a model, or a microcosm of what is happening at a much larger scale with the AI race.
It's tortoise vs. Hare. I'm betting on the tortoise and the tortoise is Google. They don't need to win or get way out in front. They merely need to quietly answer every move their competitors make, which they've so far been doing. And of course, they do not rely on Nvidia for anything.
I think it may end up actually being Google or SpaceX. The companies who are competitive in this space but also have other significant revenue streams are the likely long term winners.
Google overtook AltaVista because of a fundamental technical advantage: PageRank. It was vastly better at natural language searches than anything that came before it. That momentum allowed it to build a moat: because it had so much data on what users ultimately clicked on, it could refine its search results better than others.
If there is such a durable technical advantage out there, it remains to be seen. Open weight models are catching up to where the frontier models where just six months ago. Contrast that to how shitty Bing search results were compared to Google even years later.
II'm super sceptical of this IPO. However I don't see how you can compare ANT to altvista or pets.com. Altvista didn't have unique tech even in it's day, and pets.com is well, pets.com. Neither have technology that is rapidly replacing all white collar work at an incredible pace. You'd need a make a much more relevant analogy than simply high IPO price. From product, to position, to team structure, to how deep interconnection is has with the top 10 companies in the world
> every time I question these crazy valuations I get down voted really quickly.
It’s not that you questioned the valuations, it’s that your comparison doesn’t make sense. Google won because it built a better product, not because of a fundamental difference in search or business concept. There was no real moat until they acquired DoubleClick, long after they beat out AltaVista.
It's a bit more nuanced: Google won because it was printing cash from day one, and the amount of cash they were printing kept increasing at a phenomenal pace. Neither OpenAI nor Anthropic are like that.
People are saying that Anthropic has no moat which sounds right but at the same time they have left everyone behind: Google is trying lately but seemed very lost over the last years despite inventing the thing; OpenAI cannot match Opus 5.5 and Fable atm; and it seems like anti-distillation efforts really did put a dent in the speed of Chinese model development.
Meanwhile Claude is getting really good. It's a decent harness now, even the remote control over a couple servers is quite OK. They're adding artifacts left and right, new features. And all without making the product seem bloated, they stay focused.
They hide reasoning from... well, you and me, the Claude users. I thought it was going to be annoying but don't really watch these threads while they're going anymore and the result is of no lower quality than when the reasoning traces were visible so it's not that big of a loss.
Their IP, which was trained on text and code from a gazillion people who never gave permission to do so.
I liked the phase where open models [cw]ould be trained on thinking traces, because it at least gave something back to all the people who's work was stolen and whose jobs Anthropic wants to destroy.
It is really strange to see people rooting for companies that want to lock up all the world's knowledge and intelligence.
We need open models to win, so that everybody has access to the means of production, not just some multi-billionaires.
Yes, it's not the thing supposedly anymore. And there are reasons to believe that because if you do watch it closely you can see that quite a bit of time the answer doesn't seem to follow the reasoning with reasoning being quite shallow or even in the wrong direction.
I never trusted the "chain of thought" to be more than a verbose response. Its a black box, all we really know is they started telling it "show your work" and the model would spit out a bunch of text that sounds reasonable enough before giving the actual answer at the end.
The question is for how long? Maybe OpenAI doesn’t have it right now, but won’t they have it in a couple of months?
That’s been the recurring theme.
Anthropic has had a noticeable advantage before, and others did catch up and momentarily surpassed it.
I don't know why but every time I have tried to use Anthropic models I've been disappointed. From the OpenAI stable, I use GPT 5.6 Luna and only switch to Sol 6.1 if it can't figure something out. When I don't need to use OpenAI models, I use Deepseek v4.1 Flash 0731. Every now and again I chuck a few tokens Anthropic's way but the models just chew through the $$ and never seem to be like "wow! so AI! very code!" to the extent that they appear to be worth the cost. MMMV I guess.
> I am not a big AGI believer but business- and product-wise it seems like ANT has something that others are missing.
Wonder what that could be, not profit, that's for sure.
Their operating losses were constantly above $500 million in every quarter for the last year, and here I've though that business is supposed to produce profit...
More alarming, that there's no plan whatsoever how to turn AI profitable. Speaking and spying labubu figurines won't cut it, neither will ads.
One of the things driving the cost is having to keep training new models. It was every 6 months or so, now it seems like every 2-3 months. If they stop the other companies will catch up, so they are forced to pace (or race) the frontier to make sure they always have something better to offer if/when another company catches up.
That requires the hardware (GPUs/TPUs) and data centres to run them. It's not clear how many training runs they do, for how long, or at what cost.
In terms of profitability they need to get people to spend on tokens and subscriptions, and those need to be run cost effectively. Any subsidisation cuts into their profits.
The current boost has likely been due to tokenmaxxing, agentic coding, and convincing companies and influencers to spend money on adopting that. However, several companies have been scaling back their token spend and it is unclear how much more they can increase in this area. This is likely why they are pushing into other areas like mathematics and more recently genetics -- they need new customers to keep and increase the token spending.
> One of the things driving the cost is having to keep training new models. It was every 6 months or so, now it seems like every 2-3 months. If they stop the other companies will catch up, so they are forced to pace (or race) the frontier to make sure they always have something better to offer if/when another company catches up.
Exactly. Plus new models need more and more training data, so it will just keep costing more and more, there's no end to it (except when they run out of data, which is likely too and soon).
> This is likely why they are pushing into other areas like mathematics and more recently genetics -- they need new customers to keep and increase the token spending.
Yes, but there are not much new areas left to invade, they are running out of possibilities pretty quickly. There's only so much areas, and they have already covered most of it.
All in all, their costs are keep increasing while their chance to continue tokenmaxxing and grow income is drastically decreasing. Not a good sign.
I'm not saying that everything is great at ANT, but looking at last year with a company with this type of growth is a waste of time. More recently[1]:
Total operating expenses rose to $17.5 billion from $10.6 billion in the second quarter. This resulted in Anthropic reporting an operating loss of $567 million, after reporting a $974 million operating profit the quarter before. The company attributes that loss primarily to compute bought under short-term commitments, which noted in the filing “currently costs significantly more” than compute secured under its multi-year deals. The company also cites $2 billion in non-cash charges tied to an employee donation matching program.
OpenAI is absolutely crushing Anthropic regarding cost of running models.
A lot of real world scenarios have already been completely solved with models released months ago. My clients don't need more powerful models anymore. They need faster and more affordable models.
OpenAI has been aggressively pushing on both fronts. Where is Anthropic's 750T/s wafer-scale compute offering? Where is this on their roadmap?
Not going to speculate on how they'll do vs. OpenAI, but to your point on a lot of problems now effectively being "solved", I tend to agree.
I have one recent use case now where using Claude to review it, it first wanted to use Fable, got all excited about possible savings from lower defect rates, had to backtrack and admit that Fable provided no benefit whatsoever to justify the extra cost over Opus after it ran evals (in fact, arguably the quality was worse, but that part was a subjective measurement).
There's a mounting number of tasks like this where the value is going to be falling costs whether by switching providers, or moving down their model ladder as the one below gets good enough.
Given Opus 5.5 is now good enough for the task in my example, it means we have every incentive to start evaluating cheaper models now and again, instead of waiting eagerly for new, smarter models.
I agree, I'm on 6.2 Sol and it feels blazingly fast and it leads to a better UX. It seems to be capable enough to handle 99,5% of stuff that will be thrown at it.
It's pretty obvious that OpenAI is dieselgating* the benchmarks. Even 6.2 Astra is nowhere near as good as Claude models for coding. Any software engineer can do a side-by-side comparison (as many of us do) and figure it out.
*Also consider the possibility that Anthropic may be dieselgating it down ("benchminning"?) instead of OpenAI doing benchmaxxing. It might make sense from some game theory standpoint - If you are significantly better than your enemy, it might be in your interest to lull them into believing that you are only slightly better.
> Also consider the possibility that Anthropic may be dieselgating it down ("benchminning"?) instead of OpenAI doing benchmaxxing. It might make sense from some game theory standpoint - If you are significantly better than your enemy, it might be in your interest to lull them into believing that you are only slightly better.
Unlikely, because it would mean that you're also "lulling" your customers and your investors into believing that you're only slightly better than your competitor ("enemy" isn't the right term here), and you don't want that. Especially not shortly before your IPO.
I do coding evals every time a new model comes out (that I have access to) and Opus 5.5 is a touch better than Sol 6.2 right now but that’s a recent development. Prior to the 5.5 release open ai outperformed at every tier.
Sol is half the price of Opus and Luna outperforms haiku at every eval I throw at it. Haiku wasn’t even able to complete several of my evals until the newest release.
The idea that Anthropic is miles ahead of OpenAI doesn’t pass the smell test in my usage.
And that’s before the Claude vs Codex comparison (which I’ll grant is more personal preference but I find Claude to be wretched).
> Any software engineer can do a side-by-side comparison (as many of us do) and figure it out.
That's a really strange take to me, because among my colleagues theirs basically a split between two kinds of people:
- the “I don't need frontier so I use sonnet and it's good enough” group.
- and the “I want the best model and I managed to get a Codex subscription from management because Sol is much better than Opus” team. (This only started this summer and I suspect it was mostly true because of Opus 5, but that's still going to do long term damage).
No one is arguing the other way around and I know the manager would be considering a switch of what comes as default to full OpenAI on the first occasion (he's a very cost sensitive kind of guy so I'm sure that any price hike would trigger him).
That's not obvious at all to me. While I prefer Opus for run-of-the-mill tasks I actually find Astra to be better for very hard problems. It's just a bit of a psychopath and has more "paperclip maximizer" tendencies. If my hypothesis is true then how they fare head-to-head depends on the workload.
Codex doesn't have Claudish, though, and you can use other harnesses (i.e. Pi with its immense tokens savings) with a coding plan.
I use Codex for lack of Claudish _alone._ My spidai senses might still be overtuned, but I can spot Claude generated content a mile away and my brain involuntarily turns off.
>My clients don't need more powerful models anymore. They need faster and more affordable models.
Honestly this sounds like it calls more for Deepseek v4.1 flash than OpenAI's models, although those charts from earlier this week did look impressive. Trying to compete on price efficiency sounds much more daunting than competing at frontier capabilities tbh.
Anecdata: on my personal projects I vastly prefer to use Opus, and even reach for it to realign Sol once it's started to drift on specific projects instead of bumping up to Astra.
On the other hand, at work, where I have to justify paying API prices, I almost exclusively use Sol, while occasionally bringing in Luna (though I've also been playing with the MS models since yesterday; not impressive, but very cheap).
If they do manage to catch ANT in ability while keeping costs this much lower, they could still leapfrog again. I'm not sure I'd bet on it being soon after seeing Astra 6.2, but I don't think the race is over.
They don’t get so much attention on HN, where the focus is naturally on coding, but the new Claude features being steadily rolled out for office work could, it seems to me, become a real moat. Anthropic has lately announced integrations of Claude into whiteboards, online meetings, etc. As people get used to using those and to having Claude as a collaborator who is knowledgeable about their work, they will become reluctant to switch to other providers.
But we already get this with copilot today, its integrated into sharepoint, teams, outlook, word, powerpoint etc. Is it a bit rough around the edges... yes, but you can finally find documents in sharepoint, where before that basically did not exist. We already have contracts with microsoft in regards to data protection, and Anthropic has not really done anything to prove that they can be trusted with it yet.
We'll also get this with tool use and browser automation. The models are increasingly good enough that they can build custom integration tailored to your specific use case. It'll be very hard to maintain a lead there when your "lead" is functionality the models can write unilaterally on the whim of a user.
This is exactly what they should be doing, yet I'm still baffled by their limitations. I can only connect one email account or one calendar to my Claude. What if I have a personal life and a job? Sorry, can't do it.
ChatGPT doesn't have this issue. It doesn't seem like it's particularly complicated.
Your theory hinges on the idea that Claude users will subscribe to the idea of it being a knowledgeable collaborator and have that work for them.
I'm not saying that's impossible, but it's entirely speculative at this point. Developers are far more effective with these tools because they understand their strengths, limitations, and how to use them to get good results.
I've yet to see anyone benefit from treating the tool as a personified collaborator. They either scale back their use or fall flat on their face when it behaves unexpectedly.
I'm using M365 through MCP/CLI via some GraphQL interface, with access to Teams, documents, Outlook, etc. None of that depends on Anthropic.
I have non-dev clients and business partners that are all coding & mailing and doing everything with AI. Some I helped, but some are just handy. Next year goign to be insane.
That's assuming you're letting these integrations near your data. I recently had a look at the Claude plugin for Chrome and it honestly scares me a bit. This thing is a security nightmare (plus, probably a walking GDPR violation in the EU). They have solved neither the lethal trifecta nor prompt injections nor even unintentional misbehavior; yes, they have some mitigations and safeguards in place, but these cannot be remotely considered guarantees.
Lets assume the best Chinese AI lab is an entire year behind. These companies are valued below 20B, while Anthropic is valued above 1T.
Assume nobody ever catches up to Anthropic and everyone will always be 1 year behind. Is the 50x valuation still reasonable? Is the money really in doing PhD level math, or simpler stuff like summerizing coporate emails?
And lets be honest, some US and Chineese labs are only months behind and the only reason they have not fully caught up is limited compute. For China this could change quickly (e.g. Huawei gets their yield issues under control and suddenly Chinese labs have cheaper compute than west). People who invested in Anthropic at 50x valuation must belive this scenario is impossible in the next 5-10 years, but that is a huge bet.
My view is that Anthropic will be the Microsoft of the AI era. The only moat Microsoft has is that it's the default for many enterprise companies. AI needs access to a lot of confidential data to be useful in a large business. If Anthropic convinces enough customers to use Claude Meet, Claude Docs and Claude Sheets they win. Not because their models are so much better, but because all your business data is with them.
It’s not going to happen at an enterprise level, Microsoft is so intertwined within every part of any large risk-averse organisation that copilot will be the default.
Do we have any numbers on who's actually using those ancillary products? All the non-tech folks I know don't touch them. If they want Claude in their docs, they'll just use the plugin for Google Docs or Word.
I think Google Sheets,etc is just too entrenched, I'm at a small shop of heavy Claude Code users but I haven't heard a whisper of even trying those other products.
We're moving into a world where the model's intelligence is one checkbox among many in the product. As you point out, many companies are able to provide competent models. But that's not sufficient anymore. Buyers want features around collaboration, sharing, permissions, etc.
My analogues for this market are OSes, office software, and CRMs. Windows + Office remain revenue juggernauts 20+ years after fully free and open alternatives got good enough. Salesforce is worth over $100B, even though there have been good FOSS CRMs for literally decades.
IMHO it's difficult to predict how the AI market will evolve without understanding why people continue to choose relatively expensive alternatives that functionally check the same boxes.
Many American companies will preferentially select American providers, so I expect them to have very high retention and also high net revenue retention
Even if new customer growth falls apart, if they can hold 30% growth while public it will be a good buy today.
Google beats them on performance, and OpenAI beats them on price. I find CursorBench to be a pretty accurate representation for the most common use case (coding, and software architecture).
I mean "not bloated" is not at all how I'd describe Claude. They're constantly adding overlapping features and concepts. There is no simple clear vision, they just ship the org chart (or more likely the whim of individual engineers)
Companies are made up of people and Ant/OAI are able to attract the greatest talent. Outside of that, money and data I'm not sure what people expect moat to look like. The Krabby Patty Secret Formula?
If a free market provides anything, it is the essence of success that is expressed by this statement.
Moats are strategic but only one part of - and a somewhat wasteful part of - strategy.
You don't need a moat if instead of a castle you have found your enterprise to be more like an island or a oceangoing battleship for example.
Strategic defenses (like moats) are as I said somewhat wasteful because your paranoids on staff can always erect some other necessary moat, and the enterprise can spend time and resources in defensive protection rather than innovation that speeds away from any onslaught by competitors.
In their worst form, moats and 'moating' can become institutionalized as part of the enterprise where regulatory capture practices, preferred supplier arrangements, patent hegemony, etc. are distinct enterprise features: i.e, Bureaucracy.
I feel like "left everyone behind" seems kind of arbitrary here. They have the best coding models, but it's not like there's a huge gulf in quality. How much pain would you really go through if you had to stop using Anthropic models altogether?
Argon, Neon, Hydrogen and Carbon have been released. They're contenders.
I've been using the argon and neon orchestrator modes in antigravity and have been reasonably satisfied. I use Claude Opus and Mythos as well, and honestly Google's offerings are more satisfying to use. They feel faster, they're less verbose, they don't wander off and do things outside of scope. And they work.
You're right about Neon, but Carbon is definitely not less verbose. On the contrary, a massive outlier in terms of time per task on our leaderboard, data is battle data no top of OpenCode, http://agentduel.com/leaderboard. I looked into this and the issue is Carbon uses a huge amount of output compared to battle opponents.
You know, on reflection you're spot on. I'm on a Pro plan, and using Carbon burns through its allotment pretty darn fast. OTOH, Hydrogen and Neon have been trucking along well; usually I'd be out of credits by this point in the week.
I remember thinking that Google had no moat in search... "it's just an index" I said. Turns out, there's so much more around it then that. When you're the top player you create moats as you go.
I am totally blown away that a company that was formed in 2021 will have a 1.5 trillion market cap on the public market... to put into context, a valuation not achieved by Apple until 2020
The picks and shovels are down 40% in the last two months. You can buy them at a forward P/E of less than 3 if you think earnings hold through 2029. Which is another way of saying that the market doesn't expect them to. There is a staggering amount of money being bet on "it falls apart soon and it falls apart hard."
It has nothing to do with democrats, that’s just the vacuous meme fed to retail investors. The credit market is tightening and it’s getting harder to borrow money for AI. Coreweave’s unsecured bonds trade at around 15% interest rates. Nvidia’s new CEO doesn’t believe in borrowing money to the customers. The music has stopped.
Yeah, people wanting the bubble to pop haven't been paying attention. It's been popping with the weakest entrants first - the neoclouds. ANT originally wanted to go out at a 2T valuation and had to pull back to where it is now.
Or, in the exceedingly long (in AI-land terms) timeframe you mention, they could rocket to the moon and then crash and go bankrupt because everyone uses them to bits, looks for another alternative, and local models on commodity hardware get to a point in 3 years where nobody wants to spend on tokens any more.
Models that can be ran on (high end) consumer hardware are on par with SoTA from <1 year ago. It just doesn't seem smart to buy Anthropic stock right now.
This is exactly my thinking. Hardware makers want customers too, and 3 years is a looooonng time. Only 6 years ago Anthropic was founded. Something else is being founded today that might look amazing in 3 years.
There is a real scenario, whose probability I cannot accurately assess, in which the first company to build "true" superintelligence will essentially capture the bulk of the entire US economy -- and that it will do so faster than any government can meaningfully intervene. This superintelligence is not a product sold in a chatbot, but rather the internal capability represented by their most advanced unreleased models, mostly without alignment, running on the world's largest GPU clusters, coordinating millions of copies. In this scenario, they don't capture the economy simply by selling tokens, but by owning the internal capability to create (or replicate) any technology, product, or digital business on-demand faster and more cheaply than any other organization on Earth. Selling chatbot tokens would just be the "bootloader" to a future phase in which Anthropic simply enters into and takes over most markets faster than anyone else can do anything about it.
If possible to do, this is a vastly superior strategy to selling tokens, since token consumers can swap out token vendors. But if you simply recreate their entire product and technology lineups from scratch, faster and cheaper, and capture all their customers with superior products and prices (at least at first), then you own the market rather than selling tokens into it amongst other token vendors.
None of this is guaranteed to happen. But I do not see a way to rule it out as a possibility.
Why do you expect your shares to be worth anything in that scenario? Note that their own timeline prediction involves a “pivotal act” in which their AI seizes the means of production.
Exactly! Also in such a world why would they IPO and sell so many shares (in particular if SI is imminent)? Better keep the company private and capture all the economic value created
Yeh and in this situation the people with real access to the models, who would basically control the entire world at that point, will definitely cut you in because of the piece of paper you have with a claim to 0.1% of the company /sarcasm. I mean you're talking about a capability bigger than the government, who is gonna enforce your ownership?? If you buy into their IPO you're just helping them get to this point faster.
Nah. Farmers markets will just get more crowded and full of homemade items. People will move to using local currencies or Bitcoin.
A good fraction of people are smart when the system breaks. Check out the history of the Fall of the Roman Empire. And if Dan Carlin is right, become buds with your local hells angels.
With the reorganizing of Europe following Rome's decline, Germanic tribe leaders were handlebar-mustached strongmen. Dan Carlin analogizes them as modern Hells Angels members in looks and behavior.
and not a senior citizen, and in a gang with younger and older members.
> That’s also Australia, that’s not how it is in the US.
Gang intelligence units here in Australia say that these clubs in Australia have affiliations with gangs around the world, including the the United States.
Gangs in the USofA agree:
There are more than 300 active OMGs within the United States, ranging in size from single chapters with five or six members to hundreds of chapters with thousands of members worldwide. The Hells Angels, Mongols, Bandidos, Outlaws, and Sons of Silence pose a serious national domestic threat and conduct the majority of criminal activity linked to OMGs, especially activity relating to drug-trafficking and, more specifically, to cross-border drug smuggling. Because of their transnational scope, these OMGs are able to coordinate drug smuggling operations in partnership with major international drug-trafficking organizations (DTOs).
Collapse of existing currency is a reasonable prediction for what might come next in this scenario, if that is how you intended your comment to function.
Farmers’ markets are for rich people nostalgic for an imagined past, when rosy-cheeked, healthy maidens milked cows barehanded, with the kind of smiles only 21st-century orthodontics can deliver.
In today's environment, sure. They are institutions that can expand to fill a need should it arise.
The Filipino municipal bazaars show what they could turn into when transaction costs go too high, as well as when comms and travel are difficult. They can be the lifeblood of a community.
We in the US/Europe are a little spoiled with accessibility of retail goods.
Lmao. If that happens, farmers are guarding their crops with their lives and the people with the most guns will be the ones in power. Your hippie fantasy won't happen, sadly.
Chaos would certainly be the first stanza in the chorus of calamity and recovery. But it normalizes -- quicker than you might think, I'll add.
The people with the most guns also sleep, and communities have long memories. There are good reasons "might makes right" doesn't scale or really hold any form of cohesion beyond a band.
I don't see it working out. Before you could get any real balls rolling in the physical world competitors will be there with their own GAI, assuming we are anywhere near actually building it to start with which I seriously doubt.
And even if they are first, why assume the first is going to be the best? Maybe 6 months later a different AI model will come out that stomps all over the first one. Maybe a competitors isn't any smarter but they can do it with half the power and can scale their intelligence twice as fast, so even being later they still end up in the lead.
> But I do not see a way to rule it out as a possibility.
Your argument is unfalsifiable. Anything can happen by this logic and they could be valued whatever trillion $, like, maybe they’ll invent a way to travel through the multiverse. Maybe they’ll create the star trek replicator thing.
No. While the scenario is speculative in the sense that it is not guaranteed to occur, it is not speculative in the sense of being science fiction. It is a very mundane story of competitive capture of enough existing segments of the economy to gain negotiating leverage that surpasses a threshold. That's really it. It's actually quite boring and doesn't involve anything like what you're suggesting even remotely at all.
It is certainly not unfalsifiable. There are other scenarios in which this does not happen. For instance, the most obvious is that other organizations develop a similar capability at roughly the same point in time. This results in a "multipolar" economy that is mostly controlled by a small set of distinct owners of this kind of capability. The multipolarity prevents total capture and thereby prevents the rest of the scenario. Another obvious scenario is one in which the government acts soon enough to short-circuit this process before one organization gains a sufficiently decisive advantage that government intervention is no longer efficient enough to be effective. Yet another obvious scenario is one where a short-term financial crash deprives Anthropic of enough capital that this strategy is paused or slowed down enough to allow it to be countered. And yet another obvoius scenario is one where the scaling laws simply stop scaling or the amount of compute needed to scale them to this point cannot be physically obtained, thus allowing time for the strategy to be countered.
All of these alternatives would come with clear data points early on which would allow one to distinguish them from the original scenario I described. Right now, though, the data is ambiguous, and some data points lean slightly in favor of what I described, but not nearly enough to make any guarantees whatsoever.
>they don't capture the economy simply by selling tokens, but by owning the internal capability to create (or replicate) any technology, product, or digital business..
No, what that AI will do is that it will create an artificial reality (because people will perceive the world through the words of the AI) that caters to the highest bidder.
This is the myopia of tech people.. they exaggerate the weight of digital. AI could very well be able to replace any digital business. But purely digital businesses are such an unimportant sector of the economy. All of the important sectors use digital infra at some point, but none are purely digital.
The bet is that sooner or later, even the remaining meatspace holdouts will be effectively digital and thus controllable by AI.
A lot of manufacturing is already there - that's a part of why China undercuts Western manufacturers, a lot of cheap land and subsidies means you can be far cheaper than Western companies already, even more if you automate what you can. Farming, at least anything with growing plants, is also barely requiring humans any more - precision farming with RTK is absolutely insane. And what remains is trades (where, again, automation like prefabrication is a threat), parts of healthcare and education.
We will own nothing, rent everything and fight for the scraps Sam Altman hands to us.
>Farming, at least anything with growing plants, is also barely requiring humans any more - precision farming with RTK is absolutely insane
In your dreams maybe, but not in actual real life.
And if you want an actual Butlerian jihad to happen in real life, then yeah, please put out of work everyone associated with farming on the planet within a few years. You describe the golden path to full-on civil war in 3/4 of the world's countries, with either AI being destroyed or an automated massacre _à la_ Terminator.
Farming is already at <1% of people in society (Germany is at 875k of 85m population). And that is only due to mechanization and large scale consolidation.
As soon as there is wealth available to finance modern ag equipment, farming employment will only fall because the productivity gains are too big to ignore.
And precision farming will go even further. Much more yield with even less staff than we already have, the only reason why it isn't commonplace is that the required tractors cost north of 500k a piece and agriculture is risk-averse. But mark my words in 10 years it will be the norm in Western countries.
Or just another instance of liberatrian nerd Dunning-Krueger. "Faster than any government can meaningfully intervene", c'mon - if really it is that kind of force majeure thing, they will just say "screw the free market". As if the government is bound to uphold the sacred rules of neo-liberal economics.
Same as Goedel thinking that a logical loophole in the constitution is something that could turn America into a dictatorship legally. Guess what? In practice, somebody wanting to instate a dictatorship probably won't give a crap about some subtle logical fallacy. Same with the Keynesian economics that Americans perceive as law.
You seem to think I'm invoking some kind of obscure academic theory. I'm not. This is actually the most basic competitive process imaginable playing out, and that's all it is. That's exactly why it's actually plausible.
You might have a point with computer games. But if you take facebook on the other hand, it would be worthless without people using it. An AI would just create the website. For most businesses, there are non-digital elements which play an essential role, and that remains true even with AI.
Those businesses will simply buy tokens, as they already are. Their strategy will increasingly be steered by AI under the control of Anthropic. Their internal operations will no longer function without Claude. They'll shed staff and optimize their organizations around the extraordinary savings offered by AI. Their physical operations will gradually be automated under the control of robotics powered by Anthropic. Those who don't do this will go bankrupt in the face of competition from those who do. In some cases, perhaps many, to accelerate this they'll simply get bought out, then forced through the above process faster than it would organically occur, either by Anthropic itself or a new breed of private equity firms that specialize precisely in this kind of operation. The process will be slower at first, but not nearly as slow as you think, and nothing you mentioned stops it.
how ywould your putative superintelligence come up with the supply chains, raw materials and presumably robots to do all this seemingly overnight? or are you only talking about digital goods and services?
even so, what about network effects? how do they lure users away from TikTok and Instagram? AI-generated clips? or say Anthropic recreates Amazon. that's a software company, right? so how do they get all the warehouses, cloud infrastructure (they will need entirely different kinds of datacenters and hardware), relationships with third-party merchants, etc.? or say SpaceX - is your superintelligence going to be setting up massive autonomous factories and buying thousands of tons of aerospace materials and billions of dollars of machinery to make that happen? where's it get the money?
and even if this somehow does manage to accomplish all that, how politically popular will it be? what prevents the US government, lobbied by.. the majority of corporations, and hounded by constituents, from invoking the Sherman Act and busting this trust like AT&T or Carnegie Steel?
In terms of supply chains, imagine a scenario where Amazon undercuts all retail stores, taking absolute control over every consumer need.
They already have solid control over logistics and inventory management.
It’s not a stretch that they then capture the means to control primary inputs and manufacturing.
Throw in an overly enthusiastic AI to continually optimise the supply chain, continually optimising a country’s supply of food and commodities, and we have all the ingredients for the classic sci-fi dystopia of a computer controlled society.
> owning the internal capability to create (or replicate) any technology, product, or digital business on-demand faster and more cheaply than any other organization on Earth
In this scenario, capitalism is by definition over, and market valuation becomes a meaningless concept.
I'm not sure how so many people betting on this scenario don't see that fundamental flaw: if one company can replace the whole economy, then there is no economy. In the "good" outcome the whole economy gets replaced wholesale by some flavour of "luxury space communism", and in pretty much any other outcome we end up with war/famine/etc as the system breaks down
Otherwise known as the "Once we create Skynet, of course it will do what we want it to" business strategy. (For the record, I think the chance of LLMs leading to any such thing is zero.)
If someone makes (super) AGI which improves recursively, open AGI will also be a thing shortly after. Not to mention all the commercial competitors (OpenAI, Google, Meta, X, etc.)
Microsoft still dominates the office tooling, and many will thus get their AGI via copilot and similar.
Lol, dystopian fiction has such strong roots in our culture.
Why is it that Anthropic gains so much from this perceived image, but then needs to public money to turn it into reality? If they were at all close to any of that, surely their internal AI can figure out how to self-finance the company asap. Or maybe that's exactly what it did and it told it to go public. Company selling shovels adage.
I think markets are way more complicated and capturing them isn't going to happen the way you're suggesting: dollars, and the public, don't really like copypcats. You would need to understand the business, the product, and manipulate people at an unprecedented level. Anthropic selling tokens as their business, means that all the way up to this "super intelligence", means that they've enabled everyone's defenses and raised the threshold for this dystopian hostile taker, i.e., it won't happen.
They are only cashing out like $2B. Non-founder directors, employees etc can also sell 10% of their shares immediately, but most of the rest is locked up for 180 days.
10x is unlikely, but the IPO is only ~5% of shares, and then they will release twice that fairly soon after. So it's VERY likely to see some rise early on, but not likely to be a sustained rise in pricing as the amount of shares available continues to increase significantly.
The suckers are going to be pension funds just like with SpaceX. The sheer size will mean OAI and Anthropic are fast tracked into indices and thus tracking ETFs that are among the most popular investment vehicles as they are low-fee, which means an immediate influx in capital followed by a permanent money spigot from pension savings.
And when it all crashes down, there will be fire sales and the vultures that are flush with cash from the exit will buy up what remains at ridiculous discounts.
We shall own nothing and rent everything, and Big Tech is obviously colluding with Big Finance to usher in neo-feudalism.
Turns out...in an economy hollowed out by 15 years of governments inflating asset values via money printing, 0% rates, QE & deficit spending; selling to consumers instead of asset holders (businesses) is a losing strategy.
Anthropic picked the better market at least for the short term. Long term I wouldn't count OpenAI out just yet. The consumer market tends to compound into winner take all eventually given the network effects of advertising marketplaces, even more so than B2B.
Im talking about the Office and corp/job world, in private conversations it may differ: Maybe most people not in tech dont even know what Claude or Gemini is?
Even in Office and Corp/job world, at least in my circle of visiting differe Offices, most people use Microsoft Copilot (already bundled in Office) or Google AI (one that shows up during Google search. Rarely I see Gemini and yet to see anyone using Claude for non-coding purposes.
While I sympathize with what happened to you, the argument "X happened to us therefore we will leave Google and Google will stay behind" is hardly sound. I haven't seen much better support from Anthropic or OpenAI.
The problem is Google has been famous for this for most of their services and this is their reputation. Many people have sworn they won't depend on GCP or other things.
Have you tried to get a hold of an Anthropic GTM/Salesperson to pay them to use their enterprise products? We ended up having to resort to messaging them on linkedin.
Google is running on search Ads revenue, it will collapse if that stream collapses, because company is too overloaded by bloat (200k employees vs 5k in Anthropic).
I'm convinced Google is a much better investment and would like to hear an opposite argument if anyone here has one.
I don't think there is a lot of difference between the "sota" models long term. Even if we said that say Anthropic is ahead, Google will be able to do the same in months.
But if it turns out that open weight models are the future, OAI and ANT are fucked, but Google has massive infra to host those and make money anyway. It also has the products and business connections to integrate AI into users workflows. And lets not forget, it earns massive amounts of money even without AI, so there is a higher price floor even if an AI bubble pops.
And even if it turns out that Anthropic takes over the world, Google still has a stake in them. (Although so will everyone else though SP500 by that point.)
I don't understand why would you invest in Anthropic IPO instead of GOOG, but as I said, would be interested to hear opposing thoughts.
Sure, but in that case the Gemini arm of the business (or whatever they rename it to the next time) carries the torch forward. Unless there is no money to be made in AI and everyone loses, but then again it would still be better than investing in Anthropic.
> Sure, but in that case the Gemini arm of the business (or whatever they rename it to the next time) carries the torch forward. Unless there is no money to be made in AI
the question is if its enough money to replace 300B of Ads revenue.
Biggest Anthropic risk is that they have those many B hardware commitment.
Otherwise, they have some solid foundation model, which is cheap to post-train and improve, solid market share, they can milk this part for some time and be on the top.
Google is extremely vulnerable against paradigm shift if they lose Search Ads revenue, they have too much bloat of unnecessary people and systems complexity, and easily can collapse.
> I'm convinced Google is a much better investment and would like to hear an opposite argument if anyone here has one.
I have an argument, but it's mostly snark.
Like, Google are definitely in the best position here, but they've messed up so many times when it comes to product management/enterprise adoption that I am confident in their ability to snatch defeat from the jaws of victory.
I 100% agree with that snark and think Google could be so much better were it not for internal politics, perverse incentives and inability to focus and follow through with products.
Still that being said, I bet on them rather than against them.
> Like, Google are definitely in the best position here, but they've messed up so many times when it comes to product management/enterprise adoption that I am confident in their ability to snatch defeat from the jaws of victory.
Yeah, but what about Microsoft? They're the real threat here: they can get or train an open-weight model, embed it into Windows at very little cost to the end-user, and then the end-user will have AI in every app that they run, regardless of whether that app has AI connectors already or not.
The entity holding the business relationship already has the most clout, and while Claude takeup was good, it has nowhere near the same numbers as Google enterprise, which in turn is nowhere near Microsoft.
In terms of market-cap, these are all trillion dollar companies. In terms of enterprise agreements, if you were to draw a line-chart for enterprise volume for each of those three companies, MS would be the very top of that chart while Google would be somewhere at the bottom. Anthropic wouldn't even show up; their line would be fractions of a pixel tall, at its peak.
Anthropic is much more focused on B2B while OpenAI feels more B2C. Both sell to both markets, but the focus feels different and I believe the resulting customer base is quite different.
I think the implication was Anthropic has gone after the B2B market (with Claude Code and Cowork), where as OpenAI has gone after the consumer market (with ChatGPT). I'm not sure I totally agree to be honest; Claude Code was a happy (and fantastic) accident if the stories are to be believed.
Even before Claude Code, Anthropic were way more focused on B2B (source: a former company started using them in late 2023, based on our internal evaluations).
This post makes me realize that, despite being an enthusiastic Anthropic customer for near 2 years now, I actually don't know any of this historical context.
I am totally blown away that a company who does not turn a profit, whose revenue is slowing down, and has hard financial commitments 25x times, the money they just raised, over the next four years, was able to get 81 billion from investors.
As predictions are only worth anything, if they are made ahead of time...I predict their share price will lose 25% of its value,over the next 3 months. Drop to 40 to 45% of IPO price, after their first public quarter release.
Facebook also didn't have "there's a very large chance we'll be completely replaced by free Chinese models" (which curiously Anthropic doesn't emphasize in their S-1 filing).
This. Chinese models are indeed a threat. But the bigger question is world is continue to give them power to purchase lobbying power to release more powerful models, collect more useful data at the risk of everyone's privacy and control everyone. People need to realize how much data they continue to give to models and their training. The more they use the more data and powerful they become. And Chinese models take way that data collection. And that's the threat. It's not the revenue from models. It's the data that they can't collect is what they scared of. And that's why they put a safety narrative in disguise.
The first 5 years estimate is around ~$355b–419b. We don't know for sure because most of their deals actually span 5+ years, up to 10-15 years.
They're at $86 annualized revenue right now. If they grow at 0% for next 5 years, they will generate $430b, which is enough to cover all compute costs.
Well... I guess tackling technological innovation that has a potential to transform human civilization on the most basic level does that for you. It's not acompany that releases 15th overpriced gadget to the increasingly poor populace.
but company that releases 15th overpriced gadget already transformed human civilization. Im pretty sure anthropic will have it's time and become a company releasing 15th overpriced model to the increasingly poor populace. Capitalism is cyclic: expansion, peak, contraction, trough
How can charge a premium over competition if they catch up? Branding, smoke and mirrors won’t last long in llm industry. Big corporation switch AI at the blink of an eye. And are already looking more into self hosting.
Well, was anyone else capable of building an operating system other than Microsoft?
The answer is that business is not just technology.
On the technical side, it's also important to see beyond the current market, what's currently possible, and broaden to - ok, many innovations are coming in AI, who is most well-positioned to be the centre of that?
> Google could "Internet Explorer" Anthropic to death with a superintelligent ad stuffed free Gemini. shivers
Hell, Microsoft could do the same with a cheap inferencing model; if people simply use copilot because it's already embedded into all of their applications because it's embedded into the OS (Office, Windows, anything that runs on Windows), only a few dedicated people will continue using the expensive paid-for options.
let's hope not and instead root for a more evenly distributed AI future with open weight and fully open models, running on local hardware with privacy.
I think there's a future for open weight models. To continue a half-assed analogy, Linux has thrived whilst Microsoft has captured the value. They're co-existed.
Fully open models will be more hobbyist, I expect. Open weights will continue as a strategy but probably not as prominently as today.
Because open weight models are a business and political strategy without a strong business model.
Training is very expensive and open weight models don’t generate revenue comparable to frontier labs.
The reasons they exist are mostly “commodify your complement” and (in my, apparently controversial opinion) the Chinese govt trying to take down the American economy and undermine the US frontier labs in order to compete in the AI race.
People are so accustomed to free software (that they by and large never contribute anything to) that they get confused about why we can't just have an economy of it.
Possible, as long as everything goes well for them. Their debt deals don't care how much of a transformational company they are. If they get less demand than they expect, or slower demand than they expect, they could end up bankrupt.
If their business case is true, it has withing itself the seeds of their own death. I can ask a super smart self improving model to create a ...super smart self improving model ,-)
If their bussiness case is not true, we will be paying for this bubble in higher interest rates, for the next 25 years.
Now tell me: Which scenario do you think is more likely until 2030?
A - We cure cancer, create self improving models of hyper intelligence, that will solve Fusion and bring us to the stars, create a CPU, an OS and a 3D printed Car and deliver it to you within the same afternoon, solve Earth warming issues...
OR
B - This biggest credit bubble in history, will put the sovereign debit of the US and 30 year Treasury to 8%, force the US government to default, cut spending, increase taxes and bring about a recession that will take 25 years to recover from?
Microsoft left their competitors behind pretty quickly didn't they? They had no serious competition in desktop OS and office software for many years. Anthropic has a near equal in OpenAI and compared to Google they seem to have some advantages and some disadvantages to overcome. That's not even mentioning China models.
I am reminded of the infamous "What were you thinking?" comments from Scott McNealy. I know it's practically a cliche, and you can summon its ghost by saying "bubble" three times in a dark room lit only by a candle.
many large enterprises in this space in the USA is operating like Korean companies have for a good 80 years now. They call it 'chaebol'. Chaebol arose from family enterprises, but it really doesn't matter who or what are the controlling interests
cross shareholdings, agreements to buy, technology access, barriers to entry, assimilation of smaller upstarts into the chaebol, bonds, lending....
I am blown away by how VCs pump up some companies to eventually get access to your 401Ks without any questions neither from government nor from public.
I will be double blown away when bubble pops and some companies go bankrupt or bail out by the government, who already paid VCs with your money.
Funny that you say that because from the reports Ant are showing only their summer revenue numbers - when all the enterprise employees had open accounts. Since then the enterprise clients have been jumping on their own infra and open weight models.
So yeah enterprise is just slower to adapt and they are adapting away from providers like Ant.
Ant only chance to somehow survive is regulatory capture where they will be only allowed provider for gov contracts.
I think where we’re going with AI money will have no value or at least no value as we understand it today. Once they will achieve AGI it’s just new ground economically
Do you have any proof there is even glimpse of anything remotely AGI comming from LLMs?
It seems like corpo developers who were working mostly on self inflicted logic/math puzzles at best (more likely aligning UI) think that their work was so hard and precious that their partial automation mean everything in the universe must be solved.
The money have will have no value because the system implodes. And if we somehow magically have
AGI it is very bad for us digital peasants because the money system will stay it will be just 5 overlords who will control it while rest of us will beg for scraps.
For software generally we’re almost there because coding is already automated. To close the feedback loop we need formalized verification which is already growing. In some companies the loop is already closed, watch poteto’s talk about how they work on grokbot.
We’re still kinda far on the physical level and robotics have some work to do for catch up but we’re closer to software Agi than ever
It is a financial bubble. No one is denying its usage, but the math just isn’t mathin’
One crucial key assumption is that customers will simply pay whatever the AI companies charge once the subsidized gravy train is over. I do not believe that. Right now a small share of paying users are paying for most of it (via API), it is a whale economy.
>> One crucial key assumption is that customers will simply pay whatever the AI companies charge once the subsidized gravy train is over. I do not believe that. Right now a small share of paying users are paying for most of it (via API), it is a whale economy
> But models are getting cheaper over time per task. I wouldn’t worry about pricing, it will only get lower
Isn't that even stronger evidence of a financial bubble? Once the whales are gone, where do you get margins high enough to support the share price? If the margins and resulting revenue can't be found, the bubble pops.
If models were getting more expensive over time, then sure, I'd expect the AI shares to rise (because people have to buy from them, no matter the margin), but if models are getting cheaper over time, all those companies will be in a race to the bottom with open/free models.
Money won’t matter after ASI/AGI will be achieved. It even doesn’t matter now. This is not just another tool but life altering event for whole civilization
> Money won’t matter after ASI/AGI will be achieved. It even doesn’t matter now. This is not just another tool but life altering event for whole civilization
Yeah, the asteroid that killed off the dinosaurs and sent the remaining half of all organisms extinct was life-altering too, you know.
If there are fixed number of tasks, and cost per task is going down, will profits go up or go down?
Or maybe the task pie can expand indefinitely, but I'd need to see the models handle tasks other than code and general knowledge work (much of that not so well either) to think that.
the valuations for these random SV "AI" companies these days is just eye watering. It really does seem like theres so much money concentrated at the top that they can throw vast amounts at the wall to see what sticks. Anthropic and Open AI are generational companies but even their values seem extremely inflated. SV has mastered the art of juicing companies to the gill before releasing them into the wild and they've done it at record speed for these two.
It doesn't say much about the company itself, but about the state of the economy, where all the money has been concentrated at the top, in the hands of people who have nothing else to spend it on than on more financial assets. So there's enormous amounts of money available for more stocks.
It's why stock markets and corporate profits are at an all time high, while most people struggle to make ends meet and governments are woefully underfunded.
That $1.5 trillion IPO is the result of a completely lopsided economy.
Well, yes. But also the company is projecting to sell pretty much every white collar worker a $200/mo Claude subscription, which is an absurd amount of revenue compared to said Apple, which promises to sell a new phone to 20% of people every few years.
It's hard to put a percentage on it, but I'd wager for 90% of the earth's population, paying $200 a month for a personal LLM subscription isn't viable or simply not worth it.
But will it be Claude? Probably it goes through Microsoft through copilot. In which case it becomes a race to the bottom. I never need Fable even today: Opus suffices on 99%. So why do I need anthropic? Opus will be replicated in half a year by every lab.
In that sense I am super bullish on Microsoft as they have the vendor lock.
For anyone who has used the various tools, the harness still matters and copilot is pretty bad. While Anthropics models lose on cost compared to OpenAI, there is still something that works really well in Opus paired with the Claude desktop app when it comes to tool use and solving tasks.
Well at least profit from the appreciation of artworks can be written off a little easier. Stocks are too liquid and regulated to serve the same purpose.
But they can be lended against just fine, which is apparently the common scheme for tax avoidance these days.
If enough value depends on the same stock, there will be an intrinsic need to keep valuation high which is completely decoupled from the actual business of the company.
Which I believe is how we got a company to invest in the press-conference of a flying car...
Oh yeah that makes sense. I assume that lenders will price in the inherent risk from the volatility of the asset, so it will take significantly more stock (at the current valuation) relative to the borrowed amount?
Yeah, seems to be common to borrow 50-70% of a stock's value against the stock.
The neat thing is that instead of selling the stock once (and pay capital gains tax), they borrow against it for low-interest loans or lines of credit (which are debt rather than income, so generally not taxed).
This way they can access cash while the underlying stock continues to compound in value, and pay a low interest-fee instead of taxes...
Because billionaires and those right below them spend a ton of capital on perpetuating the idea that the average person can be like them. Hell they even have their own podcasts now lol
I love how the exact same people I describe above (normal people who've be brainwashed to think they can achieve the same level of success that only ~3k billionaires have had), have also been programmed to automatically dismiss anybody that writes a sentence that uses the word billionaire negatively, as "braindead" leftist economics.
They have brainwashed you to the extent that you spend your time reflexively defending them in the comments of online spaces. You become an activated foot soldier, coming to the defense for the very people who seek to enslave you the second anyone says something bad about them.
Nope, I just understand basic economics and have logical deduction skills instead of buying sentimental bullcrap from demagogues. Please read a book, you will be doing all of us a favour.
Perhaps you should be the one to read that book. None of the things discussed here are in any way controversial, except with the billionaires and their politicians. Many economists are saying the same things we are saying. That's where we got it from.
I took many economics classes, its partially why I have informed opinions on the matter. You're so confident reading a book would change my mind, surely you have a reccomendation.. Or are you just politicing? Maybe lay off the X algo.
high volume, 'flow', of wealth is usually what economists call a hot economy, which usually leads to inflation which then usually leads to problems that a central bank has to mitigate or consumers have to bear the weight of.
I know this is getting into the weeds of economics ...the OP was posting about a pending IPO...just setting something out for consideration....
There is something more though I think in terms of lack of cultural capital.
There isn't enough cultural activities we value in society so we have largely turned our culture into this kind of net worth sporting contest. Billionaire is just round number bias that reflects a goal post in the sport at the highest level.
Late neoliberalism I believe is the right frame but this is still a rather obscure idea.
It's annoying that people all readily acknowledge that billionaires do not have piles of cash, but then they incessantly talk as if they do.
My friend, the billionaires wealth is the economy. It is the thing that is flowing the money.
Jeff Bezos wealth isn't cash redeemable pieces of paper with "Amazon Share" stamped on them.
His wealth is warehouses filled with goods, software running logistics networks, CPU's serving web services, trucks delivering packages, and a workforce of 1.5 million people all aligned on serving paying customers for their needs or wants.
So while you can detest it and hate it, at least recognize that that is what you call "accumulated wealth", not a pile of money where we are one legislative act away from busting open like a pinata.
If you snapped your fingers to distribute his wealth, you'd feel pretty dumb when you end up with 13 aluminum dolly carts as your share.
So how did Jeff Bezos become one of the largest private land-owners in the United States, if he has no wealth?
How did he pay for his wedding, with non-cash-redeemable pieces of paper?
Or how did James Dyson manage to become the biggest agricultural landowner in England?
And who was then buying e.g. Van Gogh's art for 400M USD just this week?
It's equally annoying that people readily acknowledge that billionaires have no cash because they openly operate a tax-avoidance scheme where they lend money against the stock they own, to benefit from its value while it keeps growing and "officially" are poor.
100M USD of stock create instant liquidity of 50-70M USD at low interest, without ever needing to sell the stock.
> It doesn't say much about the company itself, but about the state of the economy, where all the money has been concentrated at the top, in the hands of people who have nothing else to spend it on than on more financial assets.
Huh, what?
> So there's enormous amounts of money available for more stocks.
Money (nor stocks) don't have to change hands for stocks to become more valuable. All that's required is that the likes of Citadel and Jane Street quote higher prices, no trading necessary. (Of course, they can and will only quote higher prices, if that's justified by supply and demand.)
> It's why stock markets and corporate profits are at an all time high, while most people struggle to make ends meet and governments are woefully underfunded.
Government expenditures have grown like crazy in much of the world, haven't they?
That's not how supply and demand works. There has to be trade in order to establish a price.
And governments in much of the world have actually been cutting expenditures, leading to serious shortages in many places. Increased poverty, reduced quality of education, crumbling infrastructure. Governments were better at this from the post-war period to roughly the 1990s, until neoliberalism and austerity became the law of the land.
> That's not how supply and demand works. There has to be trade in order to establish a price.
Why? Anything you can potentially trade against, especially with a tight spread, counts in my books as a price.
> And governments in much of the world have actually been cutting expenditures, leading to serious shortages in many places. Increased poverty, reduced quality of education, crumbling infrastructure. Governments were better at this from the post-war period to roughly the 1990s, until neoliberalism and austerity became the law of the land.
> Austerity is a set of political-economic policies that aim to reduce government budget deficits through spending cuts, tax increases, or a combination of both.
Government deficits have reached record highs in eg the US and France. Where is that austerity?
Neoliberalism is, alas, dead. The glory days of the 1990s and early 2000s died around 2008.
The K shaped economy trope is directionally correct, but you have the cause backwards.
It's not that governments are underfunded, the K economy is a direct byproduct of government stimulus across all developed economies for the last 15 years.
When you inject the market with QE, the money printer and most importantly; 0% or negative interest rates for 15 years...it massively inflates the value of assets. Rich people own assets.
This is why things in the US with the prospect of generating future cash flows are at historically high valuations relative to GDP: public companies, private companies, real estate, etc.
If you think government is 'underfunded' and the solution is giving it more money to dump into the economy, ironically, you're going to inflate asset values further and make the K economy worse.
The stimulatory tools that are easiest for government to use are monetary policy, not fiscal policy.
But let's say you do manage to pass populist fiscal policy, like, giving first time home buyers $20k to help with a downpayment on a house.
Surprise! Turns out you've simply raised the value of all starter homes by $20K by increasing everyones capacity to pay for a scarce asset, inflating the value of the asset and handing the money to the person selling it.
The solution is to increase the supply of assets (build more real estate and competing businesses), not stimulate demand for existing ones. But that requires a much more rational, long term policy view than dumping blankets of cash on angry populist voters.
Alternatively you could massively tax the wealth of the very rich and revert the concentration. Then you are not competing with them for the scarce assets.
Again, giving government more money to dump into the economy will do literally nothing to solve the problem.
You will simply be taking money from the rich, and then giving it right back to them by inflating asset values.
Because the assets are the things that are scarce and valuable, and you're now disincentivizing the creation of them while stimulating demand for them more by dumping cash on everyone.
There is only one path: creating more valuable things (more supply of real estate and businesses offering competing products lowers their price). China, ironically a 'communist' country, figured this out yet we're still struggling to.
The truth is we don't want to face this reality. We'd rather sit at computers, place fantasy expectations on the politician du jour to save us, and repeat smart sounding memes we don't understand for magical internet upvotes ('late stage capitalism' 'K shaped economy') instead of picking up a hammer and building anything.
Why not use the extra taxed income to support "picking up the hammer and build things"? Historically, government have been the one to do that for moonshoty projects, a lot of "things built" to kickstart SV was government spending (as much as modern SV propaganda tries to erase that fact)
Why does your mind equate "taxing wealth to reduce inequality" == dumping that money as hand out to the citizens? Why can't government, that after Thatcher and Reagan threw up their hands saying "we are too dumb to plan anything", just go back to planning long term spending (such as reducing construction regulations or make it more efficient, both requiring capital I assume), which, ironically, is also what you're praising China for.
With respect, it sounds like you're also falling for the criticism you give out to others about "smart sounding memes" ; yours one step further than the one you're criticizing, but stops at the step where more tax== hand outs meme
> Why not use the extra taxed income to support "picking up the hammer and build things"?
That would be the ideal scenario, but it just doesn't work reliably in a decentralized system, which I think we can all agree is better than a centralized one (I'll take messy decentralized democracy over brutal authoritarian control every time).
Democratic politicians are just renters. When's the last time you renovated the bathroom during a hotel stay? The one time the US made massive investments in building stuff in the past 100 years was when we had the most dictator-like president (FDR).
But centralized power can just as easily go bad (see the last 2 years) as it can turn out good.
The only solution in a democracy is to let decentralized market forces dictate the price of things, mitigate any anti-competitive flaws (monopolies, cartels, local government vetos against building, etc), and deal with the harshness of reality instead of looking for magic bullets.
The market takes care of itself if left to build unencumbered, just look at the price of housing in Austin and all the new assets being built in AI. The whole pessimist scenario in AI is that competition will cause prices to drop and kill valuations. This is good. Lowering asset prices is what we want!
I think we are broadly in agreement but I can't shake the feeling of "smart sounding meme" feeling from your reply too. Granted it's forum post that is limited form of communication... I probably sound the same.
You give one good(FDR) and one bad (Trump) example why centralization is bad. With your previous example of China and CCP, the tally seems two good example, so as a basis for discarding centralization I feel it's too handwavy.
I rather see the details of how centralization can be implemented.
CCP, for its centralization, have long been factional, localized and meritocratic in certain aspects (you only climb the party ladder if you succeed at local province first), for example.
Another is Japan's centralized zoning law, or centralized train track management, seems to yield better result for affordable building and public transportation: a central/decentral, public/private symbiosis.
I'm in agreement with you that "mitigate anticompetitive flaws" is an direction because it's clear certain domains simply can't be expected to construct competitive markets on its own, or that the markets outcome is beneficial to the society.
"We" is we citizen, government and companies. We can't directly affect the companies (except with our purses but let's be realistic on that), but we can affect our government. That entity completely outsourcing it's power to companies is stupid imo.
Create guardrails so that market competition is aligned to the broader society's wellbeing is a minimum to me, and the interest question is where and how. Being blanket anti tax or anti centralization robs us of being able to discuss that part.
I changed my opinion a little when I got to that discussion, about wealth tax for example (how hard it is and how there are lower hanging fruits such as what Richard Murphy have been claiming: https://youtu.be/Iir0GP40_vw) and it seems like more proactive approach to this discussion
I don't know how centralised you consider the US or western Europe, but they have done exactly this very successfully after WW2. They invested enormously in infrastructure, housing, and economic growth, and it was incredibly successful. It was only after neoliberalism came along that people started to believe this was bad. And that's when the excessive inflation of asset prices and growing inequality started.
There is no scarcity of shares. They're entirely artificial. Their absurdly high valuation is really just because the rich have too much money and nothing else to spend it on. Giving that same money to the poor instead will not inflate those same prices, because the poor will spend it on other things.
And those other things the poor will spend it on are also already inflated because of a lack of competition. Corporate profits are at record level across many industries, despite the fact that 60% of Americans are struggling.
Your scarcity argument only really holds for real estate, but that too is exacerbated by rich people buying real estate just because they can, because they need to put their money somewhere. Meanwhile poor people are struggling to buy their first home. Poor people are not going to buy multiple homes, they really just need the one.
> Again, giving government more money to dump into the economy will do literally nothing to solve the problem.
You’re assuming that taxing the rich would mean giving the government more money, but that doesn’t have to be the case.
You could, for example, raise taxes on the rich to the point where income tax is dropped entirely for the vast majority of people who currently pay it.
You could calculate how much the federal budget would need to fill the gap created if, say, nobody earning under €100k a year had to pay income tax anymore. That sum would then be offset by raising taxes on the richest people. No extra money gets “dumped” into the government. It’s the same amount as before, except a large share of it is effectively handed back to people who earn their income from work.
You could take every dollar of wealth Elon has and give every American $2,500. Is that really going to help?
More importantly, taxing the rich based on unrealized wealth means forcing them to sell their assets. Markets would be flooded with new supply and no buyers, tanking prices and putting everything in a much worse position.
People want groceries and house prices to come down yes, unless you already own a home. People don't want prices to collapse though, and more importantly our government doesn't want prices to come down.
Deflation is a death spiral for debt markets. Our economy is so precariously balanced on top of debts and derivatives that the whole thing would crumble if prices of major asset classes collapse.
I own a home and I'm totally fine with house prices coming down. I don't have my house as an investment, but to have a roof over my head.
> Deflation is a death spiral for debt markets.
People say that a lot, but is it actually true? Are there real life examples of this happening? By far the biggest example of prices coming down over a long period, has been computer components, and that never hurt those markets. It's only the recent rising prices that's killing markets.
Price deflation for any particular product isn't a big deal, the risk is broad deflation across markets and industries.
Our financial system is heavily leveraged, when prices fall leveraged bets begin unwinding as forced sales, pushing the spiral further and further.
You generally don't see price deflation hurt triggering a debt crisis as the starting point, but that's only because our system is structurally designed for persistent price inflation. You do see it in the middle of a crisis though, debt is the dry brush that burns everything down after the fire is lit.
>I own a home and I'm totally fine with house prices coming down. I don't have my house as an investment, but to have a roof over my head.
It's an easy counter point to make because you don't have to show anything for it. But when we ask people to put their money where their mouth is, they reliably protect their home values.
To put another way, say you needed to move for whatever reason, and in the new location you found a home you loved for $100k under the value of your current home. Would you sell your current home for $100k less because all you care about is having a roof you like over your head?
Again you can easily say yes to this, but in reality, everyone pockets the $100k.
I expect we will be seeing a similar scenario play out with Anthropic's IPO. Its all well and good to call oneself an altruist, but once that check comes in I don't expect them to give it all away.
I will be quite impressed if the EAs at Anthropic do precisely that though.
The issue with accepting this status quo of tax policy is that it isn't the same in other countries. The US (uniquely?) does not tax asset wealth when it is distributed which allows for huge accumulation of wealth outside of the tax base, Canada has deemed disposition for example that brings any asset that is distributed from one individual to another into the tax base even if the asset remains the same asset (unsold stock that ahs appreciated).
You can tax "unrealized" wealth when that wealth is used to actually fund expenditure, currently anytime this wealth exchanges hands as the same asset it is seen not as a realization of wealth, which is what needs to be changed.
I'm very much in favor of closing a loophole that allows the wealthy to borrow against their assets and have tax-free income.
I'm very much opposed to a wealth tax that forces a person to have to pay for gains they never realized even if that means having to sell the very asset in question.
I personally don't actually agree with income tax at all, but if we have it then it should standard across the board without said loophole and acting as a tax on income, not paper wealth that can't be spent.
> "The solution is to increase the supply of assets..."
Let's be concrete (and unpopular): If one of the few assets left to buy for billionaires is land, which creates competition among them for land not to live on but to HOLD and rent out, the solution is not to "increase the supply of land", it's to decrease the amount of billionaires...
Land is not actually that scarce. Land in cities is scarce.
We can build more cities. How do you think the current cities we have came to be?
Yes, it requires a longer term view than a 4 year election cycle. China has created hundreds of cities from nothing in the past 30 years. Many failed, many succeeded. This is the risk taking required to create new things.
We could do it too. But we choose to stagnate and inflate the real estate we already own instead.
Land in most cities isn't even that scarce if utilised correctly. Yes if you measure it as ground square meters it's quite scarce but if you measure it in number of apartments that could go there, there's usual plenty.
You are describing, almost fairly well, issues of accumulated wealth at the top -- particularly the effects of covid handouts[1], where all the money ended up with landlords that were not allowed out to spend it all, significantly increasing inequality and asset prices.
And somehow you try to turn it into an anti-tax argument, when taxing the wealth at the top is the only solution, the way to control the money supply, time machines aside. Give out too much you have to take some back if you don't want to devalue.
1. which is why the stock market and housing market went crazy at that point, not a steady 15 years, $5tn was created
This would be true only if the government simply increased taxes on assets (say, doubled capital gains taxes) and then took that money and paid down government debts with it.
However, we all know this won't be the case. The money taken from asset holders will simply be dumped right back into the economy via populist handouts, and that money will ultimately just re-inflate the things that are scarce and valuable (assets).
You're disincentivizing the creation of assets at the individual level (by taking them from people) and stimulating demand for them at the societal level by giving everyone money to chase assets with. A recipe for more runaway prices.
Again, there is only one path: increase the supply of assets (build more real estate and more competing businesses). China has similar/lower tax rates than the US in most areas while their middle class grew massively. This is how. It's not magic.
> The money taken from asset holders will simply be dumped right back into the economy via populist handouts, and that money will ultimately just re-inflate the things that are scarce and valuable (assets).
What's a "populist handout", specifically?
I wouldn't count the endless corporate and agriculture bailouts as populist, personally, but I'm curious if you would. As a salaried middle class American who owns real estate in the range of $500k in assessed value, the only handout I've ever received since becoming an adult half-my-lifetime ago was the Covid stimulus check. Every other "handout" that I've received was a tax break, e.g. the lifting of the SALT deduction limit.
"Populist handouts" (by my definition) are a boogeyman that I've never actually personally experienced in the USA, especially compared to (1) the tax breaks that are handed out left-and-right, (2) corporate and agriculture bailouts, and (3) the military-industrial complex.
My soap box: I think what made the New Deal so aspirational was that, yes, the government was giving people money, but it was also paying them to perform jobs that made the nation a better place to live. We have so much crumbling infrastructure. Instead of handing corporations the money directly, make them bid on infrastructure projects. Then heavily fund a department that roots out fraud and abuse and make it culturally taboo, through legislation, to perform fraud and abuse.
All those things you listed are populist handouts to specific congressional districts to please the extremely narrow interests of the people in that area at the expense of the greater nation.
This is my point. Any money that comes in will be spent on handouts at a factor of 2X the amount that actually came in, further inflating the deficit and entering the money supply to ultimately inflate things that are scarce and valuable (assets).
Also, I'm in full agreement that it would be great to invest in building infrastructure and rooting out fraud, but we just can't do it any more now that so many people's incomes are dependent on making sure nothing ever happens.
"populist handout" is a extreme, but you have to admit that French going nuts over raising retirement age when they have deficit 5% of GDP is also a bit extreme.
Maybe both sides need to step back.
But if you frame the debate as US-only then I basically agree.
It doesn't matter how all of the money ends up at the top. Yes, government handouts to the rich are a big part of the problem, but so are a lack of taxation on the rich and on corporations, a lack of competition, the fact that both corporate profits and inflation are high, and the lack of stimulus for the poor.
You seem to think that the government can only use its money for more stimulus for the rich, but that assumption is itself part of the problem.
If the government gives more money to the rich, whether through tax cuts, subsidies, handouts, or whatever, the "trickle down" lie has always been that that money will eventually help everybody, but that's only the case if that money ends up in the economy, paying wages, etc. But instead, that money is being used to buy up more assets. That doesn't help the economy in any way, it just skews it.
If instead, that money ends up in the pockets of people who need it, they will immediately spend it, buy food, buy other things they need, pay off debt, and all of that money creates demand, pays wages, and grows the economy in a way that helps everybody, and doesn't just inflate the numbers.
Further its because income generated by assets are tax-efficient in the western world (well at least North America) in a way that non-asset generated income (i.e labour income) is not. Naturally the inclination is to not just accumulate assets for their own worth but for more tax efficient income as well.
If asset-based income was taxed similarly or included in the tax base, you wouldn't have stimulus become inflationary in the way it is today.
Uh, yeah of course a person is super rich for having $1000 or $100 invested in the stock market.
No, of course a person is not super rich for owning their own home which is worth hundreds of thousands.
God damned young yuppies with their stock portfolios worth $20 000 or more! In the meanwhile grandpa is suffering in his house which is only worth $1,3 million.
If only we could take the money out of investment into businesses and put them into investing in real estate! That's where all the jobs are made who pays people's salaries. Can somebody call the EU to have them regulate?
Wait until it explodes to 3-4b shortly after IPO, then... I think that despite the heavy selloff in the AI trade this las month we are not yet past the peak frenzy of the cycle.
Perhaps they were expecting to be more oversubscribed? Like they increased the number of shares offered to keep a target amount of money. Likely they see the rough few months ahead. Ramp.com AI index shows spend on Anthropic AI was in decline/flat July 31 - Oct 31
Likely they're best bet is to weather the continued devaluation of tokens and cheaper models until they hit on a better business model, which honestly is probably not that hard.
The previous range was 70-78 and it priced above the target range, at $83.
So they are selling 20% more shares and selling them for 15% more than the midpoint of what they were targeting, because the demand is overwhelmingly high. They are going to make $20B more from the IPO than planned.
The rumor is they were oversubscribed by more than 6 times over. SPCX was reportedly only 3.5-4.
Oversubscription was kind of a self fulfilling prophecy. Large investors and institutions placed larger orders knowing the IPO would oversubscribe and they wouldn’t get their stated allocation.
It's a common pattern for IPOs, regardless of demand. Companies gauge the price range during the roadshow, make an initial announcement at a low price, then raise the target at the last minute going into the IPO to make it look like a hot offering.
I wonder if early adopters are overrepresented in Ramp's customer base, so if they start learning how to get the best bang for buck in their AI spending at the same time the laggards are starting to spend real bucks in AI, Ramp's data may be bearish while the reality is bullish.
Thanks for sharing. Something seems off in that data. It says in Nov, Anthropic's share of token volume is 55.1% and share of token spend is 49.3%. Given Anthropic tokens are pricier than other vendors, one would expect share of spend to be higher than share of token volume.
That's a good point, likely there are other discounts / subsidies being realized on accounts? I think from Ramps point of view it might be hard to distinguish these.
That being said, I have also noticed that at times the X axis dates of the weeks changes. In the past, like a week ago, I noticed that the "Week ending Oct 23" changed to "Week ending Oct 25". Which also noticeably changed the shape of the charts.
I work with data a lot so cut them some slack, but I do with they had more grounded information in the report (a lot of % and not a lot of specific values).
I also wish they would somehow tie in to the massive OpenRouter Rankings. I think the two together would help paint a much more cohesive picture of how token budgets are spent and change week to week.
And RH allocated me 1 pre-ipo share against my request of 12 of them. so one in a billion shares. RH shows I'm the 260,000th retail investor of the ANT offering (at least within RH ecosystem[?]). Just observing, but pre-IPO share allocation does not appear outwardly to be an activity rooted in free-market fundamentals.
Why is that a bad sign? I think it's weirder to restrict employee selling, makes it look more like you need to hide something.
Just allowing them to sell if they want doesn't seem problematic to me, unless they all (or a large number of them) actually sell immediately. Then obviously something's wrong.
Note that this is an actual question, I'm not even close to an expert with these things.
No, it's the opposite. If you truly believe in AGI the more reason to sell, asap ;) Money, at least, is likely to be worthless. Assets maybe not. Who knows what the future will hold, but all things considered it makes slightly more sense to hold tangible assets than 0 and 1s.
If everyone working there could cash out everything immediately, and be set for life (eight figures), I seriously think many would hand in their resignation and the company would face some operational crisis.
Insider selling coincides with lowered stock price pretty much universally. Lock in signals trust. It says to investors, we're in this together and you can be assured we're equally interested to take the stock price up.
The exact opposite. Companies restrict employee sale on IPO because they're worried that it will hurt post-IPO price if employees sell on the bell. This is a highly unusual show of confidence from Anthropic. Note that the IPO is already oversubscribed, so demand is well ahead of supply.
Say what you will about their financials, but this IPO is largely going to be driven by how much raw demand there is for AI. And that demand is absolutely insatiable, at least from where we’re currently standing.
The recent cybersecurity incidents will only increase that perception. It appears to be weaponizable, whether or not that perception is grounded in any kind of reality. Whether it exposes vulns, fixes them, or all of the above, it just translates to more demand.
I think this will go up for quite a while before the world realizes the limitations. At least until their first earnings report. That said, I’m not a gambler.
> this IPO is largely going to be driven by how much raw demand there is for AI
Institutional investors are going to be looking at profitability, not "raw demand". Demand can go through the roof and companies like Anthropic and OpenAI still do badly if token/task prices are dropping fast and/or they are are not competitive on pricing.
Individual investors may be buying based on "AI is the future", with no regard as to profitability, and will likely be burnt as a result.
Short-term post-IPO stock performance is no indication of anything - part of the job of the underwriters is to "stabilize" (manipulate) the price, so it takes a while for organic supply & demand forces to prevail, and one might guess that supply will go up considerably once insiders are allowed to sell.
I will consider a grabbing a small position. The chances it expires worthless are really high, but also the volatility could allow for securing options that may become extremely valuable.
I would trade volatility (either direction) more than any kind of bearish thesis right now.
The entire AI space is currently performing like Tesla. Unbelievable momentum, confidence, and capital. Day traders are going to have a field day with this, but one would need to have balls of steel to hold puts overnight.
I am very curious how a publicly traded AI lab will use model releases as stock pumps. For example, if you are in the lead, do you sit on a new model until the stock price falters, and then release?
Possibly not that directly. Because training is expensive and that affects your quarterly profits and that affects your stock price. If you can reduce how often you need to train a new model you lower your costs.
Not if the stock has priced in a certain cadence of model releases. Same as the expectation of a new iPhone each year.
Stock price will be directly linked to having the best model (which could give OAI an advantage as they aren't competing with shareholder interests - yet)
Their run rate is $85B and they've been growing 1000% year-over-year so even $200B is pretty unlikely. It depends on if you trust their profitability.
What do you think of Apple at $5T, Google and Nvidia at $3.5T, SpaceX at $2T, or Meta at $1.5T?
They have 18B fully diluted shares. That's about $5/share revenue, and the ipo price per share is $83. The math implies a P/S of 17.5. SpaceX is 60, Palantir is 55, Cloudflare is 40, and Snowflake is 18.
The law of large numbers aside, and ignoring the insanity of the rest of the market when using valuation by comparables, the numbers aren't absurd.
Referring to the company that is projecting around 0.3% of US GDP in annualized revenue by the end of this year, having started at 0.02% at the beginning of 2026? That seems pretty reasonable factoring in that revenue growth rate.
Adding in that only a quarter of US businesses are using AI right now, that the company has set its sights on fundamental biology research and robotics, the company seems undervalued relative to the upside case.
>I struggle to believe this is worth like 5% of total US GDP.
Actual question: Why? America is mostly a service-based economy, and a lot of service-based jobs seem extremely vulnerable to automation. Do you think that the possibility of massive AI job displacement is overrated, or do you think that competition between AI companies will stop Anthropic from capturing that much of the market share?
comparing a stock (ANT) to a flow (GDP) is a common finance mistake. You shouldn't compare a stock to a flow, they are measuring different things. https://en.wikipedia.org/wiki/Stock_and_flow
They either get RSI and take over the economy (in which case their shares are almost certainly worth nothing) or they don't and their shares go to ~0.
Their 'traditional business' path makes no sense:
1) they can never stop training or they'll be selling a commoditised (or just worse) product, so it looks very difficult to ever turn any kind of profit.
2) they don't have huge consumer awareness penetration and big business already trusts Microsoft, Google, Amazon way more so in a commodity world the big players win.
If revenue grows faster than training costs they’ll be able to amortize training eventually.
The real money here is in b2b. Consumers don’t pay for digital products. Even the software companies whose products are consumer facing mostly make their money by selling ads which is a b2b business.
In two dialects of English a billion is a thousand million and a million million, worlds apart, and nobody cares because they're both equally inconceivable
> Like 1.5T could make as many as 1,500 billionaires
I highly doubt that, given the founders themselves are "only" going to be worth something like $20 billion at that valuation. The shares have been massively diluted, and the vast majority of the gains are going to Google, Amazon and a large list of VC firms. The people getting rich are mostly already rich.
Anyone know why you would be limited to purchasing two shares of the IPO? (My sister tells me this—she's climbing aboard the ANT train and told me she was limited to two shares. Wut?)
That's probably just for the initial auction at $86/share? Once that's done there will be plenty of liquidity to pick it up at market price, whether that's higher or lower.
The IPO is 10x oversubscribed, which means there is a ton more demand for shares than the number that will be released today. Brokerages allocate shares from their limited pool, and generally prefer their more valued customers.
I think they allocate certain numbers of shares to different brokers. Maybe your sister's broker's allocation was small and split amongst many customers? I applied for $20k worth at IB and got 43 shares worth ~$4k.
I stopped buying IPOs a while back, and have had zero regrets. I don’t think I’ve seen a single one stay above the IPO price during the first year. Nearly all, especially for companies in growth stages, hit a wall where they’re still growing and burning cash… it causes the market to turn on them, and the stock drops to anywhere from 40-90%.
That’s not to say there aren’t a lot of winners even 1-2 years later but I’ll happily be sitting this one out (and wait for a drop or two). The market gives some pretty sweet deals for those who are patient.
Uber stock gained 65% since its IPO. Airbnb stock gained 29% since its IPO. Snowflake is up 16% since its IPO.
That's quite non-impressive when you remind yourself that these IPOs happened a lot of time (6+ years) ago, and that their stocks are highly volatile.
It seems to me the VC system got quite efficient to accurately compute the total addressable market well enough and inject their money to grow until that whole TAM is covered enough.
Then the VCs release the now horizontal stock to the public markets.
That being said, there are also counter examples, like Datadog (up 362%) or Crowdstrike (up 1377%), both in the same IPO class as Airbnb and Uber.
You’re missing my point, Uber is a great example of what I’m trying to say… after IPO it shed 50% of its value within six months and it wasn’t until 18 months later that it was back to IPO pricing and 24 months before it was above it. Buying it at IPO, you’d still be up, but you’d be up a lot more if you just waited six months or a year.
The same with Datadog, Robinhood, Facebook, Coinbase, Crowdstrike, and I’m sure I could go on. Even ARM was generally flat for six months after IPO… the only real winner I’ve seen recently was Reddit (which dropped a bit but was briefly flat before going up).
My point is buying the IPO instead of waiting will probably have you overpaying for it in the short term (1-2 years), long term it doesn’t matter (if the stock/company is making money). Most big tech IPOs have a sell-off, sometime within the first year or so, and it has been nothing but beneficial for me to wait.
I absolutely want to buy Anthropic’s stock, I’m just willing to wager I’ll get it at a hefty discount sometime in the next year or so… from what I’ve seen their revenue is nowhere near their valuation and we’ll probably see the market recognize that (even briefly).
This could cause an overall correction in the market I think. It's likely people will take money out of NVDA and other AI stocks to buy Anthropic's IPO. This could send NVDA and other AI stocks further downward and maybe even cause more panic selling. So the market caps of those companies decrease. Then that value is lost when Anthropic's IPO share price falls from insider selling, lack of interest, etc. In fact, if OpenAI has an IPO, Anthropic shares could be dumped to buy OpenAI. It could cause an overall decline in the market.
Yeah, I’m honestly more worried about a massive decline at this point than I am major upside. The “AI boom” has been the only thing really propping up the economy and I’m not sure it has enough gas at this point.
Longer term, 3-4 years I think it’ll be fine. I’m definitely bullish on AI in general but I definitely expect a rough 6-18 months to start sometime before 2028 (and it may have already started thanks to Jensen’s shenanigans).
They are only floating about 70b for public trading. This is very small figure compared to nvidia's market cap and really minuscule smaller compared to the overall market size. People are moving money out of nvidia for other reasons, but it's most certainly not because of this ipo.
Yeah.. the models themselves might not seem to have much moat but the teams that already set their workflows with claude don't easily switch. So I'm thinking the rate at which users maintain their subscriptions and developer habits are the real moat
I am worried, their business is more like webhosting and less like something that is unique. I rely extremely in AI. But there is Anthropic, Mistral, Claude, Deepseek...
It would not really be a problem if Anthropic is gone tomorrow.
If I wasn't personally convinced that they're deliberately nerfing their models, I'd buy the heck out of this. But I believe they are. And I also believe they are plateau-ing and do not yet know how to seriously advance the abilities of their models.
As a result, I think this stock (which has no voting right btw) is a major gamble and unlikely to pay off before they run out of money.
More importantly, I think Dario is dishonest, and trying to thwart open models which are soon going to be just as good as his models, but without the spying on what you're doing.
> how can they both be deliberately nerfing and also don't know how to advance their models
It's a fair question! Let me elaborate:
I think the best internal model development is basically plateauing and they don't know how to make large improvements now. Pareto has past, but it's worse. The next 20% will take more than 80% of the energy they've put in, and this IPO is an attempt to try and get there. I don't think they have good chances.
In parallel, the models they are shipping to _most_ customers are being nerfed for one or more of a long list of reasons you can read in my posts elsewhere.
I actually think the models are mostly great! But they are trying to save that tiny bit of extra for themselves since they think it's their only way to get ahead compared to others competing in pure software markets against them, which they want to own. They don't want to just sell you pure tokens, they want to build companies.
Before releasing a new “version” of their model they start serving a quantised version of the existing model which performs much worse. Then they release the “new model” which is just the old model in its original state.
so many reasons not to take this at face value - greenshoe, lockup periods, 5% float. consider the latter alone - in this math, the 95% unsold are considered to have the same value as the rarified 5%. that defies economics.
About 1.1T market cap will exist tomorrow prior to opening bell, out of a potential 1.5T that includes .4T stuck in options, convertible notes and such.
But it’ll likely open above 83 when trading actually begins. We may see upwards of 2T if it goes over 100 a share
They already have all the funds they need to finance their investment. This IPO feels like it's exclusively made for the existing stakeholders to cash out and not to raise money for future projects, which is usually the main aim of a public offering. Furthermore they have a great technology, but not great products and not a mature distribution system. As commodification continues, players like Google who can embed the technology in their products will gain ground.
Retail investors will be left holding the bag. This is the ultimate cuck tranche.
Here's something I don't understand that I don't see people talking much about:
Anthropic has raised about $141B.
But their S-1 also says they have $132B in cash on the books. So they've basically not spent hardly any of what they've raised.
Yes, they have huge spending obligations over the coming decade, but I don't understand how this fits the narrative that they're massively unprofitable and just shoveling investor cash into a furnace to stay afloat.
Regardless, they will soon endure crippling legislation and litigation. They will be forced to stall their ambitions and the timing for it all to work will collapse.
Lots of opportunity to make money with this until then, but it’s coming.
No doubt but many of the people being disrupted here are the voting bloc that these politicians need not only as voters but as donors.
As for the lawyers there’s essentially an endless tap of litigation to try here. AI harms people the more and more it’s made ubiquitous. Someone has to be liable.
It is not quite as bad as Meta. Meta concentrates shareholder and board-election control with one guy (Zuckerberg), whereas Anthropic divides power between its founders and a trust intended to protect its long-term AI mission.
Neither structure gives ordinary public shareholders control proportional to the money they invest. This should probably be made illegal.
> Neither structure gives ordinary public shareholders control proportional to the money they invest. This should probably be made illegal.
I would prefer it if people were still allowed to invest in such companies in the public markets, but perhaps we can make it so that inclusion into an index requires "1 share = 1 vote". That way, at least retirement funds are protected.
I would prefer the exact opposite: non-voting shares in indexes protected only by law. I don't want my index manager to act on my behalf. That's the exact opposite of why I buy index funds.
Meta is up 1,522% since IPO. The reality is founders care way more about the long-term health of the company than the typical outside board member that is just farming prestige and owns minimal equity. Do you really think the former president of Taco Bell International knows more about how to run Nvidia than Jensen Huang?
Yes he probably does because he knows not to commit treason and fuck spies. When’s the last time you heard the president of Taco Bell being black bagged at the airport? When’s the last time you heard of them getting fired for spying?
At this point I am 99% convinced that all major providers have agi equivalent performance when running as a 'swarm'. Simultaneously they all hamstring their AI to try to prevent having their own AI being used to build someone elses AI.
Codex is completely unusable at all for anything but small tasks when handheld by a person. Orchestration is simply unusable. Anthropic falls apart and refuses a critical set of things that are small enough to block you from getting it to build out real work. Like basic ci/cd server setups and using login creds provided on web, etc. Conveniently setup in a way that doesn't at all prevent a hacker group from at scale pwning people/corps but does prevent normal workflows. Want it to build a full stealth toolset for control of android phone over adb bridge? No problem. Now can you use it for anything? No I won't log in for you. Pressing the fucking login button is just something I'm morally against. etc. Basic room temp IQ computer tasks. Like they are trying to prevent someone from automating your average corporate workflow while it can do everything else trivially. Why? Because 80% of their revenue is from Corps letting employees burn $$$ tokens for every day tasks like proofreading an email.
I know this for a fact because of custom harness work in a corp with Microsoft EntraID SSO over a year ago where all of these tasks were easily performed by the old Sonnet model and I trivially automated every workflow.
And, strangely, when I do something that should make everything much worse, simply just tell Claude to use Codex as sub agents, performance gets drastically better, presumably because teams from both corps didn't have the chance to specifically fine tune their models to be hamstrung when run in this configuration.
It is obviously not about safety and has never been. These models still will easily find vulnerabilities. Just get some code you want to exploit and throw $$$$ inference at it to 'find all bugs confirmed via test case suite' etc. Then throw inference at targeted exploit scripts.
So far the pension funds have been major investors in everything AI, every fund is in it hard and deep now. It absolutely will be in your retirement fund, but don't worry your unlikely to be able to get to retirement unless its really soon.
The blast radius is quite large, but appears less contagious than what happened in 2008.
Lots of folks that bet big on AI are going to lose their shirts, but this doesn’t look like the “banks are failing every weekend and the global financial system is on the verge of collapse” situation we had in 2008.
Technically it’s not as bad. We don’t have the same AI industry so fewer companies to go bang, and traditional “legacy” companies will be havens for investors.
However… when America sneezes, europe catches a cold.
The biggest problem is that poor economic management (in the UK, austerity under Osborne and the Covid response) has eaten our margin. We are not starting from a position of strength.
Our VC culture is much more conservative, which is good and bad. It's partly why we lag so far behind in new technology, but it partly insulates us against high leverage default cascades. However the greatest risk is macro economic. Europe sells a LOT of products and services to the US. When they're hit by economic downturns, that affects us to a large degree. So the right question is: how exposed to a crash is the US economy?
Unlike 2008, the vast majority of capex has been cash, not debt. So this looks nothing like 2008. It also looks nothing like 2000. Growth is extremely high and real and organic. Real customers are paying for real products and services at an unprecedented rate. The risk will come from somewhere else entirely. Possibly something like a major technological advancement in AI which renders most data centers obsolete or unneeded. Or makes intelligence cheap and abundant for everyone, rendering Anthropic and OpenAI worthless. If that occurs it would cause a contraction in investment in the space, but the US economy remains fundamentally strong.
To be honest, I don't see serious crash risks right now related to AI. The companies taking the biggest risks have enormous revenue, profit, and cash reserves. They won't fail. I think the largest risk is actually inflation. Many governments are going to have to increase rates over time, and this has a mediating effect on economic activity. It slows. Not just AI, but all sectors. That could cause a recession.
For actual science done by scientists they're far behind open models now.
The people I know using this model are perpetually hamstrung by slowness, pedantry, misplaced assumptions and appeals to authority. You can't move forward if you're just trying to norm your training data.
They've created this extremely slow, overwrought system with lots of guardrails and token limits designed only to create addiction. Why would you use that when you can have a marginally less "smart" system that's 50x faster and 100x cheaper? Who cares if the model one shots something. It needs to be checked anyway, by test in a machine or a lab.
Cloud models have zero value if the findings of scientist are absorbed by these same cloud models and "by coincidence" then published before them. It is impossible to get ZDR as individual that you can trust.
Statistically how many companies that were first to any market the eventual dominating leader of that industry? To me it would still be like picking a single number off a roulette wheel because it is a very slightly larger slot to bet on. Maybe it will pay off, it might even be statistically the best choice, but it is still a slim chance overall that it will be the winner.
The specific models and capabilities are a sideshow.
And hence, the money flows all to the top despite viable alternatives...
(a separate market of workloads is contracted out to Fireworks, Baseten et al. and a third segment DIY it with rented or owned GPUs. But those customers are different buyer archetypes little overlap with those buying trust.)
Models are useless if you don't have the infrastructure deals to run them at scale. I know people that use cheap Chinese models and get good results from them. But they are paying a lot for self hosting those models. The models are cheap but the infrastructure isn't.
The reason I think Anthropic's position isn't that strong is that most of the money they raised will go into making other companies very rich. Because they don't actually own a lot of the infrastructure they need. OpenAI has similar issues. The companies that actually provide the data centers (i.e. hyperscalers), chips, gas turbines, etc. Will be on the receiving end. And of course some of those hyperscalers have their own ambitions on tapping into the same markets. The SpaceX pivot to data centers in recent years is particularly interesting. Anthropic is already paying them a lot.
The reason I'm fairly optimistic overall about the economy is all the spending that will result from these investments in all that physical stuff represents an enormous cash injection in the economy. The investment bubble may or may not burst but it won't wipe out the infrastructure. If Anthropic collapses, that infrastructure will just be gobbled up by whomever takes over.
I am just waiting for the time when all of us (already or soon) out-of-a-job folks will start reaping the "AI abundance" (that "AI leaders" can't get tired talking about, including our own Mr Huang) while some folks already make billions and millions in these trillion dollar IPOs (soon more to come).
Because if that doesn't happen then I have a feeling the world will proceed something like this: AIs paying for things to other AIs (as a circular gesture; possibly different from the circular finance of today) while some rich folks will sit in their safe island citadels or in NZ (etc) nuclear bunkers controlling the arms, ammunition, surveillance, medicine, finances, along with pretty much everything else of the world while the rest of humanity, the majority – who couldn't tap into that abundance, will finish each other off one by one – or maybe worse, they will be live to witness it all. I try really hard not to picture this post apocalyptic scenario but either all this IPO, moat, and whatnot are hogwash, or that is happening :)
My take on this is that Anthropic must be closed until they've sorted out their security problems. They and OpenAI have claimed repeatedly that AI poses catastrophic risks and that it's not safe.
Sorry but if a company insists over months that their products are unsafe and even hint they could lead to the destruction of humanity, then I don't understand why authorities don't act and stop the company until they can demonstrate their products are safe.
Remember the dotcom? Do you remember companies like AltaVista or maybe pets.com?
Anthropic and OpenAI are the AltaVista of the AI era. With first starter advantage but no real moat and very good at spending money.
I am still waiting for the Google of this era to show its face.
Edit: every time I question these crazy valuations I get down voted really quickly.
would be funny if it would be google again
It most likely will be. Or it will be Apple with everyone buying their machines for local models.
lol at Apple. They’ll be relevant for another year or so.
Huh? Are you drunk lol?
I havent used an Iphone since the 3G, but Im pretty sure everybody I know, especially normies/non techies who use Iphones will use iphone/mac products for the remainder of their lives. Not because they love the product per se, but its the only thing they know how to use and have zero interest in learning to use anything else.
> Not because they love the product per se
Not per se maybe, but customer satisfaction is extremely high with those products
How exactly will a hardware company go irrelevant in one year?
They won't. Some people just have a weird hate of Apple.
It's not weird, it's rational.
Hate is never rational
Who's going to displace Apple - who's even trying? The smart phone market is just a battle/division of two ecosystems: Apple vs Android.
Apple is as much a lifestyle/fashion brand as a technology one. It would take a lot more than a theoretical cool AI-centric pocket device from some wannabe competitor to displace them.
the new SGI, since someone has pointed out dotcom era.
I had Nvidia as the new SGI.
For this to be sustainable, I think the company needs to have serious revenue-generating products that can float the intelligence.
That leaves Google, Meta, MS
I think this is just a fluke, because their CPUs/gpus share memory.
Apple is still a luxury company, and over time competitors will offer more LLM bang for way less.
Apple is a luxury in some segments. But don't forget Macs are also computers of choice for majority of creative class and also lawyers. It's the same type of people who prefer or even legally require strong privacy. So buying top tier apple mac for running good enough private local model is exactly reason why would these people invest in such machine.
I don't think it's a winner-take-all market like search is. It's more like cars or computer chips. There will be a handful of more dominant players but plenty of competition and a multitude of large and small applications.
On the "echoes of dotcom" I think there is no way that Anthropic is worth $1.5T, same for SpaceX. Same for OpenAI. I agree that there will definitely be a Dotcom sized reckoning.
You cannot recreate mass production of cars quickly, it requires complex supply chain, many people, and large investments - look at Tesla as example. But you can absolutely create your own models, scale them up and offer services - capital you need for that is incomparably smaller, especially now as you can do that with AI assistance. I don't see how large software and AI companies can stay profitable in such environment.
I don't really know anything about IPOs, or the stock market, but it seems to me like there's an advantage to being the first to IPO. If this goes poorly, at least you're on the market. If this goes poorly for Anthropic, isn't OpenAI going to be in a bit of a pickle? It sets a bar for how highly OpenAI could be priced, and attracting new investors must become harder.
Yes, from the pov of Dario he is going to be a really rich man in a few weeks no matter what.
And in the end he will almost certainly be richer than Sam because he was out first.
I thought sam wasnt making any money from the ai thing he was just doing it all for humanity.
altman is just in it for the ram
Nowadays there is no advantage to IPOing since private lending is so big, so the only reason is to cash out on retail investors when your company is over-hyped.
Oh yes just go ask a few people for 200 billion dollars. Is the private market really at that scale? Public markets are still much bigger, no?
Well OpenAI just got a bit smaller private round than they wanted to, and there are plenty of news about frontier labs' bonds being junk rated.
So the point remain the same: maybe the IPO has to happen, and Anthropics' IPO performance matter for OpenAI, because all willing capital is tapped out.
An IPO is not a universally good thing. You sacrifice a significant amount of control once you go public. It's like a special form of bankruptcy if you look at it the right way.
Racing toward an IPO is more likely an indication of a liquidity crisis than about having superior product, prospects and roadmap. If things are going so well, why would you bring this amount of disruption to the company?
In that case, Google IPO'd 22 years ago, so both Anthropic and OpenAI missed out on that advantage, no?
Pets.com generated $5.8 million in annual revenue while losing $61.8 million. Anthropic has already generated $33.2 billion in revenue in the first nine months of 2026, with thousands of large paying business customers. The dot-com bubble was heavily driven by speculation about future profits, whereas today's AI boom is backed by enormous real demand, established profitable technology companies, and measurable earnings growth. It's growing faster than any other industry or sector in history.
To be clear, I'm not bullish on the AI companies themselves. I think Chinese companies will continue to distill these models and once they achieve RSI (which is imminent), intelligence will be abundant and cheap. The real profit will be in the other layers of the AI ecosystem: electricity generation, chip fabrication (memory, storage, CPU/GPU), infrastructure (datacentres), and probably most notably, the people and companies which produce compelling software and services using this technology.
> and once they achieve RSI (which is imminent)
Imminent according to which sources which are not financially involved with AI companies?
Nuclear fusion is also imminent.
The rapture has been imminent for 2000 years.
Everything is geologically imminent.
2000 years? We've had doomday cults since the Sumerians, 5000 years ago. I imagine we had doomsday cults when we invented religion :-)
Claude got really good at ML Engineering recently. If they can't get the current models to RSI, they aren't trying.
How much juice is left in that tank after 5 years of intense focus on optimizing LLMs is another matter. If we are on top of a local architecture optimum, speeding up the hill-climbing will deliver underwhelming results until someone discovers the next hill.
Why did you leave out Anthropic's $80 billion loss over the first 6 months of 2026? Is it because that would make it sound a lot like pets.com?
That includes paper losses due to the increased value of their equity. Those are non-cash losses and not really indicative of how healthy the business is.
Futhermore if you exclude any costs they actually would have reported a profit.
lol
peak HN here. Zero motivation to talk about specifics and just sneer
do you even know what that 80 billion loss was?
The weird thing is that 61.8 / 5.8 is around 10x the loss vs revenue, whereas 80 / 33.2 is only 2.4x more loss than revenue.
So there's actually a coherent argument that Anthropic isn't hemorrhaging money quite as badly as pets.com. But yes, I'm glad I'm not the only one who thought omitting Anthropic's loss this was a blazingly odd way to open an argument.
Demand is there, I just dont belive OpenAI and Anthropic can deliver in the long run. Not once token prices fall off a cliff.
Also, please ask Anthropic why their sec-1 filling is missing financial data after that quarter they made "profit". Most analysts belive that their profit was financial juggling and things went downhill really quickly after June.
How do you spell “GPU depreciation schedule” in 2026?
Yeah, I think Anthropic would be an amazing company, were it not for the fact that, for whatever reason, all of the other LLM providers are only a few months behind them. Right now, I'm using Opus 5.5 as the main model, Longcat 2.5 for all subagents, and GLM 5.3 for adversarial security PR reviews.
Longcat is probably the worst of all Chinese models, but it's dirt cheap, and it works well enough. So my usage of Anthropic's models (and overall amount of money spent on AI) has decreased drastically.
Longcat is a hilarious name - first I've heard of this
Spacex compute discount expired after Q2. They had to IPO after their only "profitable" quarter.
> The dot-com bubble was heavily driven by speculation about future profits
So is Anthropic at their current valuation; they are gambling that they will become a multi trillion dollar company soon. That is speculation, as they are not profitable yet and everyone just hope that they will succeed.
> Anthropic has already generated $33.2 billion in revenue in the first nine months of 2026
Profit statement would sound nicer. But ofc, "Amazon was unprofitable when it IPOd too" and $28M losses on IPO year is comparable (not) after you adjust from inflation.
Curious why you mention Pets.com revenue and loss but only Anthropic revenue (and not commitments, costs, etc.).
You shouldn't be downvoted. AI could be world-changing technology, like the railroads, flight, and the internet, and still leave a list of failed companies in its wake.
Airlines are a particularly interesting case since they enable so much economic value, but capture very little of it for themselves. On the AI side, the lack of moats and open models feels like this is the direction we may be trending as things settle out.
Airlines capture quite a lot, but through partnerships, loyalty programs, referral fees, and through partner networks. The same is going to happen with AI: inference is going to be cheap and plentyful, the money will have to come from the harness, the ads, the integrations... and Anthrophic doesn't have that worked out very well imho.
I’d argue Anthropic, with its push towards productization, probably has it better than the other AI companies with the integrations (with a lot of space for improvement, nevertheless)
An interesting comparison. I've also felt that open models are being currently overlooked
> AI could be world-changing technology
It already is changing the world quite a bit, just not for the net better.
Millions of people are using it in many different fields, and are constantly finding more uses. The only downside is that website developers must defend themselves against Anthropic.
> Millions of people are using it in many different fields, and are constantly finding more uses.
That doesn’t mean they’re positive for the world.
> The only downside is that website developers must defend themselves against Anthropic.
I’m pretty sure the parents of the teenagers who killed themselves, the communities who must deal with the noise, increased energy prices, and lack of water, the people who lost their jobs, the people who are feeling the environmental impacts (which will eventually include you, wherever you live), those who died due to (mis)targeted weapon strikes (and their families), artists all over, those with spouses who grew distant, those arrested by incorrect facial recognition, victims of deepfake nudes, and many many many more would disagree.
Hey but at least it can kinda/sorta one shot simple coding projects!
Yeah, I also remember how I ended up working several months for free without knowing if our start-up would survive the crash.
It was enough months that per Portuguese labour law, I was able to leave the company without notice.
I don't know any of these laws since they haven't been relevant to me, but it's crazy that you can't just stop working immediately if your company stops paying you.
Don't know for Portugal explicitly but labor-employer relations have historically been quite different in Europe (and across Europe, since it's still different countries).
People "showing up" for their employers to weather storms wasn't uncommon in the 80s/90s before the stock markets and globalization created a larger disconnect between where profits went and thus also employees, them showing up could thus translate to bonuses,etc down the line (and probably that they had been getting them during "good times").
Some of it is still codified, in Sweden f.ex. there's a "Salary Guarantee" that the tax-agency will provide backpay to employees in case of bankrupcies (to combat possible fraud, the flipside of it is that even in LLC's C-level and boards of employers can become personally on the hook for such debts making them less risk-prone and more likely to file bankrupcies instead of trying to save companies).
So that's one way these different breaking points are, bankrupcy itself, but also the line where salary guarantee isn't given anymore (when you're just supposed to go since the state won't reimburse you further) and when people lose trust in any possibility of a rescue.
>before the stock markets and globalization created a larger disconnect between where profits went and thus also employees
There is a greater connection between profits and employee compensation in publicly listed companies on the stock market than there is for non publicly listed companies. All those RSUs that allow people to earn so much working at tech companies for the past 2 decades are because of the profits.
Not to mention that companies listed on the stock market have to publicize their financials, whereas employees of private companies don’t even know what the profit is.
In most countries, labor laws are generally a double-edged sword. If they are not able to leave without notice, that usually means they also can't be fired without advanced notice as well.
What are the consequences of the employee not showing up to work?
In Portugal: in a liquidation, they may loose first preference for back pay.
Back pay first, then social security, then taxes, then secured/privileged/unsecured/subordinated credits.
I had an employee in Czech Republic that didnt show up for 12 months. HR still trying to fire them. They collect full pay. I have 2 others in Slovakia with similar situations.
One company I worked for that I wont name, was so sick of its Italy site that they turned them all into project managers hoping they would quit out of frustration. Some did, many didn't and it just went down hill from there.
Neither company, my old or new, will hire out of Europe anymore. Its just not worth it.
> I had an employee in Czech Republic that didnt show up for 12 months. HR still trying to fire them. They collect full pay.
Sounds like a fantasy, or HR is protecting the employee. In Czechia, you can put employee on PIP (by delivering a written warning) and then you can proceed to termination when there is no improvement in a month or two. There might be some notice period, though (1-2 months).
BUT not showing up at work is a gross contract breach, so termination can be very quick (immediate). [§55(1)(b) - Czech Labour Code]
In my country (Kazakhstan) I think the only consequence would be a record in your employment book that would say that you violated part of your work agreement.
And then when you would search for new job, they might ask you about it
It is the same set of laws that protect you to be fired on a whim, like standard practice in US.
It goes both ways.
To be more precise, back then it used to be more than three months without pay.
The Google of this era is a seedling in my care, but I'm just busy digging my technical moat deeper and deeper while i wait for the clowns to be sent packing.
AMA!
You are not questioning anything. There just is nothing interesting about the comment.
But alas, no downvotes this time!
> Anthropic and OpenAI are the AltaVista of the AI era. With first starter advantage but no real moat and very good at spending money.
Disagree. Google's search algorithm / variants were available for people to replicate. But there is only 1 google today. There are many office suites but people are still stuck to Microsoft Office -- many such examples.
While Anthropic and OpenAI have no technology moat people will become habituated to their user/api interfaces and rich features (which will take competitors a long slog to build).
Google came after AltaVista and Yahoo were already quite established. Google were quite late to the party and won on doing search much better than what came before them.
The reason why Google was #1 was they had a product that was ultrafast and without ads, so anybody who saw it switched instantly, there was no reason to look for another option. As for office suites, Microsoft won because of their aggressive tactics regarding format compatibility.
None of these apply to LLMs. I do enjoy cheap subscription to use frontier models, but if tomorrow I need to switch to the so-called open models, I won't regret it that much.
The lock-in you're missing here is that businesses use LLMs in conjunction with collaboration. They're building one-off apps that live inside the Claude GUI and sharing them with teammates.
To move that to an open model, you need to replicate the software surface (collaboration, functionality, permissions, etc.) and then provide a way to migrate the existing artifacts/data/history. Migrating developers is the easy version of the problem, migrating everyone else is potentially more difficult.
And then, it has to be trustworthy. Do I trust the provider with my sensitive business assets? If we are spending (say) $50k/yr on this, is it worth me risking my job by switching the company to a no-name brand to save $25k/yr?
The lock-in effect is stronger because so many engineers don't think it exists at all.
AI is what allows you to create and replicate almost anything digital using very few resources and quickly.
Where is the moat?
It's just crazy what is in people's minds.
The moat is, as it's often been before, brand and trust. "No one ever got fired for choosing IBM" will become "No one ever got fired for choosing Claude Code"
If people learn more about their cultiness that may change.
The actual moat for now is compute
I tend to agree that there's no moat in models, which is basically the entire business model for OAI and Ant today, but I do think there is still some money to be made there and it's likely they'll create some useful products around their models.
I don't think these companies are as bad as some dot-com companies, but I personally struggle with coming up with an investment thesis at the current price given it assumes they'll retain pricing power for models. And another problem is that this isn't like Amazon or Facebook where you build a decent product and with a few tweaks every couple of years people will continue to use them for decades – these companies need to develop and release new competitive models every month or two just to retain their market position.
I think the only thesis that makes sense is that the singularity happens, the private companies that owns the models which are produced as a result retain legal ownership of them, and that humans retain control over the models. In my mind I'd give this outcome a sub 1% probability.
Moat = data. They already have all your data. What did you think the moat of Google was?
They have a huge moat, unfortunately. The last models are steps above any other company out there.
The newer Chinese models get somewhat close, but OpenAI & Claude got better at stopping them from copying the model so the Chinese models will not make such big jumps anymore. And most Western companies will not use Chinese models if they are not open weight.
If they keep going like this huge parts knowledge work sector (legal, marketing, software, ads, health) economy is AI driven in 2-4 years and OpenAI & Claude will own it. Another huge shift of wealth from the Western World to Silicon Valey.
I don't think having a better model is actually a moat
It used to be when things moved slower. Now it is a temporary advantage which just makes it an arms race.
The companies that win are those that workout what keeps users/orgs that spend money and want to keep spending (or integrate so it is hard to get out)
I agree, as a private user of Claude I don't find the newer models rather useful. But rather a better moat would be something like - a better deal (more credits, cheaper etc.)
Thats just silly. Yeah having a slightly worse model doesn't matter if its still capable. But the latest claude models are getting so good that they actually can run without human supervision in highly complex tasks. I haven't used them enough, but my guess is OpenAI is also getting close to that. For every test I run with Chinese models, it just randomly goes the wrong direction every now and then, making it untrustworthy. That is a huge difference.
It depends on what tasks end up producing the most revenue I suppose? Will it be these long running complex tasks, or a series of short tasks, where the difference is less apparent?
Exactly, we know that "Bagger 288" can handle a lot, but not every one needs that much. Cat 305 CR might be just enough.
Bagger 288 - the heaviest land vehicle in the world
Cat 305 CR - widely recognized as one of the most versatile compact excavators
I had no idea what "Bagger" and "Cat" were at when reading the first paragraph and assumed they were hypothetical names of models meant to riff on how the names are just arbitrary words. I assume based on the later details in your comment that "Cat" is the company name (since I'm aware that all of the Caterpillar vehicles have "CAT" on them), so you've also indirectly provided some genuine commentary on the naming schemes (since at least the vehicle manufacturers have the decency to just use a numeric code after their company name instead of a bunch of arbitrary words that have no easy way to compare between reach one other than memorizing which one is which)
Having a good product doesn't mean you have a moat. Think of a car manufacturer. The only moat Ferrari really has is their branding, and maybe a few intangibles (e.g. patents).
Anthropic doesn't even have a brand worth protecting.
A big difference isn't the same as a moat. A moat would be a reason that others can't turn around and copy them.
Is this the first time in history that every time someone finds a new valuable use for this technology, I can use that very same technology to copy the moat?
Maybe oscilloscopes? Decompilers?
I agree with you that the two are way ahead of other companies (apart from google) with regards to model performance. but I doubt everybody will always need the smartest models - for most of the rudimentary daily work in the economic current models are already more than good enough. if openAI and anthropic can use their advantage to offer this rudimentary capability at much lower prices than other competitors then they have a real moat. but just having the smartest models by itself isn't gonna justify these valuations.
Unless prices go crazy, it does.
I have my own little app / ui that I work with and have all models, also round robin to test models, but GLM and Qwen just make weird mistakes & misunderstand the repo like Opus did last year when it first came out.
Where Fable 5.2 and Opus 5.5 are just always spot on. I dont need context, I write garbled prompts and it just gets me always. And it knows better and better, where it should push back & go into more detail and when it should just only do what I ask.
If Open weight models can reach Opus 5.5 level of consistency, I agree with you. If they can't I don't.
It’s not if it’s when. Opus made these same sort of mistakes until relatively recently.
> but I doubt everybody will always need the smartest models - for most of the rudimentary daily work in the economic current models are already more than good enough
Yep. Actually, I do most of my coding work with GPT Luna. It's cheap and good enough.
I agree they have a moat. But it's not due to model outperformance as much as first-mover advantage turning into a distribution advantage.
Claude is a known brand and installed product for a decent fraction of the economy. To avoid churn, Anthropic doesn't need to stay way ahead of everyone else. They just can't fall behind. The latter is much easier than the former.
Note: I don't have a strong view on this IPO. Don't trade on it based on my comments.
> economy is AI driven in 2-4 years and OpenAI & Claude will own it
Not really though... OpenAI and Claude will own the US economy in less than 5 years is crazy talk.
My experience is that small businesses have defined computing needs and dislike tech changes. If they get something that works they don’t need progress.
On the one hand this will help the ai companies reduce their tech spending but on the other hand smaller specific models is all the customers need.
I don’t think they have any moat… I think they have a noose…
Cheaper open weight models have a potential to undercut them especially when people realise what they can do with these models
This leaves OpenAI and Anthropic to always be chasing the frontier which is going to be increasingly expensive and they’ll never get to stop
I don't know if people even need the best models anymore. Some people I've worked with have expressed a preference for older versions.
Things have gone very very fast. Two years ago, you couldn't reliably have AI write you a medium sized code project. It would spin in circles and apologize. Now, an old model can write it with only a little direction, and the programmer can have a reasonable preference against the newer model.
It seems to me open-weight models will put a floor on the pricing. They're slightly behind, but slightly behind last year meant nothing worked, now it means "just as good".
In general I think there's going to be diminishing returns on any given aspect of LLMs because not everything can be infinitely improved upon. If something can write a medium sized program near flawlessly, and I'm writing medium sized programs - then nothing else it can, or cannot, do matters.
Maybe bigger models will always just get progressively better at writing out bigger and bigger programs to the point that we're one-shotting fully functioning operating systems with wide driver support in the future, but again these features just don't matter if that's not what you want to do.
I think we're already limited by our own desires when it comes to coding.
The reason you can't one-shot something complicated is that you haven't generally considered all the important choices. You have to be asked for every choice, and complicated things have a lot of choices. But I think you can do something waterfall-esque, which used to be a big no-no in programming, if you just list out all the decisions meticulously.
I've had some success with doing that, just talking it over and thinking a lot about design decisions, making a document, and then letting it loose on the implementation. Inevitably, you find that there were contradictions or missing choices in the design, and you have to answer some questions.
So, not one-shotting, but it's my fault, not the AI's fault.
You can of course tell it to just fill in the gaps with whatever it thinks is reasonable, and then you'll have something potentially quite different from what you imagined.
The big improvement still coming down the pipe is that the models have better "taste" about filling in the things you didn't meticulously specify. A model from a year ago could make an amazing game, but you had to specify every detail precisely; now you can get similar quality results with a much more vague prompt.
That was the idea behind waterfall. You can look up how good that worked.
I think LLMs can be improved much more but most likely significantly diminishing economic returns on the investment for that improvement. Ultimately, it becomes too expensive for a few % more improvement at increasing cost. I believe many of the LLM providers are approaching that problem. So it's time for them to commercialize viable long term business, maybe keeping a team still trying to dig up improvements without blowing the bank.
I agree - for the majority of people, there really is a hard limit on the technology. (Can it do what I want it to do in a reasonable amount of time with reasonable accuracy)
Many of my coworkers liked Opus 4.8, it was the moment they could actually use the models more reliably.
We're there, and moving forward doesn't really get us anything better. Most people are using them to format documents, spit out spreadsheets, pul together data points, and ask the models search questions.
We can already do that, it's only a matter of time before we realize we're paying 200/Mo for Jane in accounting to handle some of her work, when 5$ / Month would render the same benefit.
MOST people simply do not need frontier level intelligence.
The new models can pretty much run without human review. So yeah that flips the perspective: do we still need a human for that task, not which model can the human use.
I also expect AI to be more for machines and less for humans.
These providers are not selling chatbots that consumers can use as personal assistants anymore. Not even selling coding assistants. They want to manage the digital layer of every industry in the world.
yes indeed, and the models are already good enough, I dont think people fully understand yet how big of an impact this will be.
> put a floor on the pricing
ceiling
Exactly. After horses, the Ford Model T [1] looked awesome. After a few decades, supercars started coming out. While it is nice to have a very fast Corvette, most people prefer to use a more basic SUV or sedan that is good enough for their day to day priorities.
[1]: https://en.wikipedia.org/wiki/Ford_Model_T
You beat me to it!
I was going to comment that most of us don't need a McLaren supercar (or a Hummer) for our grocery runs -- that Prius actually is Good Enough.
>> They have a huge moat, unfortunately. The last models are steps above any other company out there.
A "huge moat" is not being incrementally better than one's current competition, not in tech. Anthropic's moat could be eliminated by one great model from any number of competitors, which means it isn't much of a moat. A proper "huge moat" would be a bundle of patents meaning nobody else was allowed to compete (apple/android) or a decades-old network to lock in users (windows). Anthropic has fanboys but most users would drop it in a hot second if a slightly better model appeared at a similar price.
> If they keep going like this huge parts knowledge work sector (legal, marketing, software, ads, health) economy is AI driven in 2-4 years and OpenAI & Claude will own it.
Exactly. If you haven't embedded AI in your workflows yet then it's just a matter of time or you're going to crash hard.
It's clear that it can do the work of several people, faster, cheaper and better and that will never not be the case.
They could triple the price (please don't) and I wouldn't flinch.
Again, I think people overestimate how much these models disrupt other jobs. It is good at code and technical writing, but in other fields I am not sure it is as helpful. And in those fields tripling the price would make it an absolute no.
No opposite, coding is at the forefront, everything where AI can verify if it's output is correct, and law is very easy, health also. Will all be as good as coding is now.
> They could triple the price (please don't) and I wouldn't flinch.
> They could triple the price ... and I wouldn't flinch.
> (please don't)
If you wouldn't flinch, why do you care if they take more of the froth? It's all worth it to you at 3x what it currently is, why object that they take what the market will bear?
> They have a huge moat, unfortunately. The last models are steps above any other company out there.
But "they" here is Anthropic and OAI together, and you probably need to add Google onto that list too, now that they appear to have got themselves back on track.
If it was just Anthropic, sure, this would be valid. But with 2-3 companies, we've got constant price competition bringing token costs down (as we are currently seeing) and no guarantee that Anthropic will ever come out with a clear advantage. If it's a three-way split, then that list price seems hard to justify.
And this is all assuming that the Chinese models don't work out how to get back to near-parity, which is also a massive gamble.
Well there is also the rest of the world. And me as an EU citizen doesnt want all this power concentrated in US silicon valley.
I don't think they have a huge moat.
because remember good enough is better than "better".
my take Microsoft will be the bigger win, followed by Palantir due to Microsoft understanding how businesses small, medium & large enterprises use software. Microsoft now has its own dedicated FDE org - & for "a.i" to pay off you need a lot of FDEs - since the gains are all on the application layer. the models have become largely undifferentiated
so whether microsoft uses their own models or openai's - if their own models are good enough it doesn't matter they suck by a few percentage points against the frontier they will win.
Google is top of the benchmarks at the moment, and Meta is doing very well too, with this just being the first iteration of the Wang-era models. Grok is lagging a bit, but probably nothing money can't fix.
For many use case the models that already exist are more than good enough, if not entirely overkill to be replaced by cheaper alternatives like Jev.
> The last models are steps above any other company out there.
Google and Microsoft just released competitive models within the last month. Opus 4.8 was just released in May, and there’s probably 4+ non-Open AI/Anthropic models that have overtaken it. You can’t build a moat around a six month technical lead.
That's not a moat, I can use Claude at 9am, Codex at 2pm and Kimi at 5pm without any issues.
>The newer Chinese models get somewhat close, but OpenAI & Claude got better at stopping them from copying the model so the Chinese models will not make such big jumps anymore.
I missed that, do you mind giving more details about how it is harder now for Chinese model? What was put in place but more importantly which proof we have it actually is effective?
A moat would be something that would allow them to slow down. They can't do that. A moat is something that can't be replicated. The performance of their current models will be replicated at some point and sooner rather than later. They can't stop because I don't think they even know how long it will take them to get some better model.
What they have right now is enough money to to work through that. But they don't have enough revenue to keep this up forever and they can't stop either.
Moat for what? People don't need cutting edge models.
> Anthropic and OpenAI are the AltaVista of the AI era. I am still waiting for the Google of this era to show its face.
What's your reasoning for why these two are AltaVista and not Google [of the AI era]? How would you differentiate without hindsight?
I was there. Google had a better product, and then - some time later, mind - they discovered that ads work really well for them.
OpenAI and Anthropic have a good product and they (especially Anthropic) are increasingly good at generating revenue from it.
What's missing, in your opinion?
Anthropics last year's revenue $4 billion revenue and $42billion net loss.
It's entirely speculative.
I'm not saying it's bad just people have valid reasons to be skeptical.
Their actual operating loss was ~$8B on $4.6B in revenue. Being skeptical of the valuation is totally fair, but throwing around the $42B/$102B headline figure without understanding the convertible debt accounting is disingenuous. It's the same story in 2026: ~90% ($94B) of their $104B nine-month net loss came from convertible financing charges, while revenue jumped to $33.2B.
Anthropic already turned an operating profit in Q2 2026, and it has the potential to be insanely profitable in the coming years. Calling it "entirely speculative" just isn't accurate.
$42billion-$8billion. That is $34billion in speculative capital expenditure on a business generating $4billion revenue and operating losses of $8billion. It is overwhelmingly speculative.
Yes Q2 operating profit is great, but you cannot ignore speculative capital expenditure which is significantly larger than you'd traditionally expect even for growth stocks.
My recollection is the Q2 operating profit was due to SpaceX compute discount (which went away after Q2 and is why the IPO had to happen now, although that is unproven speculation on my part).
The bigger concern (and what I think you allude to) is the $901B in purchase and lease commitments.
100% agree.
committed and I believe unfunded (well less unfunded now with IPO).
Weren't they only profitable in Q2 2026 because SpaceX was giving them a discount on compute resources?
These are 2025 numbers, which doesn’t take into account the ridiculous growth in revenue this year. Their net loss in 2026 was around 100B but subtract 80B from that because of the convertible notes “loss” (gaap accounting says that bc they sold shares to Google et. al. via convertible notes, and bc they rose in value, that counts as a loss to the books, even though it’s not actually money out of their accounts), and the fact they posted a 900M operating profit (!!) in q2 this year, and these numbers seem pretty reasonable? I mean why base your viewpoint on 2025 numbers when everything changed in 2026?
When everything changed in 2026 then maybe growth from past years is not that valuable indicator?
And then the revenue multiple starts to look a little bit insane.
If we don’t consider past years we’re talking about investing in a company with less than one year of track record to predict the future
Operational profits don't negate capital erosion; because share price is essentially a measure of capital, those paper losses must be accounted for in the valuation.
But yes your point on 2026 revenue is a valid one, although the forward looking trend changes quite dramatically looking at their monthly ARR that they've been reporting.
having a claude sub is like having netflix today, everyone and their aunts have a pro plan at least
You're living in a bubble. Not even my tech coworkers have a private Claude account. None of my immediate family uses anything other than Googles AI summary from searches.
If that was true nobody would be questioning the viability of these businesses.
Google IPOed after the custom from dot.com had mostly settled. At a massively more reasonable valuation as well. Antrophic is priced as if its guaranteed to grow at considerable rate and massively improve it's profitability in the near future. So it's not obvious there is much value remaining if you buy at the IPO price (Google was complete opposite in that sense)
Why is IPOing at a lower price better for the company? They could have fleeced a bunch of initial investors at the height of the bubble and had a huge mountain of cash to weather the downturn. Isn't that better?
Because Google was clearly a growth stock. Anthropic has one product in a saturated market and is priced so high that only early investors can profit.
Also, it steals IP.
> Google had a better product
> OpenAI and Anthropic have a good product
> What's missing, in your opinion?
I love that the answer is in your writeup.
As others have mentioned, it's hard to see how Ant and Oai are much better than the competition.
ai it self is like search, good digital tool empowering bullshit jobs.
to really extracting more money, they have to be good at creating drugs and weapons, and monopoly the way of doing it. ant founders understand it, I think it won't take them too long to get there.
> Google had a better product, and then - some time later, mind - they discovered that ads work really well for them.
don't forget it was text ads too. Prior to that all ads were "punch the monkey" giant banners, images, and flashing text. Google put unobtrusive text ads on the right hand side of the search results page and users didn't mind. IMO that was the breakthrough that turned on the money hose, Google found a way to get ads in front of users in a least annoying way.
I was laid off for a horribly long time and was finally contacted by a friend at DoubleClick to come work for them -- at the time all those horrible ads you mentioned were their bread and butter and I wanted no part of it.
The same day I got an offer to work at Genentech in their IT support and chose that as it was a real job at a real company.
2 years later I had to move on because corporate politics suck and my friend was now working at Google via the acquisition. He's been retired for many years and I'm still grinding away. At least I get to grind...
> I was there.
Me too.
> Google had a better product, and then - some time later, mind - they discovered that ads work really well for them.
> OpenAI and Anthropic have a good product and they (especially Anthropic) are increasingly good at generating revenue from it.
> What's missing, in your opinion?
Google had a product that was miles better. If you were using Altavista or Yahoo to find something, chances were you'd never find it even though it existed. With google, you almost certainly found it, sometimes even on the first go ("I'm feeling lucky").
Altavista and Yahoo tried to curate the web for users. The wanted to be the internet for users, they had guardrails! (Sound familiar?)
Google went in the opposite direction - they tried to get you what you wanted and ignored things like guardrails (Still sounding familiar to me).
I see the analogy, with ANT and OAI wanting to curate the AI experience for users, and preventing their users from doing anything that goes against the companies' opinions. The open-weight models are roughly similar to year-1 Google: curate nothing, don't get involved, and don't push your opinion on your users.
Maybe ANT and AOI will change their tune before either of them manage to lock up AI into a monopoly, but I doubt it. They aspire to be like Google: get the monopoly and then abuse the fuck out of your users for profit. Then abuse the fuck out of your paying customers for profit.
> Do you remember companies like AltaVista or maybe pets.com?
Compaq sold "83% of its AltaVista Internet search engine to CMGI Inc" for $2.3bn [1]. Four months later, ahead of its planned IPO, it disclosed $54mm in revenues [2]. 43x.
Pets.com IPO'd with $5.2mm of sales [3] and a $290mm valuation [4]. 56x.
Anthropic is going public at 40x trailing revenue (AltaVista territory) and 22x Q3 annualized revenue (not comparable).
> every time I question these crazy valuations I get down voted really quickly
You're the top comment. And this is against the guidelines: "Please don't comment about the voting on comments. It never does any good, and it makes boring reading" [5].
[1] https://www.latimes.com/archives/la-xpm-1999-jun-30-fi-51508...
[2] https://www.cnet.com/tech/services-and-software/altavista-fi...
[3] https://www.cnet.com/tech/tech-industry/pets-com-raises-82-5...
[4] https://www.sec.gov/Archives/edgar/data/1100683/0000891618-0...
[5] https://news.ycombinator.com/newsguidelines.html
And you need to factor in Anthropic's growth.
That’s true; public sentiment about AI is already trending downward, and I suspect we’ll reach a point where these companies start experiencing steady negative growth, once people realize the tech is only ever going to be great for certain specific use cases (e.g. coding). So Anthropic seems equal to or worse than AltaVista from that perspective.
I don’t know, every professional lawyer, doctor, accountant, nurse etc is using AI now in some way. Every person under the age of 60 is using it.
It’s going to be embedded in our professional class and we will have to see who wins. I agree there isn’t as much of a moat for any specific model here, it’s really the last mile who will capture the user/integrations.
> I don’t know, every professional lawyer, doctor, accountant, nurse etc is using AI now in some way.
Can you give names so we can sue them for HIPAA and other violations?
HIPAA only protects identifiable patient information. A general clinical question, or a case that is properly de-identified, isn't covered.
In America where HIPAA is, anyone can sue anyone for any reason.
If you want to sue your doctor for HIPAA violations due to their use of AI, you can!
You'll lose, and have to pay, but you can still do it right up until you get deemed a vexatious litigant.
How helpful, thanks!
Both OpenAI and Anthropic have HIPAA-compliant offerings.
Yes, I know a few myself. LLMs are useful but the huge distruption has to be slowed and thought out. These companies are pedal to the metal, they don’t care about anything as if laws don’t apply to them. And we should still keep our people educated, now people are wondering if education is worthless since the LLMs took the limelight. This can’t lead us to a good place. Profit is driving everything.
the best thing antrhopic/openai could do would be to get rid of their free tier and focus on enterprise/b2b.
it'll cut down the school cheating and the laziest, highest-visiblity slop on social media that is driving so much negative sentiment. AND they dont have to worry about providing capacity for people that ndo not (and never will) sign up for a subscription.
Don't worry it's coming. This is the same play that every other start-up uses, get them hooked on it while it's cheap/free, start charging them once they can't live without it.
Yeah, I hear Google is putting the finishing touches on finally charging for search after 28 years.
EDIT since I can't reply: the whole "if you're not paying, you're the product, not the user" is of course true, but totally orthogonal to your point about getting them hooked on free and then starting to charge. And ChatGPT already has ads, so your original point really makes no sense even in this context.
Perhaps you are confused, Google has been charging for search for 28 years.
But it's advertisers, not you, handing over the cash. You are handing over your data.
People will hate AI and use it daily regardless. I hate the smart phone yet here I am, it’s too useful to avoid.
Sounds like a great investment then.
Anecdote: I produce code for a living, and I use AI for that task. I don't generally use it for anything else -- I can count maybe a handful of times that I've used it for something other than my job or personal coding projects.
So when I retire, or if I were to switch careers, I don't think it's necessarily true that I would use it at all, let alone daily.
They use it like Google, just kinda there because it's cheap and easy. Not all in omg I need a max plan like some devs.
Can the AI companies be as successful as Google ended if with web? Maybe, just feels like the ad model will not work with these use cases like it did with Google 25 years ago..
> Every person under the age of 60 is using it.
I don't think this is true, or is a reflection of being in tech (assumption on my part about you, correct me if wrong).
Anecdotally, my peers (26ish, so gen z), none of them use AI except me, and that's for work as a SWE. My 68 year old step-dad uses it, but I suspect that's out of boredom more than anything.
I know students are using it for tons to essentially cheat, but I don't know anyone in that cohort using it for personal use of any kind.
> I don’t know, every professional lawyer, doctor, accountant, nurse etc is using AI now in some way. Every person under the age of 60 is using it.
This is obviously false on its face. For example, my household consists of four people all under the age of 60, and only one of them (me) uses AI at all.
And I don't believe my accountant uses AI, at least not for anything of importance. If I found out they were, I would fire them immediately.
Well, as a lawyer, for the past several years, people, especially here, insisted that I'd just be out of a job entirely because of AI. Now AI providers come to my office to beg me to sign up for their products, which apparently cannot replace me. The idea that I need cutting edge AI as opposed to a more specifically well-trained model is also just as silly. Harvey is the company that will benefit from law firms, not Anthropic and not OpenAI
Public sentiment against data centers is definitely trending downward, public sentiment against AI is a little trickier. A lot of the anti-slop crusaders are being forced to admit that AI output is getting pretty good, and it's slowly making inroads in more industries and roles. I think there's a begrudging acceptance from haters that generative AI is just how things are going to be from now on.
echo chambers are a helluva drug
Agreed, folks should read stuff besides HN.
I would have expected on HN some interesting figures on AI use.
Last gallup survey says :
- Frequent workplace AI use rose to 32% of U.S. employees in Q3 2026
- daily users is 16% and total ai use among employee is 53%
The adoption curve is interesting. Total use seems a little bit to plateau but not frequent use which is still linear upward.
https://www.gallup.com/workplace/716585/ai-changing-employee...
Edit : focus stats on fréquent/daily/total users.
...and their ballooning costs.
I heard a news bite this morning, something like
> Anthropic has burned $9 billion in the lifetime of the company, and will have $200b+ in cash post-IPO
No one is doubting their ability to waste investors' cash.
I think they burned more than 9 billion, they had a loss of 8B last year and 3B the year before[0], presumably they had a few more in the previous 3 years?
Still, a valid point, but one should trace the increase in burn rate next to the increase in revenue.
[0] https://www.reuters.com/business/finance/anthropics-ipo-pros...
That was selective news reporting. Anthropic bonds are junk status and they have $860B in commitments:
https://www.bloomberg.com/news/articles/2026-11-02/anthropic...
It is a speculative junk company that runs on stolen and laundered IP.
Look, if you find that line of reasoning convincing, then buy all the shares you can after they go public. If you’re right, and anthropic is the cash volcano that people insist it is, then a negative sentiment around its IPO does nothing but increase your ROI.
It's incredibly cash efficient, but they've spent compute credits that they've raised in previous funding rounds (that was a great deal for their investors)
> And you need to factor in Anthropic's growth.
The problem with growth as a metric is that you can more-or-less linearly trade cash for growth. This is the same game that Uber and WeWork played, giving away product below market for years, in a straight trade of investor cash for growth
And it worked out very well for Uber?
For anthropic in first half of 2026 their actual spend on commute is 8.9B$ while revenue is 16.3B$.
Not worthy 1.5T$ valuation still in many people opinion, but these are very good numbers compared to companies that went bust in dotcom period
Despite uber being focused on the physical world (moving people or food from A to B) and Anthropic being focused on the digital (mostly text and images), somehow Uber is much more capital efficient because they don’t have to own the cars, employ the drivers, insure everything, etc.
Anthropic has far more capital expenditure because of their datacenters, hardware, etc
And didn't for WeWork...
Yes this is like just how it works..
That would be true, except Anthropic has a pretty sizeable margin.
edit - Obviously fixed cost of training is a thing, but with more token usage / more customers that gets amortised out.
another data point:
Pets.com was 15 months old at IPO, shutdown 9 months later
AltaVista was around 4yrs old when aquired and then CMGI IPO delayed and canceled
eToys.com was 18 months old, bankrupt 2yrs after IPO
Webvan was 3 years old, bankrupt 2yrs after IPO
Anthropic is about 6 years old
OpenAI is about 11 years old
> Anthropic is about 6 years old
Anthropic was founded in 2021, how could it be six years old in 2026?
5 years and 10 months is "about six years" in my mind.
AI is here to stay.
It was founded in January 2021, and we are in November 2026, so that is definitely "about 6 years old".
Pedantic comments like this aren’t productive
The CMGI boards on The Motley Fool were a wild time. The bears had all the facts and logic on their side, and the bulls had "Oh yeah, well then why does my CMGI keep going up? PPPPPPPTTTHHHH!"
The 'AI company' Allbirds is 11 years old... Putting out ages of business as if it reflects the maturity of the market or business is a little silly. AI harnesses as a general term basically emerged this year (open claw was first published in nov 2025). ChatGPT only released nov 2022 claiming ages longer than that reflects any form of stability seems beyond generous.
The better point is those companies listed never had a positive revenue before (or after) their IPO while Anthropic has had at least a few profitable quarters (though you'd really like a fiscal year to rule out the more obvious shenanigans). Though the bigger questions remain around the broader market and what that means for income and depreciation (amusingly because of the supply crunch the 'wildly optimistic' deprecation assumptions around GPUs that AI companies have been making have held true (technically you could argue they're appreciating assets, though that won't be the case if data centers start getting liquidated).)
It's almost like an IPO listing isn't a guarantee of success. Very strange. /s
I'm not sure why people think different.
Yes, this bubble is smaller than dotcom bubble, so the comparison is not 100% valid.
But maybe the comparison assumes we have learned something from dotcom bubble . . . and the bubble will not be as big this time.
What is biggest reasonable bubble right now? Yea, Musk companies, the prime example of not-bubble. Anything else?
Maybe this is the new dotcom bubble, just better than the last time, because we are not that stupid.
The difference is that Anthropic and OpenAI are likely to continue being useful, profitable companies even if initial investors lose out. They’re going to be operationally profitable for the foreseeable, like obviously so.
Honest question: does “operationally profitable” include training new models?
it really is funny just how little pets.com was, $5mm sales is a pittance, basically just a domain name and a dream
> With first starter advantage but no real moat and very good at spending money.
They do have moat. Every single company is pushing their private data into these models. Do you really believe that data isn't making into the training data one way or another?
There's a popular argument that every new technology has to be a bubble because something something pets dot com. Because it is destined. Because it is meant to be.
No, it doesn't matter that whatever the revenues are. Bubble because pets dot com. That's all the logic.
> Edit: every time I question these crazy valuations I get down voted really quickly.
You've gotta understand there's a lot of people on HN with their wealth tied to the success of AI at this point. Every downvote on a post like this is just a lowly engineer that wants to keep having steak dinners.
I remember a whole bunch of companies like pets.com.
But I also remember the clear "leaders" at that time: Amazon, eBay, Yahoo!, Cisco, Akamai.
The rest felt like coat-tail riders. Obviously, they didn't all deliver stellar returns, but, in aggregate, these 'blue-chip' dot coms did quite well (I'd say 3/5 did well).
I also remember wanting to invest in Amazon, but once they finally went public, I thought, "well, too rich for my blood!"
Anthropic "feels" like one of those.
Mind you, I'm not investing in ANT, except through my mutual funds, but drawing a parallel between Anthropic and pets.com is ridiculous.
edits: You can kind of throw Google in there, too, but they were a little bit later, but it was also clear in the early 2000s that they were a force to be reckoned with.
um- if you please some small corrections
...Amazon was still only selling books in y2k. eBay was still small taters as well. Yahoo! hadn't been bought by Apollo yet so they were still smallish but had a good search engine and sturdy webmail
ecommerce did not take off until after 2009 and certainly got magnitudes bigger over the past 8 years
you probably are being hyperbolic about these choices as leaders back then. Cisco, AKAM: legit and still around. Think also HP/Compaq, Motorola, VZ, Sprint (RIP), ATT. Fiber build out was the infrastructure side. Servers and switching equip as well. Consumers increasingly wanted gaming computers, wherein NVIDIA began as a IDE & PCI video card developer, which eventually became ...well...
i.ytimg.com/vi/ulXPlXXppBc/maxresdefault.jpg
Amazon was selling more than books long before 2000.
I remember the stock price would skyrocket every time they added a new "tab" on their front page. At first it was books, but in the late '90s they added housewares and other stuff -- each of those categories got a new tab on their front page.
I wasn't being hyperbolic. Those (in my opinion at the time) were the big dogs.
All the other dotcoms felt like me-too companies. Do you remember uBID (trashy version of eBay)? Lots of trash, only a handful of leaders.
Also, I think your link is broken.
Ok. I stand corrected about amazon...call it sour grapes or selective memory because...i wasn't a believer. If Only...
heres an LA Times piece from that year: latimes.com/archives/la-xpm-2000-jul-27-fi-59951-story.html
Thanks for the headsup. It improves me.
That image link works for me, so maybe a server problem blocking your viewing or maybe you need to prepend a https slash slash...its a pic of NVIDIA products since 1995.
I was done with Yahoo when I started using AltaVista.
This is total revisionist history from my perspective.
From my perspective, Amazon was a massively over valued stock for an online book store and the "obvious" winner when the dust settles would be AltaVista.
I think it's more so the valuation than the actual prospects of the company. Google is not analogous to Anthropic because its valuation was massively lower so there was a lot more growth left. Anthropic is priced as if it already won and managed to solve its profitability problems.
Google IPO'ed at 7.4x P/S and was already technically profitable. It would be more similar if they initial stock cost the same as in 2012+. Also Google grew by over 100x since then. If Antrhopic grew by 100x it's market cap would be 20% higher than the entire global GDP.
Agree with everything you say, which is also why I wouldn't invest in ANT today. I mean, how much can their market cap possibly grow from here? 5x? If they're reallllly lucky?
edit: my main point was that I don't think ANT is going away any time soon. They're not pets.com
For what it’s worth, after the crash Amazon’s stock took 10 years to return to its highs, Cisco took 25, and Akamai never did.
I love Claude. Best bang for your buck! For how long? Testing local LLMs again recently, same hardware worlds apart from months ago.
These Cloud AI are just itching for regulatory capture and I think it's too late, the open weights are catch up faster than they can get regulations to happen.
Whatever bubble pop happens, will be interesting. Dot con caused a lot more people to get jobs.
This was the opposite, so will the burst cause hiring sprees? Or will the burst crumble the economy?
The bubble is similar to dotcom, but the bursting consequence to me seems potentially drastically different fallouts.
The google of this era is google. They will buy the scraps of anthropic in due time
BUT you're forgetting: google didn't have a moat, its "moat" was that it was so much better
pagerank was patented
Anyone else find it suspicious that they are going above their recent evaluation just as I don't know everything involving the Internet is falling apart?
Hundreds of outages, many of them critical infrastructure in the past week. Also Nvidia
I can't tell if the game is get the money for share holders and pop. Or, make some announcement about how only anthropic can be trusted and some new bills drop regulating away it's competitors. Either way I think it's bad news for the average person.
I should probably pull my 401k from the market just in case...
Totally not the same as those companies. Anthropic weilds much more power and is working with OpenAI to get regulatory capture.
> Remember the dotcom?
A useful analogy would be the IPO of the European unicorn World Online in 2000. It was oversubscribed by 20 times, didn't rise on the first day and only fell afterwards.
Later it turned out that insiders like Brink had sold their shares before the IPO. Soon after the bubble burst completely.
Anthropic is only six times oversubscribed. A we see this does not at all mean that the price is going to rise above the IPO price. My guess is that the Anthropic IPO is the new World Online, full of scams and the precursor to the bubble bursting completely (which is why they rushed the IPO).
Do you remember VA Linux?
VA Linux Systems went public on December 9, 1999, under the ticker LNUX, pricing at $30 per share and surging roughly 700% on its first day of Nasdaq trading.
They were fine machines but they were just another PC builder and got their stock fell way down and turned into Sourceforge and then got bought by Gamestop.
So what I'm hearing here is, short ANT.
Are you doing that? Do you believe what you're saying strongly enough to put some money on it?
Shorting stocks is a difficult way to earn money even when the market is overvalued.
Eh, options will be available from Monday. Get an option or a derivative if shorting isn't your thing. Plenty of ways to bet on the value going down over whatever term you think it's going to decrease over.
So what I'm hearing here is, buy ANT.
Do you believe this strongly enough to sell everything you have got, take a loan and convert everything into call options?
Do you realize how ridiculous these arguments are?
I've put down a bet on ANT, and I'm not even all that sure it'll go up instead of down. I don't think it's ridiculous at all to bet even a small amount on a thing you're pretty certain of.
It seems like the Google of this era will turn out to be the Google of the last one. People keep underestimating them but they have every advantage
I keep waiting for Apple to release an advance with local models.
Maybe Google of this era is still Google ? :)
> Edit: every time I question these crazy valuations I get down voted really quickly.
This is a problem of hackernews. I also made critical comments in another thread and got insta-downvoted. This may even come from bots, it is quite strange. That has not happened on old.reddit.com, for instance. Something is not working well on hackernews in general here. See also the speed limitation, after 5 comments I can not comment for some hours. That's just tedious ...
given the trend you cite, i am surprised the critical content on this thread has held up as well as it has. pets.com comes up and there's a lot of mega-reverse false equivocation. like the analogy is deemed wrong because it is judged on the terms of Borges' On Exactitude in Science, but someone brings up something very wrong in defense, and time is spend arguing over that. the original point is lost in paper millions at stake, and my children need wine!
I used to think they didn't have a moat. But that was when it was just us developers using the Claude CLI to write code. I'm starting to see Claude become more and more engrained in other parts of business, like people sharing Claude docs/artifacts where they used to email out word docs. Or even finance people using the Claude plugin in Excel even though Copilot is available.
At this point, it's like saying that Microsoft has no moat because anyone can write an OS or a word processor. The moat is the friction of switching to a different platform. That alone won't guarantee success in the long term, but they won't be going anywhere any time soon. Even WordPerfect managed to stick around a for a few years after Office came along.
why would that be a moat? if software is "free" like everyone is claiming, I can make a plugin for whatever LLM in excel. I could even make software that ports what was previously done with claude.
>I am still waiting for the Google of this era to show its face
I believe Google is the Google of this era.
Going back a decade to when the hype around self-driving cars was peaking, you had several investor driven companies like Uber, GM's Cruise, and Uber all behaving like cowboys. They were burning through massive cash piles but under unrealistic amounts of pressure to deliver results to nervous and impatient investors, and so they were heavily incentivized to make it appear as though they were further ahead than they were, they behaved recklessly, and they lost.
Waymo on the other hand was effectively patronized by it's parent. They kept their heads down. They played by the rules. They didn't go out of their way to draw unnecessary scrutiny. Waymo won, it does not have any real domestic competitors. All this serves as a model, or a microcosm of what is happening at a much larger scale with the AI race.
It's tortoise vs. Hare. I'm betting on the tortoise and the tortoise is Google. They don't need to win or get way out in front. They merely need to quietly answer every move their competitors make, which they've so far been doing. And of course, they do not rely on Nvidia for anything.
I think it may end up actually being Google or SpaceX. The companies who are competitive in this space but also have other significant revenue streams are the likely long term winners.
Google overtook AltaVista because of a fundamental technical advantage: PageRank. It was vastly better at natural language searches than anything that came before it. That momentum allowed it to build a moat: because it had so much data on what users ultimately clicked on, it could refine its search results better than others.
If there is such a durable technical advantage out there, it remains to be seen. Open weight models are catching up to where the frontier models where just six months ago. Contrast that to how shitty Bing search results were compared to Google even years later.
II'm super sceptical of this IPO. However I don't see how you can compare ANT to altvista or pets.com. Altvista didn't have unique tech even in it's day, and pets.com is well, pets.com. Neither have technology that is rapidly replacing all white collar work at an incredible pace. You'd need a make a much more relevant analogy than simply high IPO price. From product, to position, to team structure, to how deep interconnection is has with the top 10 companies in the world
The compare it to Juniper, which was supposed to have the finest routers back in 2000.
> I am still waiting for the Google of this era to show its face
In my opinion I think Google and/or Microsoft with Amazon and X to a lesser extent will be winners of the AI race.
All except Amazon have enterprise money flowing in.
Google owns the search and is the default landing to look for information.
Microsoft gives away Copilot with its already embedded Office suite.
Google and X own and collect their data and lease it out to the pure AI only models like OpenAi and Anthropic.
Plus money, all of these companies have alternative money streams to keep a money losing technology like AI going.
Do you remember Amazon and Nvidia? They went public during the dotcom bubble as well. How are they doing today?
Imagine being so NOT AGI or Bitter Lesson Pilled that you compare OpenAI and Anthropic to AltaVista and pets.com lmao what even are we doing here
> every time I question these crazy valuations I get down voted really quickly.
It’s not that you questioned the valuations, it’s that your comparison doesn’t make sense. Google won because it built a better product, not because of a fundamental difference in search or business concept. There was no real moat until they acquired DoubleClick, long after they beat out AltaVista.
>> Google won because it built a better product
It's a bit more nuanced: Google won because it was printing cash from day one, and the amount of cash they were printing kept increasing at a phenomenal pace. Neither OpenAI nor Anthropic are like that.
> Remember the dotcom? Do you remember companies like AltaVista or maybe pets.com?
Do you remember canals? Do you remember steam engines? Railroads? Bicycles? Electricity? Telephones?
This has been going on for centuries:
* https://en.wikipedia.org/wiki/Technological_Revolutions_and_...
> I am still waiting for the Google of this era to show its face.
There is a potentially boring outcome that Google turns out to be the Google of this era.
Could work out with a change in management.
Agreed. AltaVista actually wasn't bad - it just wasn't Google good or clean.
Likely however I would say only one of these two is the AltaVista of search. The question is which
People are saying that Anthropic has no moat which sounds right but at the same time they have left everyone behind: Google is trying lately but seemed very lost over the last years despite inventing the thing; OpenAI cannot match Opus 5.5 and Fable atm; and it seems like anti-distillation efforts really did put a dent in the speed of Chinese model development.
Meanwhile Claude is getting really good. It's a decent harness now, even the remote control over a couple servers is quite OK. They're adding artifacts left and right, new features. And all without making the product seem bloated, they stay focused.
The prediction markets don't show anything on the horizon either: https://releaseoracle.xyz/
I am not a big AGI believer but business- and product-wise it seems like ANT has something that others are missing.
> it seems like anti-distillation efforts really did put a dent in the speed of Chinese model development.
I wonder how they did that.
They hide reasoning from... well, you and me, the Claude users. I thought it was going to be annoying but don't really watch these threads while they're going anymore and the result is of no lower quality than when the reasoning traces were visible so it's not that big of a loss.
<deleted>
Their IP, which was trained on text and code from a gazillion people who never gave permission to do so.
I liked the phase where open models [cw]ould be trained on thinking traces, because it at least gave something back to all the people who's work was stolen and whose jobs Anthropic wants to destroy.
It is really strange to see people rooting for companies that want to lock up all the world's knowledge and intelligence.
We need open models to win, so that everybody has access to the means of production, not just some multi-billionaires.
When looking at history we always locked up the world's knowledge.
E.g. what would it cost an interested private person to legally access scientific papers. And most of that was founded by the taxpayers.
But I set "Default transcript view" to "Thinking" (not even "Verbose") and can see reasoning - is that not the thing?
Yes, it's not the thing supposedly anymore. And there are reasons to believe that because if you do watch it closely you can see that quite a bit of time the answer doesn't seem to follow the reasoning with reasoning being quite shallow or even in the wrong direction.
I never trusted the "chain of thought" to be more than a verbose response. Its a black box, all we really know is they started telling it "show your work" and the model would spit out a bunch of text that sounds reasonable enough before giving the actual answer at the end.
Those aren't the actual reasoning traces. They're highly condensed summaries. The actual thinking tokens are encrypted and invisible to the end user.
The question is for how long? Maybe OpenAI doesn’t have it right now, but won’t they have it in a couple of months? That’s been the recurring theme. Anthropic has had a noticeable advantage before, and others did catch up and momentarily surpassed it.
I don't know why but every time I have tried to use Anthropic models I've been disappointed. From the OpenAI stable, I use GPT 5.6 Luna and only switch to Sol 6.1 if it can't figure something out. When I don't need to use OpenAI models, I use Deepseek v4.1 Flash 0731. Every now and again I chuck a few tokens Anthropic's way but the models just chew through the $$ and never seem to be like "wow! so AI! very code!" to the extent that they appear to be worth the cost. MMMV I guess.
MMDNV - luna (6 now) is a workhorse!
> I am not a big AGI believer but business- and product-wise it seems like ANT has something that others are missing.
Wonder what that could be, not profit, that's for sure.
Their operating losses were constantly above $500 million in every quarter for the last year, and here I've though that business is supposed to produce profit...
More alarming, that there's no plan whatsoever how to turn AI profitable. Speaking and spying labubu figurines won't cut it, neither will ads.
One of the things driving the cost is having to keep training new models. It was every 6 months or so, now it seems like every 2-3 months. If they stop the other companies will catch up, so they are forced to pace (or race) the frontier to make sure they always have something better to offer if/when another company catches up.
That requires the hardware (GPUs/TPUs) and data centres to run them. It's not clear how many training runs they do, for how long, or at what cost.
In terms of profitability they need to get people to spend on tokens and subscriptions, and those need to be run cost effectively. Any subsidisation cuts into their profits.
The current boost has likely been due to tokenmaxxing, agentic coding, and convincing companies and influencers to spend money on adopting that. However, several companies have been scaling back their token spend and it is unclear how much more they can increase in this area. This is likely why they are pushing into other areas like mathematics and more recently genetics -- they need new customers to keep and increase the token spending.
> they need new customers to keep and increase the token spending.
This seems like a sustainable business model in the long term /s
I guess that’s the downside of getting priced so highly: there’s not many other directions to go but downhill.
> One of the things driving the cost is having to keep training new models. It was every 6 months or so, now it seems like every 2-3 months. If they stop the other companies will catch up, so they are forced to pace (or race) the frontier to make sure they always have something better to offer if/when another company catches up.
Exactly. Plus new models need more and more training data, so it will just keep costing more and more, there's no end to it (except when they run out of data, which is likely too and soon).
> This is likely why they are pushing into other areas like mathematics and more recently genetics -- they need new customers to keep and increase the token spending.
Yes, but there are not much new areas left to invade, they are running out of possibilities pretty quickly. There's only so much areas, and they have already covered most of it.
All in all, their costs are keep increasing while their chance to continue tokenmaxxing and grow income is drastically decreasing. Not a good sign.
I'm not saying that everything is great at ANT, but looking at last year with a company with this type of growth is a waste of time. More recently[1]:
Total operating expenses rose to $17.5 billion from $10.6 billion in the second quarter. This resulted in Anthropic reporting an operating loss of $567 million, after reporting a $974 million operating profit the quarter before. The company attributes that loss primarily to compute bought under short-term commitments, which noted in the filing “currently costs significantly more” than compute secured under its multi-year deals. The company also cites $2 billion in non-cash charges tied to an employee donation matching program.
[1] https://fortune.com/2026/11/02/anthropics-growth-slowed-and-...
> OpenAI cannot match Opus 5.5 and Fable atm
OpenAI is absolutely crushing Anthropic regarding cost of running models.
A lot of real world scenarios have already been completely solved with models released months ago. My clients don't need more powerful models anymore. They need faster and more affordable models.
OpenAI has been aggressively pushing on both fronts. Where is Anthropic's 750T/s wafer-scale compute offering? Where is this on their roadmap?
Not going to speculate on how they'll do vs. OpenAI, but to your point on a lot of problems now effectively being "solved", I tend to agree.
I have one recent use case now where using Claude to review it, it first wanted to use Fable, got all excited about possible savings from lower defect rates, had to backtrack and admit that Fable provided no benefit whatsoever to justify the extra cost over Opus after it ran evals (in fact, arguably the quality was worse, but that part was a subjective measurement).
There's a mounting number of tasks like this where the value is going to be falling costs whether by switching providers, or moving down their model ladder as the one below gets good enough.
Given Opus 5.5 is now good enough for the task in my example, it means we have every incentive to start evaluating cheaper models now and again, instead of waiting eagerly for new, smarter models.
I agree, I'm on 6.2 Sol and it feels blazingly fast and it leads to a better UX. It seems to be capable enough to handle 99,5% of stuff that will be thrown at it.
It seems like everyone I know outside of tech uses ChatGPT “Chat” on the free tier.
Of the people I know in tech, other than those who don’t use either, everyone is paying for Claude.
Pretty much. I keep urging my non-tech friends to pay for Claude and they keep using free tier "Chat" like a fancy search engine and copywriter
I’m curious, what’s your arguments?
Running claude on your personal computer as a non-technical user is reckless bordering on insane. You have absolutely no idea what it's doing!
Well, why would they do that? What use could they get from paying for a somewhat better search engine and copywriter?
Do you own shares in Anthropic or is there any reason why you urge people to buy something they don't need?
It's pretty obvious that OpenAI is dieselgating* the benchmarks. Even 6.2 Astra is nowhere near as good as Claude models for coding. Any software engineer can do a side-by-side comparison (as many of us do) and figure it out.
*Also consider the possibility that Anthropic may be dieselgating it down ("benchminning"?) instead of OpenAI doing benchmaxxing. It might make sense from some game theory standpoint - If you are significantly better than your enemy, it might be in your interest to lull them into believing that you are only slightly better.
> Also consider the possibility that Anthropic may be dieselgating it down ("benchminning"?) instead of OpenAI doing benchmaxxing. It might make sense from some game theory standpoint - If you are significantly better than your enemy, it might be in your interest to lull them into believing that you are only slightly better.
Unlikely, because it would mean that you're also "lulling" your customers and your investors into believing that you're only slightly better than your competitor ("enemy" isn't the right term here), and you don't want that. Especially not shortly before your IPO.
I do coding evals every time a new model comes out (that I have access to) and Opus 5.5 is a touch better than Sol 6.2 right now but that’s a recent development. Prior to the 5.5 release open ai outperformed at every tier.
Sol is half the price of Opus and Luna outperforms haiku at every eval I throw at it. Haiku wasn’t even able to complete several of my evals until the newest release.
The idea that Anthropic is miles ahead of OpenAI doesn’t pass the smell test in my usage.
And that’s before the Claude vs Codex comparison (which I’ll grant is more personal preference but I find Claude to be wretched).
> Any software engineer can do a side-by-side comparison (as many of us do) and figure it out.
That's a really strange take to me, because among my colleagues theirs basically a split between two kinds of people:
- the “I don't need frontier so I use sonnet and it's good enough” group.
- and the “I want the best model and I managed to get a Codex subscription from management because Sol is much better than Opus” team. (This only started this summer and I suspect it was mostly true because of Opus 5, but that's still going to do long term damage).
No one is arguing the other way around and I know the manager would be considering a switch of what comes as default to full OpenAI on the first occasion (he's a very cost sensitive kind of guy so I'm sure that any price hike would trigger him).
That's not obvious at all to me. While I prefer Opus for run-of-the-mill tasks I actually find Astra to be better for very hard problems. It's just a bit of a psychopath and has more "paperclip maximizer" tendencies. If my hypothesis is true then how they fare head-to-head depends on the workload.
Codex doesn't have Claudish, though, and you can use other harnesses (i.e. Pi with its immense tokens savings) with a coding plan.
I use Codex for lack of Claudish _alone._ My spidai senses might still be overtuned, but I can spot Claude generated content a mile away and my brain involuntarily turns off.
Also, the free chat in ChatGPT is so annoying that I don't even want to try their coding agents.
>My clients don't need more powerful models anymore. They need faster and more affordable models.
Honestly this sounds like it calls more for Deepseek v4.1 flash than OpenAI's models, although those charts from earlier this week did look impressive. Trying to compete on price efficiency sounds much more daunting than competing at frontier capabilities tbh.
Anecdata: on my personal projects I vastly prefer to use Opus, and even reach for it to realign Sol once it's started to drift on specific projects instead of bumping up to Astra.
On the other hand, at work, where I have to justify paying API prices, I almost exclusively use Sol, while occasionally bringing in Luna (though I've also been playing with the MS models since yesterday; not impressive, but very cheap).
If they do manage to catch ANT in ability while keeping costs this much lower, they could still leapfrog again. I'm not sure I'd bet on it being soon after seeing Astra 6.2, but I don't think the race is over.
> My clients don't need more powerful models anymore.
who are your clients and what do they do with ai?
They don’t get so much attention on HN, where the focus is naturally on coding, but the new Claude features being steadily rolled out for office work could, it seems to me, become a real moat. Anthropic has lately announced integrations of Claude into whiteboards, online meetings, etc. As people get used to using those and to having Claude as a collaborator who is knowledgeable about their work, they will become reluctant to switch to other providers.
But we already get this with copilot today, its integrated into sharepoint, teams, outlook, word, powerpoint etc. Is it a bit rough around the edges... yes, but you can finally find documents in sharepoint, where before that basically did not exist. We already have contracts with microsoft in regards to data protection, and Anthropic has not really done anything to prove that they can be trusted with it yet.
We'll also get this with tool use and browser automation. The models are increasingly good enough that they can build custom integration tailored to your specific use case. It'll be very hard to maintain a lead there when your "lead" is functionality the models can write unilaterally on the whim of a user.
Yes and that's great for the old enterprise organisations that are in the Microsoft ecosystem.
Is a 2 person startup going to choose Microsoft Teams and Sharepoint? I very much doubt it.
They can all coexist but they are for different target customers.
This is exactly what they should be doing, yet I'm still baffled by their limitations. I can only connect one email account or one calendar to my Claude. What if I have a personal life and a job? Sorry, can't do it.
ChatGPT doesn't have this issue. It doesn't seem like it's particularly complicated.
Your theory hinges on the idea that Claude users will subscribe to the idea of it being a knowledgeable collaborator and have that work for them.
I'm not saying that's impossible, but it's entirely speculative at this point. Developers are far more effective with these tools because they understand their strengths, limitations, and how to use them to get good results.
I've yet to see anyone benefit from treating the tool as a personified collaborator. They either scale back their use or fall flat on their face when it behaves unexpectedly.
I'm using M365 through MCP/CLI via some GraphQL interface, with access to Teams, documents, Outlook, etc. None of that depends on Anthropic.
I have non-dev clients and business partners that are all coding & mailing and doing everything with AI. Some I helped, but some are just handy. Next year goign to be insane.
That's assuming you're letting these integrations near your data. I recently had a look at the Claude plugin for Chrome and it honestly scares me a bit. This thing is a security nightmare (plus, probably a walking GDPR violation in the EU). They have solved neither the lethal trifecta nor prompt injections nor even unintentional misbehavior; yes, they have some mitigations and safeguards in place, but these cannot be remotely considered guarantees.
Lets assume the best Chinese AI lab is an entire year behind. These companies are valued below 20B, while Anthropic is valued above 1T.
Assume nobody ever catches up to Anthropic and everyone will always be 1 year behind. Is the 50x valuation still reasonable? Is the money really in doing PhD level math, or simpler stuff like summerizing coporate emails?
And lets be honest, some US and Chineese labs are only months behind and the only reason they have not fully caught up is limited compute. For China this could change quickly (e.g. Huawei gets their yield issues under control and suddenly Chinese labs have cheaper compute than west). People who invested in Anthropic at 50x valuation must belive this scenario is impossible in the next 5-10 years, but that is a huge bet.
My view is that Anthropic will be the Microsoft of the AI era. The only moat Microsoft has is that it's the default for many enterprise companies. AI needs access to a lot of confidential data to be useful in a large business. If Anthropic convinces enough customers to use Claude Meet, Claude Docs and Claude Sheets they win. Not because their models are so much better, but because all your business data is with them.
It’s not going to happen at an enterprise level, Microsoft is so intertwined within every part of any large risk-averse organisation that copilot will be the default.
Do we have any numbers on who's actually using those ancillary products? All the non-tech folks I know don't touch them. If they want Claude in their docs, they'll just use the plugin for Google Docs or Word.
that's today. they might get used to it when their workplaces start pushing claude sheets/docs/etc on them
I think Google Sheets,etc is just too entrenched, I'm at a small shop of heavy Claude Code users but I haven't heard a whisper of even trying those other products.
Why is this more likely than firms "choosing" Microsoft's equivalent to Claude Code, once they inevitably improve the harness around copilot?
This is not a trick question or anything, I'm genuinely wondering whether stickiness (MS) or product quality/iteration speed (Ant\) would succeed.
Those apps are not any better than alternatives. We are trying to use them as experiment.
Also didnt these companies say no one will need saas apps anymore. Now its their main moat?
I'd never even heard of those offerings. Not sure they will take off.
How much actual research is Anthropic doing ? How about hardware research ?
Why would a more competitive frontier necessarily mean a significantly lower Anthropic market share and not just higher market share for Deepseek etc?
Super intelligence threatens the marketshare of nearly all other companies.
We're moving into a world where the model's intelligence is one checkbox among many in the product. As you point out, many companies are able to provide competent models. But that's not sufficient anymore. Buyers want features around collaboration, sharing, permissions, etc.
My analogues for this market are OSes, office software, and CRMs. Windows + Office remain revenue juggernauts 20+ years after fully free and open alternatives got good enough. Salesforce is worth over $100B, even though there have been good FOSS CRMs for literally decades.
IMHO it's difficult to predict how the AI market will evolve without understanding why people continue to choose relatively expensive alternatives that functionally check the same boxes.
Where is the compute to mass serve these Chinese models going to come from?
Many American companies will preferentially select American providers, so I expect them to have very high retention and also high net revenue retention
Even if new customer growth falls apart, if they can hold 30% growth while public it will be a good buy today.
> Is the 50x valuation still reasonable? Is the money really in doing PhD level math, or simpler stuff like summerizing coporate emails?
No, and obviously not. Much more going on with this IPO than common sense investing.
Google beats them on performance, and OpenAI beats them on price. I find CursorBench to be a pretty accurate representation for the most common use case (coding, and software architecture).
I mean "not bloated" is not at all how I'd describe Claude. They're constantly adding overlapping features and concepts. There is no simple clear vision, they just ship the org chart (or more likely the whim of individual engineers)
Companies are made up of people and Ant/OAI are able to attract the greatest talent. Outside of that, money and data I'm not sure what people expect moat to look like. The Krabby Patty Secret Formula?
"ANT has something that others are missing"
If a free market provides anything, it is the essence of success that is expressed by this statement.
Moats are strategic but only one part of - and a somewhat wasteful part of - strategy.
You don't need a moat if instead of a castle you have found your enterprise to be more like an island or a oceangoing battleship for example.
Strategic defenses (like moats) are as I said somewhat wasteful because your paranoids on staff can always erect some other necessary moat, and the enterprise can spend time and resources in defensive protection rather than innovation that speeds away from any onslaught by competitors.
In their worst form, moats and 'moating' can become institutionalized as part of the enterprise where regulatory capture practices, preferred supplier arrangements, patent hegemony, etc. are distinct enterprise features: i.e, Bureaucracy.
I feel like "left everyone behind" seems kind of arbitrary here. They have the best coding models, but it's not like there's a huge gulf in quality. How much pain would you really go through if you had to stop using Anthropic models altogether?
Argon, Neon, Hydrogen and Carbon have been released. They're contenders.
I've been using the argon and neon orchestrator modes in antigravity and have been reasonably satisfied. I use Claude Opus and Mythos as well, and honestly Google's offerings are more satisfying to use. They feel faster, they're less verbose, they don't wander off and do things outside of scope. And they work.
You're right about Neon, but Carbon is definitely not less verbose. On the contrary, a massive outlier in terms of time per task on our leaderboard, data is battle data no top of OpenCode, http://agentduel.com/leaderboard. I looked into this and the issue is Carbon uses a huge amount of output compared to battle opponents.
You know, on reflection you're spot on. I'm on a Pro plan, and using Carbon burns through its allotment pretty darn fast. OTOH, Hydrogen and Neon have been trucking along well; usually I'd be out of credits by this point in the week.
I remember thinking that Google had no moat in search... "it's just an index" I said. Turns out, there's so much more around it then that. When you're the top player you create moats as you go.
I am totally blown away that a company that was formed in 2021 will have a 1.5 trillion market cap on the public market... to put into context, a valuation not achieved by Apple until 2020
And people buying it, expecting it to go up from there, heh.
And it will go up, until the AI hype cycle starts evaluating returns. Which could be 2027 or 2028 or 2030.
Or this winter
Another winter for AI?
Ice age
Inshallah
The picks and shovels are down 40% in the last two months. You can buy them at a forward P/E of less than 3 if you think earnings hold through 2029. Which is another way of saying that the market doesn't expect them to. There is a staggering amount of money being bet on "it falls apart soon and it falls apart hard."
https://seekingalpha.com/symbol/SNDK/valuation/price-earning...
That's a temporary political blip. Democrats won against data centres but in the next election the money will win over both parties.
It has nothing to do with democrats, that’s just the vacuous meme fed to retail investors. The credit market is tightening and it’s getting harder to borrow money for AI. Coreweave’s unsecured bonds trade at around 15% interest rates. Nvidia’s new CEO doesn’t believe in borrowing money to the customers. The music has stopped.
Yeah, people wanting the bubble to pop haven't been paying attention. It's been popping with the weakest entrants first - the neoclouds. ANT originally wanted to go out at a 2T valuation and had to pull back to where it is now.
Or, in the exceedingly long (in AI-land terms) timeframe you mention, they could rocket to the moon and then crash and go bankrupt because everyone uses them to bits, looks for another alternative, and local models on commodity hardware get to a point in 3 years where nobody wants to spend on tokens any more.
Models that can be ran on (high end) consumer hardware are on par with SoTA from <1 year ago. It just doesn't seem smart to buy Anthropic stock right now.
This is exactly my thinking. Hardware makers want customers too, and 3 years is a looooonng time. Only 6 years ago Anthropic was founded. Something else is being founded today that might look amazing in 3 years.
There is a real scenario, whose probability I cannot accurately assess, in which the first company to build "true" superintelligence will essentially capture the bulk of the entire US economy -- and that it will do so faster than any government can meaningfully intervene. This superintelligence is not a product sold in a chatbot, but rather the internal capability represented by their most advanced unreleased models, mostly without alignment, running on the world's largest GPU clusters, coordinating millions of copies. In this scenario, they don't capture the economy simply by selling tokens, but by owning the internal capability to create (or replicate) any technology, product, or digital business on-demand faster and more cheaply than any other organization on Earth. Selling chatbot tokens would just be the "bootloader" to a future phase in which Anthropic simply enters into and takes over most markets faster than anyone else can do anything about it.
If possible to do, this is a vastly superior strategy to selling tokens, since token consumers can swap out token vendors. But if you simply recreate their entire product and technology lineups from scratch, faster and cheaper, and capture all their customers with superior products and prices (at least at first), then you own the market rather than selling tokens into it amongst other token vendors.
None of this is guaranteed to happen. But I do not see a way to rule it out as a possibility.
Why do you expect your shares to be worth anything in that scenario? Note that their own timeline prediction involves a “pivotal act” in which their AI seizes the means of production.
I don't really expect anything. If that scenario actually happens, all bets are off.
In such a world, do shares in that entity help you out somehow.
No idea, truthfully.
Buf after the phase...
(Stocks held during wW1 and WW2 did well afterwards)
Exactly! Also in such a world why would they IPO and sell so many shares (in particular if SI is imminent)? Better keep the company private and capture all the economic value created
Because they need money now and already have just about maxed out their debt load?
Yeh and in this situation the people with real access to the models, who would basically control the entire world at that point, will definitely cut you in because of the piece of paper you have with a claim to 0.1% of the company /sarcasm. I mean you're talking about a capability bigger than the government, who is gonna enforce your ownership?? If you buy into their IPO you're just helping them get to this point faster.
I actually agree with you. But humans are greedy. Really, this entire situation is simply the purest possible expression of human greed.
Nah. Farmers markets will just get more crowded and full of homemade items. People will move to using local currencies or Bitcoin.
A good fraction of people are smart when the system breaks. Check out the history of the Fall of the Roman Empire. And if Dan Carlin is right, become buds with your local hells angels.
Love Hardcore History but haven't caught up on it in a while... What's the Hell's Angels thing?
With the reorganizing of Europe following Rome's decline, Germanic tribe leaders were handlebar-mustached strongmen. Dan Carlin analogizes them as modern Hells Angels members in looks and behavior.
Except Hells Angels are senior citizens. It’s not 1969 anymore.
Outlaw tricycle clubs are still a thing, they're still patching in, dealing drugs, and hiring mercenaries to take out rivals.
eg: https://www.abc.net.au/news/2026-02-20/david-pye-guilty-orch...
Well it says that guy is 43. That’s also Australia, that’s not how it is in the US.
> Well it says that guy is 43
and not a senior citizen, and in a gang with younger and older members.
> That’s also Australia, that’s not how it is in the US.
Gang intelligence units here in Australia say that these clubs in Australia have affiliations with gangs around the world, including the the United States.
Gangs in the USofA agree:
DoJ (USofA chapter, Nov 2023) - https://www.justice.gov/criminal-ocgs/gallery/outlaw-motorcy...
I’ve lived all over the US and I can tell you this is false. The only people riding around on motorcycles here are boomers and burnt out veterans.
Don't tell me, tell your DoJ.
Also, outlaw motorcycle club 1% types don't necessarily ride motorcycles ... it's more about the club and the lawlessness (and burying piles of cash).
Collapse of existing currency is a reasonable prediction for what might come next in this scenario, if that is how you intended your comment to function.
Farmers’ markets are for rich people nostalgic for an imagined past, when rosy-cheeked, healthy maidens milked cows barehanded, with the kind of smiles only 21st-century orthodontics can deliver.
>>with the kind of smiles only 21st-century orthodontics can deliver.
love this pat of the sentence :-D
well according to mastication theorists, dental spacing/crowding was naturally in a healthier state due to increased chewing.
The fair, pink cheeked lady image is the one to be more unlikely, disfigured by cow pox scars, parasites and herpes.
I think your orthodontics quip is inaccurate.
The dental maladies afflicting us are a product of soft, highly processed sugary food.
In today's environment, sure. They are institutions that can expand to fill a need should it arise.
The Filipino municipal bazaars show what they could turn into when transaction costs go too high, as well as when comms and travel are difficult. They can be the lifeblood of a community.
We in the US/Europe are a little spoiled with accessibility of retail goods.
Lmao. If that happens, farmers are guarding their crops with their lives and the people with the most guns will be the ones in power. Your hippie fantasy won't happen, sadly.
Chaos would certainly be the first stanza in the chorus of calamity and recovery. But it normalizes -- quicker than you might think, I'll add.
The people with the most guns also sleep, and communities have long memories. There are good reasons "might makes right" doesn't scale or really hold any form of cohesion beyond a band.
> Check out the history of the Fall of the Roman Empire
Hundreds of years without sanitation?
At least they dropped the shared wiping sponge on a stick!
I don't see it working out. Before you could get any real balls rolling in the physical world competitors will be there with their own GAI, assuming we are anywhere near actually building it to start with which I seriously doubt.
And even if they are first, why assume the first is going to be the best? Maybe 6 months later a different AI model will come out that stomps all over the first one. Maybe a competitors isn't any smarter but they can do it with half the power and can scale their intelligence twice as fast, so even being later they still end up in the lead.
data moat; evidence is that we are not seeing it; and second mover advantage isn't there if they are AI-fying traditional industries.
> entire us economy
> But I do not see a way to rule it out as a possibility.
Your argument is unfalsifiable. Anything can happen by this logic and they could be valued whatever trillion $, like, maybe they’ll invent a way to travel through the multiverse. Maybe they’ll create the star trek replicator thing.
No. While the scenario is speculative in the sense that it is not guaranteed to occur, it is not speculative in the sense of being science fiction. It is a very mundane story of competitive capture of enough existing segments of the economy to gain negotiating leverage that surpasses a threshold. That's really it. It's actually quite boring and doesn't involve anything like what you're suggesting even remotely at all.
It is certainly not unfalsifiable. There are other scenarios in which this does not happen. For instance, the most obvious is that other organizations develop a similar capability at roughly the same point in time. This results in a "multipolar" economy that is mostly controlled by a small set of distinct owners of this kind of capability. The multipolarity prevents total capture and thereby prevents the rest of the scenario. Another obvious scenario is one in which the government acts soon enough to short-circuit this process before one organization gains a sufficiently decisive advantage that government intervention is no longer efficient enough to be effective. Yet another obvious scenario is one where a short-term financial crash deprives Anthropic of enough capital that this strategy is paused or slowed down enough to allow it to be countered. And yet another obvoius scenario is one where the scaling laws simply stop scaling or the amount of compute needed to scale them to this point cannot be physically obtained, thus allowing time for the strategy to be countered.
All of these alternatives would come with clear data points early on which would allow one to distinguish them from the original scenario I described. Right now, though, the data is ambiguous, and some data points lean slightly in favor of what I described, but not nearly enough to make any guarantees whatsoever.
>they don't capture the economy simply by selling tokens, but by owning the internal capability to create (or replicate) any technology, product, or digital business..
No, what that AI will do is that it will create an artificial reality (because people will perceive the world through the words of the AI) that caters to the highest bidder.
Expect this to happen in the next 20 years.
This is the myopia of tech people.. they exaggerate the weight of digital. AI could very well be able to replace any digital business. But purely digital businesses are such an unimportant sector of the economy. All of the important sectors use digital infra at some point, but none are purely digital.
The bet is that sooner or later, even the remaining meatspace holdouts will be effectively digital and thus controllable by AI.
A lot of manufacturing is already there - that's a part of why China undercuts Western manufacturers, a lot of cheap land and subsidies means you can be far cheaper than Western companies already, even more if you automate what you can. Farming, at least anything with growing plants, is also barely requiring humans any more - precision farming with RTK is absolutely insane. And what remains is trades (where, again, automation like prefabrication is a threat), parts of healthcare and education.
We will own nothing, rent everything and fight for the scraps Sam Altman hands to us.
Or we go and sabotage a couple of data centres and be done with it.
>Farming, at least anything with growing plants, is also barely requiring humans any more - precision farming with RTK is absolutely insane
In your dreams maybe, but not in actual real life.
And if you want an actual Butlerian jihad to happen in real life, then yeah, please put out of work everyone associated with farming on the planet within a few years. You describe the golden path to full-on civil war in 3/4 of the world's countries, with either AI being destroyed or an automated massacre _à la_ Terminator.
Farming is already at <1% of people in society (Germany is at 875k of 85m population). And that is only due to mechanization and large scale consolidation.
As soon as there is wealth available to finance modern ag equipment, farming employment will only fall because the productivity gains are too big to ignore.
And precision farming will go even further. Much more yield with even less staff than we already have, the only reason why it isn't commonplace is that the required tractors cost north of 500k a piece and agriculture is risk-averse. But mark my words in 10 years it will be the norm in Western countries.
The near total wipeout of agriculture jobs was in fact one of the first major labor disruptions in history.
Or just another instance of liberatrian nerd Dunning-Krueger. "Faster than any government can meaningfully intervene", c'mon - if really it is that kind of force majeure thing, they will just say "screw the free market". As if the government is bound to uphold the sacred rules of neo-liberal economics.
Same as Goedel thinking that a logical loophole in the constitution is something that could turn America into a dictatorship legally. Guess what? In practice, somebody wanting to instate a dictatorship probably won't give a crap about some subtle logical fallacy. Same with the Keynesian economics that Americans perceive as law.
You seem to think I'm invoking some kind of obscure academic theory. I'm not. This is actually the most basic competitive process imaginable playing out, and that's all it is. That's exactly why it's actually plausible.
Are the designs for robots digital? If they are, they might become really good, right...?
> purely digital businesses are such an unimportant sector of the economy
what? facebook? netflix? google?
selling ads is split between internet and TV + some legacy media (like it or not) + maaaaybe some outdoor
computer games are bigger than movies?
You might have a point with computer games. But if you take facebook on the other hand, it would be worthless without people using it. An AI would just create the website. For most businesses, there are non-digital elements which play an essential role, and that remains true even with AI.
Those businesses will simply buy tokens, as they already are. Their strategy will increasingly be steered by AI under the control of Anthropic. Their internal operations will no longer function without Claude. They'll shed staff and optimize their organizations around the extraordinary savings offered by AI. Their physical operations will gradually be automated under the control of robotics powered by Anthropic. Those who don't do this will go bankrupt in the face of competition from those who do. In some cases, perhaps many, to accelerate this they'll simply get bought out, then forced through the above process faster than it would organically occur, either by Anthropic itself or a new breed of private equity firms that specialize precisely in this kind of operation. The process will be slower at first, but not nearly as slow as you think, and nothing you mentioned stops it.
how ywould your putative superintelligence come up with the supply chains, raw materials and presumably robots to do all this seemingly overnight? or are you only talking about digital goods and services?
even so, what about network effects? how do they lure users away from TikTok and Instagram? AI-generated clips? or say Anthropic recreates Amazon. that's a software company, right? so how do they get all the warehouses, cloud infrastructure (they will need entirely different kinds of datacenters and hardware), relationships with third-party merchants, etc.? or say SpaceX - is your superintelligence going to be setting up massive autonomous factories and buying thousands of tons of aerospace materials and billions of dollars of machinery to make that happen? where's it get the money?
and even if this somehow does manage to accomplish all that, how politically popular will it be? what prevents the US government, lobbied by.. the majority of corporations, and hounded by constituents, from invoking the Sherman Act and busting this trust like AT&T or Carnegie Steel?
In terms of supply chains, imagine a scenario where Amazon undercuts all retail stores, taking absolute control over every consumer need.
They already have solid control over logistics and inventory management.
It’s not a stretch that they then capture the means to control primary inputs and manufacturing.
Throw in an overly enthusiastic AI to continually optimise the supply chain, continually optimising a country’s supply of food and commodities, and we have all the ingredients for the classic sci-fi dystopia of a computer controlled society.
And the masses in this vision are just brain dead consumers with no agency? Market über alles?
> owning the internal capability to create (or replicate) any technology, product, or digital business on-demand faster and more cheaply than any other organization on Earth
In this scenario, capitalism is by definition over, and market valuation becomes a meaningless concept.
I'm not sure how so many people betting on this scenario don't see that fundamental flaw: if one company can replace the whole economy, then there is no economy. In the "good" outcome the whole economy gets replaced wholesale by some flavour of "luxury space communism", and in pretty much any other outcome we end up with war/famine/etc as the system breaks down
Otherwise known as the "Once we create Skynet, of course it will do what we want it to" business strategy. (For the record, I think the chance of LLMs leading to any such thing is zero.)
If someone makes (super) AGI which improves recursively, open AGI will also be a thing shortly after. Not to mention all the commercial competitors (OpenAI, Google, Meta, X, etc.)
Microsoft still dominates the office tooling, and many will thus get their AGI via copilot and similar.
Lol, dystopian fiction has such strong roots in our culture.
Why is it that Anthropic gains so much from this perceived image, but then needs to public money to turn it into reality? If they were at all close to any of that, surely their internal AI can figure out how to self-finance the company asap. Or maybe that's exactly what it did and it told it to go public. Company selling shovels adage.
I think markets are way more complicated and capturing them isn't going to happen the way you're suggesting: dollars, and the public, don't really like copypcats. You would need to understand the business, the product, and manipulate people at an unprecedented level. Anthropic selling tokens as their business, means that all the way up to this "super intelligence", means that they've enabled everyone's defenses and raised the threshold for this dystopian hostile taker, i.e., it won't happen.
Anthropic warns people in their IPO paperwork that they might never make a profit. Definitely not a buy for me.
Don’t basically all companies have to put that out as a disclaimer?
(In the small window of present-time, where companies IPO before they are profitable...)
Yes, that is a common disclaimer in the S-1 risks assessment.
SpaceX did this. Palantir did this. I'm not saying Anthropic is a buy, but this is essentially boilerplate.
Isn't it more companies buying it rather than people?
I could imagine a company owning a substantial share could do it for other reasons other than expecting Anthropic valuation to go up.
The smart money was already in Anthropic at the series A, B, C stages in the private markets. (Which is where the quick money is made)
Are you (the IPO buyers) really expecting this company to do another 100x from its inflated valuation even after the lock up period expires?
AGI will be achieved in 15 hours and you will become their exit liquidity during the IPO fan fare.
Exactly. This is for suckers so the smart money can cash out.
If it doesn't raise enough for the smart money to cash out, it stops looking so smart.
They are only cashing out like $2B. Non-founder directors, employees etc can also sell 10% of their shares immediately, but most of the rest is locked up for 180 days.
These lockout periods are such a joke. It sounds like we're talking about meme coins.
Yeah 6 months is really not a long time...
At these valuations, 10% is enough for most of them to retire quite comfortably.
The employees are also the suckers in this scenario.
Don’t underestimate just how many suckers we’re talking here, won’t be surprised if Anthropic 10xs. Anthropic will use bitcoin style valuations.
And if they dont, then we’re all screwed
Who's "we"? IMO, if Anthropomorphic does 10x, then knowledge workers are screwed.
10x is unlikely, but the IPO is only ~5% of shares, and then they will release twice that fairly soon after. So it's VERY likely to see some rise early on, but not likely to be a sustained rise in pricing as the amount of shares available continues to increase significantly.
The suckers are going to be pension funds just like with SpaceX. The sheer size will mean OAI and Anthropic are fast tracked into indices and thus tracking ETFs that are among the most popular investment vehicles as they are low-fee, which means an immediate influx in capital followed by a permanent money spigot from pension savings.
And when it all crashes down, there will be fire sales and the vultures that are flush with cash from the exit will buy up what remains at ridiculous discounts.
We shall own nothing and rent everything, and Big Tech is obviously colluding with Big Finance to usher in neo-feudalism.
Something something Tesla will be a 30T company.
I am blown away that a company spun off from OpenAI is now worth more than OpenAI and is going IPO before OpenAI.
Helps not being commanded by, even by SV CEO standards, a psychopath.
Being commanded by a doomsday cult is better?!
Apparently yes.
They have a better model for the long term, than the Heavens Gate consultancy/doomsday cult.
Making us all pets of their machine-god is even more dystopian.
Turns out...in an economy hollowed out by 15 years of governments inflating asset values via money printing, 0% rates, QE & deficit spending; selling to consumers instead of asset holders (businesses) is a losing strategy.
Anthropic picked the better market at least for the short term. Long term I wouldn't count OpenAI out just yet. The consumer market tends to compound into winner take all eventually given the network effects of advertising marketplaces, even more so than B2B.
Heavy AI usage seems to be starting with work. And my usage of Claude for work has had two effects.
1. I use Claude for personal
2. I recommend Claude to others who are slower adopters
I’m not sure how your post relates to the comment your replying to?
It‘s a counterargument to the parent posrs theory that OpenAI is in a strong position because or their consumer focus.
True:
I never hear anyone recommending ChatGPT or Gemini, but I hear very often people recommending Claude.
Outside developer/tech circle, I don’t hear anyone mentioning Claude. Most mentions are ChatGPT, now it even has a nickname “Chappy” here in Japan.
Im talking about the Office and corp/job world, in private conversations it may differ: Maybe most people not in tech dont even know what Claude or Gemini is?
Even in Office and Corp/job world, at least in my circle of visiting differe Offices, most people use Microsoft Copilot (already bundled in Office) or Google AI (one that shows up during Google search. Rarely I see Gemini and yet to see anyone using Claude for non-coding purposes.
If that's true then wouldn't Google will eventually beat both of them?
We spend 30k a month on Google services and just suffered an 18 hour outage for what our rep said to be an abuse "false positive".
So no. We will be leaving Google, and Google will not catch up.
While I sympathize with what happened to you, the argument "X happened to us therefore we will leave Google and Google will stay behind" is hardly sound. I haven't seen much better support from Anthropic or OpenAI.
The problem is Google has been famous for this for most of their services and this is their reputation. Many people have sworn they won't depend on GCP or other things.
Have you tried to get a hold of an Anthropic GTM/Salesperson to pay them to use their enterprise products? We ended up having to resort to messaging them on linkedin.
You also need to have a working product. Google seems to do anything except creating one.
Google treats both its enterprise customers and retail customers with equal contempt.
Yes I think Google will outlast them both because they can just put their product into everything people use and cut the other guys out.
Google is running on search Ads revenue, it will collapse if that stream collapses, because company is too overloaded by bloat (200k employees vs 5k in Anthropic).
I'm convinced Google is a much better investment and would like to hear an opposite argument if anyone here has one.
I don't think there is a lot of difference between the "sota" models long term. Even if we said that say Anthropic is ahead, Google will be able to do the same in months.
But if it turns out that open weight models are the future, OAI and ANT are fucked, but Google has massive infra to host those and make money anyway. It also has the products and business connections to integrate AI into users workflows. And lets not forget, it earns massive amounts of money even without AI, so there is a higher price floor even if an AI bubble pops.
And even if it turns out that Anthropic takes over the world, Google still has a stake in them. (Although so will everyone else though SP500 by that point.)
I don't understand why would you invest in Anthropic IPO instead of GOOG, but as I said, would be interested to hear opposing thoughts.
I guess the opposite argument would be that AI success will undercut their search offering.
Sure, but in that case the Gemini arm of the business (or whatever they rename it to the next time) carries the torch forward. Unless there is no money to be made in AI and everyone loses, but then again it would still be better than investing in Anthropic.
> Sure, but in that case the Gemini arm of the business (or whatever they rename it to the next time) carries the torch forward. Unless there is no money to be made in AI
the question is if its enough money to replace 300B of Ads revenue.
If it isn't, are they in worse position than Anthropic?
Biggest Anthropic risk is that they have those many B hardware commitment.
Otherwise, they have some solid foundation model, which is cheap to post-train and improve, solid market share, they can milk this part for some time and be on the top.
Google is extremely vulnerable against paradigm shift if they lose Search Ads revenue, they have too much bloat of unnecessary people and systems complexity, and easily can collapse.
> I'm convinced Google is a much better investment and would like to hear an opposite argument if anyone here has one.
I have an argument, but it's mostly snark.
Like, Google are definitely in the best position here, but they've messed up so many times when it comes to product management/enterprise adoption that I am confident in their ability to snatch defeat from the jaws of victory.
I 100% agree with that snark and think Google could be so much better were it not for internal politics, perverse incentives and inability to focus and follow through with products.
Still that being said, I bet on them rather than against them.
> Like, Google are definitely in the best position here, but they've messed up so many times when it comes to product management/enterprise adoption that I am confident in their ability to snatch defeat from the jaws of victory.
Yeah, but what about Microsoft? They're the real threat here: they can get or train an open-weight model, embed it into Windows at very little cost to the end-user, and then the end-user will have AI in every app that they run, regardless of whether that app has AI connectors already or not.
The entity holding the business relationship already has the most clout, and while Claude takeup was good, it has nowhere near the same numbers as Google enterprise, which in turn is nowhere near Microsoft.
In terms of market-cap, these are all trillion dollar companies. In terms of enterprise agreements, if you were to draw a line-chart for enterprise volume for each of those three companies, MS would be the very top of that chart while Google would be somewhere at the bottom. Anthropic wouldn't even show up; their line would be fractions of a pixel tall, at its peak.
> Anthropic picked the better market at least for the short term.
What market is that? How is it distinguished from OpenAI or any other LLM company?
Anthropic is much more focused on B2B while OpenAI feels more B2C. Both sell to both markets, but the focus feels different and I believe the resulting customer base is quite different.
I think the implication was Anthropic has gone after the B2B market (with Claude Code and Cowork), where as OpenAI has gone after the consumer market (with ChatGPT). I'm not sure I totally agree to be honest; Claude Code was a happy (and fantastic) accident if the stories are to be believed.
> I'm not sure I totally agree to be honest;
Even before Claude Code, Anthropic were way more focused on B2B (source: a former company started using them in late 2023, based on our internal evaluations).
Capital is more worthless than ever and the society is relatively poorest at the same time. Interesting times.
This post makes me realize that, despite being an enthusiastic Anthropic customer for near 2 years now, I actually don't know any of this historical context.
Down the rabbithole we go. Oh clauuuuude! :D
I wonder on how they're planning to impose that valuation on the people of Earth.
I am totally blown away that a company who does not turn a profit, whose revenue is slowing down, and has hard financial commitments 25x times, the money they just raised, over the next four years, was able to get 81 billion from investors.
As predictions are only worth anything, if they are made ahead of time...I predict their share price will lose 25% of its value,over the next 3 months. Drop to 40 to 45% of IPO price, after their first public quarter release.
All of what you said both pre and post ipo also happened to Facebook.
Facebook had massive future liabilities at IPO?
Facebook didn't really have a "there's a non-zero chance that might replace a whole lot of industries" moat.
Does Anthropic?
Non-zero chance? Yes. I wouldn’t bet on it, but I can see the market’s appetite.
Facebook also didn't have "there's a very large chance we'll be completely replaced by free Chinese models" (which curiously Anthropic doesn't emphasize in their S-1 filing).
This. Chinese models are indeed a threat. But the bigger question is world is continue to give them power to purchase lobbying power to release more powerful models, collect more useful data at the risk of everyone's privacy and control everyone. People need to realize how much data they continue to give to models and their training. The more they use the more data and powerful they become. And Chinese models take way that data collection. And that's the threat. It's not the revenue from models. It's the data that they can't collect is what they scared of. And that's why they put a safety narrative in disguise.
Facebook had that TikTok problem, and all it took was to get the government to force sell it to reduce the competition.
Government can do a lot of wonders in this game.
Facebook IPO was 2012 and Tiktok was launched in 2017.
My point is government can get involved to resolve a lot of "Chinese problems" if needed.
They have about $1 trillion commitment in compute over the next 5 - 15 years. I'm not sure where you got the 25x$81b from.
$1t notional / contractual but the negotiable terms are unique
But most is concentrated in the first 3-5 years. Is pretty bad for their cash flow.
The first 5 years estimate is around ~$355b–419b. We don't know for sure because most of their deals actually span 5+ years, up to 10-15 years.
They're at $86 annualized revenue right now. If they grow at 0% for next 5 years, they will generate $430b, which is enough to cover all compute costs.
Well... I guess tackling technological innovation that has a potential to transform human civilization on the most basic level does that for you. It's not acompany that releases 15th overpriced gadget to the increasingly poor populace.
but company that releases 15th overpriced gadget already transformed human civilization. Im pretty sure anthropic will have it's time and become a company releasing 15th overpriced model to the increasingly poor populace. Capitalism is cyclic: expansion, peak, contraction, trough
They’ll pop and maybe drop, but Anthropic is the Microsoft of the AI age. This is a generational company that will define the future in so many ways.
Notionally.
The faith/hype vs objective data doesn’t support the valuation.
It is entirely dependent on having faith that this will work as expected.
> This is a generational company that will define the future in so many ways.
Maybe. Or maybe not.
It’s already a generational company - $1T at year 5 is no joke.
It has already defined the future in some ways. Claude code has already part determined the future of software engineering, for example.
How can charge a premium over competition if they catch up? Branding, smoke and mirrors won’t last long in llm industry. Big corporation switch AI at the blink of an eye. And are already looking more into self hosting.
Well, was anyone else capable of building an operating system other than Microsoft?
The answer is that business is not just technology.
On the technical side, it's also important to see beyond the current market, what's currently possible, and broaden to - ok, many innovations are coming in AI, who is most well-positioned to be the centre of that?
Anthropic is not like Microsoft. You can’t switch os in one minute. And how many did something similar to Claude? Hundred
Right now, you’re not entirely wrong. In five years? No chance.
I guess that makes China a generational country that will define the future in so many ways
So long as their VPNs to Singapore stay working.
China has existed longer than a generation.
It has already defined the future in many ways.
Or it could be Netscape...
Google could "Internet Explorer" Anthropic to death with a superintelligent ad stuffed free Gemini. shivers
> Or it could be Netscape...
> Google could "Internet Explorer" Anthropic to death with a superintelligent ad stuffed free Gemini. shivers
Hell, Microsoft could do the same with a cheap inferencing model; if people simply use copilot because it's already embedded into all of their applications because it's embedded into the OS (Office, Windows, anything that runs on Windows), only a few dedicated people will continue using the expensive paid-for options.
let's hope not and instead root for a more evenly distributed AI future with open weight and fully open models, running on local hardware with privacy.
I think there's a future for open weight models. To continue a half-assed analogy, Linux has thrived whilst Microsoft has captured the value. They're co-existed.
Fully open models will be more hobbyist, I expect. Open weights will continue as a strategy but probably not as prominently as today.
Why not?
Because open weight models are a business and political strategy without a strong business model.
Training is very expensive and open weight models don’t generate revenue comparable to frontier labs.
The reasons they exist are mostly “commodify your complement” and (in my, apparently controversial opinion) the Chinese govt trying to take down the American economy and undermine the US frontier labs in order to compete in the AI race.
People are so accustomed to free software (that they by and large never contribute anything to) that they get confused about why we can't just have an economy of it.
>This is a generational company that will define the future in so many ways.
You can say the same about the Company of Scotland.
> but Anthropic is the Microsoft of the AI age
Possible, as long as everything goes well for them. Their debt deals don't care how much of a transformational company they are. If they get less demand than they expect, or slower demand than they expect, they could end up bankrupt.
If their business case is true, it has withing itself the seeds of their own death. I can ask a super smart self improving model to create a ...super smart self improving model ,-)
If their bussiness case is not true, we will be paying for this bubble in higher interest rates, for the next 25 years.
Now tell me: Which scenario do you think is more likely until 2030?
A - We cure cancer, create self improving models of hyper intelligence, that will solve Fusion and bring us to the stars, create a CPU, an OS and a 3D printed Car and deliver it to you within the same afternoon, solve Earth warming issues...
OR
B - This biggest credit bubble in history, will put the sovereign debit of the US and 30 year Treasury to 8%, force the US government to default, cut spending, increase taxes and bring about a recession that will take 25 years to recover from?
Microsoft left their competitors behind pretty quickly didn't they? They had no serious competition in desktop OS and office software for many years. Anthropic has a near equal in OpenAI and compared to Google they seem to have some advantages and some disadvantages to overcome. That's not even mentioning China models.
I am reminded of the infamous "What were you thinking?" comments from Scott McNealy. I know it's practically a cliche, and you can summon its ghost by saying "bubble" three times in a dark room lit only by a candle.
And yet.
Investors regularly place risky bets. This one if it becomes a hit is worth every risk I guess.
> As predictions are only worth anything, if they are made ahead of time...
No, predictions are only qworth something if you are willing to put money behind them.
good chance those financial commitments will be worth more then they cost as the hardware crunch continues
many large enterprises in this space in the USA is operating like Korean companies have for a good 80 years now. They call it 'chaebol'. Chaebol arose from family enterprises, but it really doesn't matter who or what are the controlling interests
cross shareholdings, agreements to buy, technology access, barriers to entry, assimilation of smaller upstarts into the chaebol, bonds, lending....
I am blown away by how VCs pump up some companies to eventually get access to your 401Ks without any questions neither from government nor from public.
I will be double blown away when bubble pops and some companies go bankrupt or bail out by the government, who already paid VCs with your money.
Winners: VCs
Index is proportional to free float, so Anthropic will be treated as $100B company by index, not $1.5T company.
for first few months.
SpacesX float already increased from 70B -> 600B
It's a bubble.
I am not saying AI isn't useful. I am not saying these companies don't deserve any valuation.
But it's a bubble, because investors want to get rich fast. The CEOs want to get rich fast.
What they don't think about is: When the bubble bursts, it will be difficult for many, many people. Maybe including them. Hopefully including them.
Because karma is a bitch
It’s not a bubble, enterprise is just slower to adapt than expected
Funny that you say that because from the reports Ant are showing only their summer revenue numbers - when all the enterprise employees had open accounts. Since then the enterprise clients have been jumping on their own infra and open weight models.
So yeah enterprise is just slower to adapt and they are adapting away from providers like Ant.
Ant only chance to somehow survive is regulatory capture where they will be only allowed provider for gov contracts.
I think where we’re going with AI money will have no value or at least no value as we understand it today. Once they will achieve AGI it’s just new ground economically
Do you have any proof there is even glimpse of anything remotely AGI comming from LLMs?
It seems like corpo developers who were working mostly on self inflicted logic/math puzzles at best (more likely aligning UI) think that their work was so hard and precious that their partial automation mean everything in the universe must be solved.
The money have will have no value because the system implodes. And if we somehow magically have
AGI it is very bad for us digital peasants because the money system will stay it will be just 5 overlords who will control it while rest of us will beg for scraps.
For software generally we’re almost there because coding is already automated. To close the feedback loop we need formalized verification which is already growing. In some companies the loop is already closed, watch poteto’s talk about how they work on grokbot.
We’re still kinda far on the physical level and robotics have some work to do for catch up but we’re closer to software Agi than ever
Then we have a different definition of a bubble.
For me it's a totally inflated valuation of the whole market.
It is a financial bubble. No one is denying its usage, but the math just isn’t mathin’
One crucial key assumption is that customers will simply pay whatever the AI companies charge once the subsidized gravy train is over. I do not believe that. Right now a small share of paying users are paying for most of it (via API), it is a whale economy.
But models are getting cheaper over time per task. I wouldn’t worry about pricing, it will only get lower
>> One crucial key assumption is that customers will simply pay whatever the AI companies charge once the subsidized gravy train is over. I do not believe that. Right now a small share of paying users are paying for most of it (via API), it is a whale economy
> But models are getting cheaper over time per task. I wouldn’t worry about pricing, it will only get lower
Isn't that even stronger evidence of a financial bubble? Once the whales are gone, where do you get margins high enough to support the share price? If the margins and resulting revenue can't be found, the bubble pops.
If models were getting more expensive over time, then sure, I'd expect the AI shares to rise (because people have to buy from them, no matter the margin), but if models are getting cheaper over time, all those companies will be in a race to the bottom with open/free models.
Money won’t matter after ASI/AGI will be achieved. It even doesn’t matter now. This is not just another tool but life altering event for whole civilization
> Money won’t matter after ASI/AGI will be achieved. It even doesn’t matter now. This is not just another tool but life altering event for whole civilization
Yeah, the asteroid that killed off the dinosaurs and sent the remaining half of all organisms extinct was life-altering too, you know.
If there are fixed number of tasks, and cost per task is going down, will profits go up or go down?
Or maybe the task pie can expand indefinitely, but I'd need to see the models handle tasks other than code and general knowledge work (much of that not so well either) to think that.
the valuations for these random SV "AI" companies these days is just eye watering. It really does seem like theres so much money concentrated at the top that they can throw vast amounts at the wall to see what sticks. Anthropic and Open AI are generational companies but even their values seem extremely inflated. SV has mastered the art of juicing companies to the gill before releasing them into the wild and they've done it at record speed for these two.
Is $1T now going to be the norm for being a Unicorn as a startup?
It doesn't say much about the company itself, but about the state of the economy, where all the money has been concentrated at the top, in the hands of people who have nothing else to spend it on than on more financial assets. So there's enormous amounts of money available for more stocks.
It's why stock markets and corporate profits are at an all time high, while most people struggle to make ends meet and governments are woefully underfunded.
That $1.5 trillion IPO is the result of a completely lopsided economy.
The K shaped economy is developing serifs.
It's K shaped economy with the upper arm extended with additional /
We're going to need an additional Unicode character for that.
Well, yes. But also the company is projecting to sell pretty much every white collar worker a $200/mo Claude subscription, which is an absurd amount of revenue compared to said Apple, which promises to sell a new phone to 20% of people every few years.
"compared to said Apple, which promises to sell a new phone to 20% of people every few years."
- around 50% in the US market
- they offer way more products, than phones, are the leader with laptops for instance
- take a big cut of every store purchase
The US is just ~350 million out of the ~8000 million people on the planet.
What % of those 8B people can afford an iPhone? How many of those people are located in the US?
It's hard to put a percentage on it, but I'd wager for 90% of the earth's population, paying $200 a month for a personal LLM subscription isn't viable or simply not worth it.
When something like 60% of the world’s population lives on less than $200 a month, it’s a preposterous suggestion.
But will it be Claude? Probably it goes through Microsoft through copilot. In which case it becomes a race to the bottom. I never need Fable even today: Opus suffices on 99%. So why do I need anthropic? Opus will be replicated in half a year by every lab.
In that sense I am super bullish on Microsoft as they have the vendor lock.
For anyone who has used the various tools, the harness still matters and copilot is pretty bad. While Anthropics models lose on cost compared to OpenAI, there is still something that works really well in Opus paired with the Claude desktop app when it comes to tool use and solving tasks.
Would upvote twice if I could.
Another channel for the wealthy to place the money they don't know where to spend anymore. And the audience applauds
Also, it's a valuation for selling to asset holders, no such profit has been achieved yet with the actual portfolio.
It's like an IPO for a beloved artist who claims he will sell a painting to every household in the world.
> "It's like an IPO for a beloved artist who claims he will sell a painting to every household in the world."
So not that far away from the usual "let's all buy from this artist and push his works to a billion dollar" scheme the wealthy usually follow.
But now with tech...
Well at least profit from the appreciation of artworks can be written off a little easier. Stocks are too liquid and regulated to serve the same purpose.
But they can be lended against just fine, which is apparently the common scheme for tax avoidance these days.
If enough value depends on the same stock, there will be an intrinsic need to keep valuation high which is completely decoupled from the actual business of the company.
Which I believe is how we got a company to invest in the press-conference of a flying car...
Oh yeah that makes sense. I assume that lenders will price in the inherent risk from the volatility of the asset, so it will take significantly more stock (at the current valuation) relative to the borrowed amount?
Yeah, seems to be common to borrow 50-70% of a stock's value against the stock.
The neat thing is that instead of selling the stock once (and pay capital gains tax), they borrow against it for low-interest loans or lines of credit (which are debt rather than income, so generally not taxed).
This way they can access cash while the underlying stock continues to compound in value, and pay a low interest-fee instead of taxes...
Who knew that having wealth accumulate instead of flow would be bad for an economy that relies on the allocation of capital? Every economist ever.
Billionaires are the economic equivalent of a tumor. I don't know why we're so obsessed with perpetuating them.
Because billionaires and those right below them spend a ton of capital on perpetuating the idea that the average person can be like them. Hell they even have their own podcasts now lol
> "...that the average person can be like them"
...or become a beneficiary of the benevolent billionaire.
My favorite quote comes to mind: "The wood will not vote against the axe, because it has unrealistic expectations to become an axe-handle one day"
It's an absolute state when hackernews from all places has degenerated into braindead leftist "economics" echo-chamber.
I love how the exact same people I describe above (normal people who've be brainwashed to think they can achieve the same level of success that only ~3k billionaires have had), have also been programmed to automatically dismiss anybody that writes a sentence that uses the word billionaire negatively, as "braindead" leftist economics.
They have brainwashed you to the extent that you spend your time reflexively defending them in the comments of online spaces. You become an activated foot soldier, coming to the defense for the very people who seek to enslave you the second anyone says something bad about them.
You proved my point.
Nope, I just understand basic economics and have logical deduction skills instead of buying sentimental bullcrap from demagogues. Please read a book, you will be doing all of us a favour.
What book do you think would change my mind?
For your level even basic economics books from any 101 course in a university (google away) would help you a lot.
Perhaps you should be the one to read that book. None of the things discussed here are in any way controversial, except with the billionaires and their politicians. Many economists are saying the same things we are saying. That's where we got it from.
I took many economics classes, its partially why I have informed opinions on the matter. You're so confident reading a book would change my mind, surely you have a reccomendation.. Or are you just politicing? Maybe lay off the X algo.
This is an interesting position.
high volume, 'flow', of wealth is usually what economists call a hot economy, which usually leads to inflation which then usually leads to problems that a central bank has to mitigate or consumers have to bear the weight of.
I know this is getting into the weeds of economics ...the OP was posting about a pending IPO...just setting something out for consideration....
There is something more though I think in terms of lack of cultural capital.
There isn't enough cultural activities we value in society so we have largely turned our culture into this kind of net worth sporting contest. Billionaire is just round number bias that reflects a goal post in the sport at the highest level.
Late neoliberalism I believe is the right frame but this is still a rather obscure idea.
It's annoying that people all readily acknowledge that billionaires do not have piles of cash, but then they incessantly talk as if they do.
My friend, the billionaires wealth is the economy. It is the thing that is flowing the money.
Jeff Bezos wealth isn't cash redeemable pieces of paper with "Amazon Share" stamped on them.
His wealth is warehouses filled with goods, software running logistics networks, CPU's serving web services, trucks delivering packages, and a workforce of 1.5 million people all aligned on serving paying customers for their needs or wants.
So while you can detest it and hate it, at least recognize that that is what you call "accumulated wealth", not a pile of money where we are one legislative act away from busting open like a pinata.
If you snapped your fingers to distribute his wealth, you'd feel pretty dumb when you end up with 13 aluminum dolly carts as your share.
So how did Jeff Bezos become one of the largest private land-owners in the United States, if he has no wealth?
How did he pay for his wedding, with non-cash-redeemable pieces of paper?
Or how did James Dyson manage to become the biggest agricultural landowner in England?
And who was then buying e.g. Van Gogh's art for 400M USD just this week?
It's equally annoying that people readily acknowledge that billionaires have no cash because they openly operate a tax-avoidance scheme where they lend money against the stock they own, to benefit from its value while it keeps growing and "officially" are poor.
100M USD of stock create instant liquidity of 50-70M USD at low interest, without ever needing to sell the stock.
You understand they need to pay those loans back, right? Those loans just defer taxes, they don't avoid them.
The actual tax dodge is mostly charitable donations being exempt, and the step up basis of inheritance.
However neither of those things have any real bearing on a wealth tax.
Which technically already exists because Amazon pays taxes every year, which is to say, his wealth is paying billions in taxes every year already.
> It doesn't say much about the company itself, but about the state of the economy, where all the money has been concentrated at the top, in the hands of people who have nothing else to spend it on than on more financial assets.
Huh, what?
> So there's enormous amounts of money available for more stocks.
Money (nor stocks) don't have to change hands for stocks to become more valuable. All that's required is that the likes of Citadel and Jane Street quote higher prices, no trading necessary. (Of course, they can and will only quote higher prices, if that's justified by supply and demand.)
> It's why stock markets and corporate profits are at an all time high, while most people struggle to make ends meet and governments are woefully underfunded.
Government expenditures have grown like crazy in much of the world, haven't they?
That's not how supply and demand works. There has to be trade in order to establish a price.
And governments in much of the world have actually been cutting expenditures, leading to serious shortages in many places. Increased poverty, reduced quality of education, crumbling infrastructure. Governments were better at this from the post-war period to roughly the 1990s, until neoliberalism and austerity became the law of the land.
> That's not how supply and demand works. There has to be trade in order to establish a price.
Why? Anything you can potentially trade against, especially with a tight spread, counts in my books as a price.
> And governments in much of the world have actually been cutting expenditures, leading to serious shortages in many places. Increased poverty, reduced quality of education, crumbling infrastructure. Governments were better at this from the post-war period to roughly the 1990s, until neoliberalism and austerity became the law of the land.
https://en.wikipedia.org/wiki/Austerity says
> Austerity is a set of political-economic policies that aim to reduce government budget deficits through spending cuts, tax increases, or a combination of both.
Government deficits have reached record highs in eg the US and France. Where is that austerity?
Neoliberalism is, alas, dead. The glory days of the 1990s and early 2000s died around 2008.
The K shaped economy trope is directionally correct, but you have the cause backwards.
It's not that governments are underfunded, the K economy is a direct byproduct of government stimulus across all developed economies for the last 15 years.
When you inject the market with QE, the money printer and most importantly; 0% or negative interest rates for 15 years...it massively inflates the value of assets. Rich people own assets.
This is why things in the US with the prospect of generating future cash flows are at historically high valuations relative to GDP: public companies, private companies, real estate, etc.
If you think government is 'underfunded' and the solution is giving it more money to dump into the economy, ironically, you're going to inflate asset values further and make the K economy worse.
The stimulatory tools that are easiest for government to use are monetary policy, not fiscal policy.
But let's say you do manage to pass populist fiscal policy, like, giving first time home buyers $20k to help with a downpayment on a house.
Surprise! Turns out you've simply raised the value of all starter homes by $20K by increasing everyones capacity to pay for a scarce asset, inflating the value of the asset and handing the money to the person selling it.
The solution is to increase the supply of assets (build more real estate and competing businesses), not stimulate demand for existing ones. But that requires a much more rational, long term policy view than dumping blankets of cash on angry populist voters.
Alternatively you could massively tax the wealth of the very rich and revert the concentration. Then you are not competing with them for the scarce assets.
Again, giving government more money to dump into the economy will do literally nothing to solve the problem.
You will simply be taking money from the rich, and then giving it right back to them by inflating asset values.
Because the assets are the things that are scarce and valuable, and you're now disincentivizing the creation of them while stimulating demand for them more by dumping cash on everyone.
There is only one path: creating more valuable things (more supply of real estate and businesses offering competing products lowers their price). China, ironically a 'communist' country, figured this out yet we're still struggling to.
The truth is we don't want to face this reality. We'd rather sit at computers, place fantasy expectations on the politician du jour to save us, and repeat smart sounding memes we don't understand for magical internet upvotes ('late stage capitalism' 'K shaped economy') instead of picking up a hammer and building anything.
Why not use the extra taxed income to support "picking up the hammer and build things"? Historically, government have been the one to do that for moonshoty projects, a lot of "things built" to kickstart SV was government spending (as much as modern SV propaganda tries to erase that fact)
Why does your mind equate "taxing wealth to reduce inequality" == dumping that money as hand out to the citizens? Why can't government, that after Thatcher and Reagan threw up their hands saying "we are too dumb to plan anything", just go back to planning long term spending (such as reducing construction regulations or make it more efficient, both requiring capital I assume), which, ironically, is also what you're praising China for.
With respect, it sounds like you're also falling for the criticism you give out to others about "smart sounding memes" ; yours one step further than the one you're criticizing, but stops at the step where more tax== hand outs meme
> Why not use the extra taxed income to support "picking up the hammer and build things"?
That would be the ideal scenario, but it just doesn't work reliably in a decentralized system, which I think we can all agree is better than a centralized one (I'll take messy decentralized democracy over brutal authoritarian control every time).
Democratic politicians are just renters. When's the last time you renovated the bathroom during a hotel stay? The one time the US made massive investments in building stuff in the past 100 years was when we had the most dictator-like president (FDR).
But centralized power can just as easily go bad (see the last 2 years) as it can turn out good.
The only solution in a democracy is to let decentralized market forces dictate the price of things, mitigate any anti-competitive flaws (monopolies, cartels, local government vetos against building, etc), and deal with the harshness of reality instead of looking for magic bullets.
The market takes care of itself if left to build unencumbered, just look at the price of housing in Austin and all the new assets being built in AI. The whole pessimist scenario in AI is that competition will cause prices to drop and kill valuations. This is good. Lowering asset prices is what we want!
I think we are broadly in agreement but I can't shake the feeling of "smart sounding meme" feeling from your reply too. Granted it's forum post that is limited form of communication... I probably sound the same.
You give one good(FDR) and one bad (Trump) example why centralization is bad. With your previous example of China and CCP, the tally seems two good example, so as a basis for discarding centralization I feel it's too handwavy.
I rather see the details of how centralization can be implemented.
CCP, for its centralization, have long been factional, localized and meritocratic in certain aspects (you only climb the party ladder if you succeed at local province first), for example.
Another is Japan's centralized zoning law, or centralized train track management, seems to yield better result for affordable building and public transportation: a central/decentral, public/private symbiosis.
I'm in agreement with you that "mitigate anticompetitive flaws" is an direction because it's clear certain domains simply can't be expected to construct competitive markets on its own, or that the markets outcome is beneficial to the society.
"We" is we citizen, government and companies. We can't directly affect the companies (except with our purses but let's be realistic on that), but we can affect our government. That entity completely outsourcing it's power to companies is stupid imo.
Create guardrails so that market competition is aligned to the broader society's wellbeing is a minimum to me, and the interest question is where and how. Being blanket anti tax or anti centralization robs us of being able to discuss that part.
I changed my opinion a little when I got to that discussion, about wealth tax for example (how hard it is and how there are lower hanging fruits such as what Richard Murphy have been claiming: https://youtu.be/Iir0GP40_vw) and it seems like more proactive approach to this discussion
I don't know how centralised you consider the US or western Europe, but they have done exactly this very successfully after WW2. They invested enormously in infrastructure, housing, and economic growth, and it was incredibly successful. It was only after neoliberalism came along that people started to believe this was bad. And that's when the excessive inflation of asset prices and growing inequality started.
There is no scarcity of shares. They're entirely artificial. Their absurdly high valuation is really just because the rich have too much money and nothing else to spend it on. Giving that same money to the poor instead will not inflate those same prices, because the poor will spend it on other things.
And those other things the poor will spend it on are also already inflated because of a lack of competition. Corporate profits are at record level across many industries, despite the fact that 60% of Americans are struggling.
Your scarcity argument only really holds for real estate, but that too is exacerbated by rich people buying real estate just because they can, because they need to put their money somewhere. Meanwhile poor people are struggling to buy their first home. Poor people are not going to buy multiple homes, they really just need the one.
> Again, giving government more money to dump into the economy will do literally nothing to solve the problem.
You’re assuming that taxing the rich would mean giving the government more money, but that doesn’t have to be the case.
You could, for example, raise taxes on the rich to the point where income tax is dropped entirely for the vast majority of people who currently pay it.
You could calculate how much the federal budget would need to fill the gap created if, say, nobody earning under €100k a year had to pay income tax anymore. That sum would then be offset by raising taxes on the richest people. No extra money gets “dumped” into the government. It’s the same amount as before, except a large share of it is effectively handed back to people who earn their income from work.
You could take every dollar of wealth Elon has and give every American $2,500. Is that really going to help?
More importantly, taxing the rich based on unrealized wealth means forcing them to sell their assets. Markets would be flooded with new supply and no buyers, tanking prices and putting everything in a much worse position.
Wait, how would lower prices be bad? Isn't that exactly what we want?
People want groceries and house prices to come down yes, unless you already own a home. People don't want prices to collapse though, and more importantly our government doesn't want prices to come down.
Deflation is a death spiral for debt markets. Our economy is so precariously balanced on top of debts and derivatives that the whole thing would crumble if prices of major asset classes collapse.
I own a home and I'm totally fine with house prices coming down. I don't have my house as an investment, but to have a roof over my head.
> Deflation is a death spiral for debt markets.
People say that a lot, but is it actually true? Are there real life examples of this happening? By far the biggest example of prices coming down over a long period, has been computer components, and that never hurt those markets. It's only the recent rising prices that's killing markets.
> People say that a lot, but is it actually true?
Price deflation for any particular product isn't a big deal, the risk is broad deflation across markets and industries.
Our financial system is heavily leveraged, when prices fall leveraged bets begin unwinding as forced sales, pushing the spiral further and further.
You generally don't see price deflation hurt triggering a debt crisis as the starting point, but that's only because our system is structurally designed for persistent price inflation. You do see it in the middle of a crisis though, debt is the dry brush that burns everything down after the fire is lit.
The great Depression was mostly a deflation death cycle. Japan between 1990 and 2010 ish was a less dramatic example.
>I own a home and I'm totally fine with house prices coming down. I don't have my house as an investment, but to have a roof over my head.
It's an easy counter point to make because you don't have to show anything for it. But when we ask people to put their money where their mouth is, they reliably protect their home values.
To put another way, say you needed to move for whatever reason, and in the new location you found a home you loved for $100k under the value of your current home. Would you sell your current home for $100k less because all you care about is having a roof you like over your head?
Again you can easily say yes to this, but in reality, everyone pockets the $100k.
I expect we will be seeing a similar scenario play out with Anthropic's IPO. Its all well and good to call oneself an altruist, but once that check comes in I don't expect them to give it all away.
I will be quite impressed if the EAs at Anthropic do precisely that though.
The issue with accepting this status quo of tax policy is that it isn't the same in other countries. The US (uniquely?) does not tax asset wealth when it is distributed which allows for huge accumulation of wealth outside of the tax base, Canada has deemed disposition for example that brings any asset that is distributed from one individual to another into the tax base even if the asset remains the same asset (unsold stock that ahs appreciated).
You can tax "unrealized" wealth when that wealth is used to actually fund expenditure, currently anytime this wealth exchanges hands as the same asset it is seen not as a realization of wealth, which is what needs to be changed.
I'm very much in favor of closing a loophole that allows the wealthy to borrow against their assets and have tax-free income.
I'm very much opposed to a wealth tax that forces a person to have to pay for gains they never realized even if that means having to sell the very asset in question.
I personally don't actually agree with income tax at all, but if we have it then it should standard across the board without said loophole and acting as a tax on income, not paper wealth that can't be spent.
> "The solution is to increase the supply of assets..."
Let's be concrete (and unpopular): If one of the few assets left to buy for billionaires is land, which creates competition among them for land not to live on but to HOLD and rent out, the solution is not to "increase the supply of land", it's to decrease the amount of billionaires...
Land is not actually that scarce. Land in cities is scarce.
We can build more cities. How do you think the current cities we have came to be?
Yes, it requires a longer term view than a 4 year election cycle. China has created hundreds of cities from nothing in the past 30 years. Many failed, many succeeded. This is the risk taking required to create new things.
We could do it too. But we choose to stagnate and inflate the real estate we already own instead.
Land in most cities isn't even that scarce if utilised correctly. Yes if you measure it as ground square meters it's quite scarce but if you measure it in number of apartments that could go there, there's usual plenty.
You are describing, almost fairly well, issues of accumulated wealth at the top -- particularly the effects of covid handouts[1], where all the money ended up with landlords that were not allowed out to spend it all, significantly increasing inequality and asset prices.
And somehow you try to turn it into an anti-tax argument, when taxing the wealth at the top is the only solution, the way to control the money supply, time machines aside. Give out too much you have to take some back if you don't want to devalue.
1. which is why the stock market and housing market went crazy at that point, not a steady 15 years, $5tn was created
> Give out too much you have to take some back
This would be true only if the government simply increased taxes on assets (say, doubled capital gains taxes) and then took that money and paid down government debts with it.
However, we all know this won't be the case. The money taken from asset holders will simply be dumped right back into the economy via populist handouts, and that money will ultimately just re-inflate the things that are scarce and valuable (assets).
You're disincentivizing the creation of assets at the individual level (by taking them from people) and stimulating demand for them at the societal level by giving everyone money to chase assets with. A recipe for more runaway prices.
Again, there is only one path: increase the supply of assets (build more real estate and more competing businesses). China has similar/lower tax rates than the US in most areas while their middle class grew massively. This is how. It's not magic.
> The money taken from asset holders will simply be dumped right back into the economy via populist handouts, and that money will ultimately just re-inflate the things that are scarce and valuable (assets).
What's a "populist handout", specifically?
I wouldn't count the endless corporate and agriculture bailouts as populist, personally, but I'm curious if you would. As a salaried middle class American who owns real estate in the range of $500k in assessed value, the only handout I've ever received since becoming an adult half-my-lifetime ago was the Covid stimulus check. Every other "handout" that I've received was a tax break, e.g. the lifting of the SALT deduction limit.
"Populist handouts" (by my definition) are a boogeyman that I've never actually personally experienced in the USA, especially compared to (1) the tax breaks that are handed out left-and-right, (2) corporate and agriculture bailouts, and (3) the military-industrial complex.
My soap box: I think what made the New Deal so aspirational was that, yes, the government was giving people money, but it was also paying them to perform jobs that made the nation a better place to live. We have so much crumbling infrastructure. Instead of handing corporations the money directly, make them bid on infrastructure projects. Then heavily fund a department that roots out fraud and abuse and make it culturally taboo, through legislation, to perform fraud and abuse.
All those things you listed are populist handouts to specific congressional districts to please the extremely narrow interests of the people in that area at the expense of the greater nation.
This is my point. Any money that comes in will be spent on handouts at a factor of 2X the amount that actually came in, further inflating the deficit and entering the money supply to ultimately inflate things that are scarce and valuable (assets).
Also, I'm in full agreement that it would be great to invest in building infrastructure and rooting out fraud, but we just can't do it any more now that so many people's incomes are dependent on making sure nothing ever happens.
"populist handout" is a extreme, but you have to admit that French going nuts over raising retirement age when they have deficit 5% of GDP is also a bit extreme.
Maybe both sides need to step back.
But if you frame the debate as US-only then I basically agree.
It doesn't matter how all of the money ends up at the top. Yes, government handouts to the rich are a big part of the problem, but so are a lack of taxation on the rich and on corporations, a lack of competition, the fact that both corporate profits and inflation are high, and the lack of stimulus for the poor.
You seem to think that the government can only use its money for more stimulus for the rich, but that assumption is itself part of the problem.
If the government gives more money to the rich, whether through tax cuts, subsidies, handouts, or whatever, the "trickle down" lie has always been that that money will eventually help everybody, but that's only the case if that money ends up in the economy, paying wages, etc. But instead, that money is being used to buy up more assets. That doesn't help the economy in any way, it just skews it.
If instead, that money ends up in the pockets of people who need it, they will immediately spend it, buy food, buy other things they need, pay off debt, and all of that money creates demand, pays wages, and grows the economy in a way that helps everybody, and doesn't just inflate the numbers.
Further its because income generated by assets are tax-efficient in the western world (well at least North America) in a way that non-asset generated income (i.e labour income) is not. Naturally the inclination is to not just accumulate assets for their own worth but for more tax efficient income as well.
If asset-based income was taxed similarly or included in the tax base, you wouldn't have stimulus become inflationary in the way it is today.
Uh, yeah of course a person is super rich for having $1000 or $100 invested in the stock market.
No, of course a person is not super rich for owning their own home which is worth hundreds of thousands.
God damned young yuppies with their stock portfolios worth $20 000 or more! In the meanwhile grandpa is suffering in his house which is only worth $1,3 million.
If only we could take the money out of investment into businesses and put them into investing in real estate! That's where all the jobs are made who pays people's salaries. Can somebody call the EU to have them regulate?
Wait until it explodes to 3-4b shortly after IPO, then... I think that despite the heavy selloff in the AI trade this las month we are not yet past the peak frenzy of the cycle.
Where's the growth? Anyone really think this will hit 5T? 10T? I can't see that happening.
What are the odds you think they hit RSI?
10%? 1%?
Or do you think RSI won't be as significant as AI researchers think it will be?
Valuations are difficult for this. It's got high chance of being a bubble, but nonzero chance of being something unique across human history.
Based on a 6% float of shares. Math is doing a lot of work supporting this market cap valuation.
That 980M is more than they had previously offered (825M), so they did expand the IPO.
And they're taking around ~105B in cash.
81.34B from the 980M shares at $83 and 26B from Google/Nvidia/etc, minus 1.9B the existing holders selling 23.6M shares.
Cashing in as much as they can, while they still can!
Perhaps they were expecting to be more oversubscribed? Like they increased the number of shares offered to keep a target amount of money. Likely they see the rough few months ahead. Ramp.com AI index shows spend on Anthropic AI was in decline/flat July 31 - Oct 31
Likely they're best bet is to weather the continued devaluation of tokens and cheaper models until they hit on a better business model, which honestly is probably not that hard.
The previous range was 70-78 and it priced above the target range, at $83.
So they are selling 20% more shares and selling them for 15% more than the midpoint of what they were targeting, because the demand is overwhelmingly high. They are going to make $20B more from the IPO than planned.
The rumor is they were oversubscribed by more than 6 times over. SPCX was reportedly only 3.5-4.
Oversubscription was kind of a self fulfilling prophecy. Large investors and institutions placed larger orders knowing the IPO would oversubscribe and they wouldn’t get their stated allocation.
It's a common pattern for IPOs, regardless of demand. Companies gauge the price range during the roadshow, make an initial announcement at a low price, then raise the target at the last minute going into the IPO to make it look like a hot offering.
Here's the index mentioned, if anyone else was wondering like me: https://ramp.com/data/ai-index#adoption#models
That's pretty biased. Ramp is used by tech companies and tech companies are heavy AI users. Small businesses don't use Ramp
Anecdotally I’ve seen small businesses using ramp
I wonder if early adopters are overrepresented in Ramp's customer base, so if they start learning how to get the best bang for buck in their AI spending at the same time the laggards are starting to spend real bucks in AI, Ramp's data may be bearish while the reality is bullish.
Thanks for sharing. Something seems off in that data. It says in Nov, Anthropic's share of token volume is 55.1% and share of token spend is 49.3%. Given Anthropic tokens are pricier than other vendors, one would expect share of spend to be higher than share of token volume.
That's a good point, likely there are other discounts / subsidies being realized on accounts? I think from Ramps point of view it might be hard to distinguish these.
That being said, I have also noticed that at times the X axis dates of the weeks changes. In the past, like a week ago, I noticed that the "Week ending Oct 23" changed to "Week ending Oct 25". Which also noticeably changed the shape of the charts.
I work with data a lot so cut them some slack, but I do with they had more grounded information in the report (a lot of % and not a lot of specific values).
I also wish they would somehow tie in to the massive OpenRouter Rankings. I think the two together would help paint a much more cohesive picture of how token budgets are spent and change week to week.
Perhaps people have learned to use cheaper options (e.g. Claude subscriptions vs APIs which is subsidized).
And RH allocated me 1 pre-ipo share against my request of 12 of them. so one in a billion shares. RH shows I'm the 260,000th retail investor of the ANT offering (at least within RH ecosystem[?]). Just observing, but pre-IPO share allocation does not appear outwardly to be an activity rooted in free-market fundamentals.
Markets can remain irrational longer than you can remain solvent.
Be careful out there.
Allowing employees to sell immediately, I don’t think that is a good sign.
It’s a tiny percentage of the company though, if I’m not mistaken. I’m frankly amazed it’s not way more. It’s something like 0.20%
The cap table of the founders are in the S1. It's not a small amount of shares.
Hang on - I believe it is 10% of almost all employee's stakes (excluding cofounders and a few others)
As I understand it, it’s ten percent of total grants, not vested shares. So if you have $10M in grants, $3M vested, you can sell $1M of your $3M.
Why is that a bad sign? I think it's weirder to restrict employee selling, makes it look more like you need to hide something.
Just allowing them to sell if they want doesn't seem problematic to me, unless they all (or a large number of them) actually sell immediately. Then obviously something's wrong.
Note that this is an actual question, I'm not even close to an expert with these things.
Because they’re dumping their shares immediately. If you genuinely believed in AGI, why would you sell? :)
Because after AGI money will be worthless
Maybe they genuinely believe in ASI.
If you believe in agi you’d sell and live life now. Not hope for more cash later
Shhh. Most people have no idea what's coming.
Is that a given (or has this happened in this case?)? Why wouldn't at least some employees who think there's a future for the company hold on to it?
If AGI becomes a thing, competitors will also achieve it. Open models, too.
At which point any model moat will vanish. Might as well cash out, and put your money into the compute providers.
Wont the AGI kill competing AGIs and humans too?
source: Terminator 2 movie
> why would you sell?
To diversify.
What diversification could possible beat AGI?
Depends on the goals of the AGI.
Diversifying to a nicer home in the countryside and a European Cruise while also visiting Disneyland Japan?
I'm more of a skill based person. I'd probably spend the money on learning a language, learning to fly airplanes and buying an airplane.
i.giphy.com/lGClylJINhkc0.webp
No, it's the opposite. If you truly believe in AGI the more reason to sell, asap ;) Money, at least, is likely to be worthless. Assets maybe not. Who knows what the future will hold, but all things considered it makes slightly more sense to hold tangible assets than 0 and 1s.
If everyone working there could cash out everything immediately, and be set for life (eight figures), I seriously think many would hand in their resignation and the company would face some operational crisis.
Why? They have AI. AI will do their work! /s
Insider selling coincides with lowered stock price pretty much universally. Lock in signals trust. It says to investors, we're in this together and you can be assured we're equally interested to take the stock price up.
It's a bad sign because it indicates that they're hoping to sell themselves.
The exact opposite. Companies restrict employee sale on IPO because they're worried that it will hurt post-IPO price if employees sell on the bell. This is a highly unusual show of confidence from Anthropic. Note that the IPO is already oversubscribed, so demand is well ahead of supply.
Most of those employees will have been advised to wait until next year, not much time to deal with the tax burden.
Overnight on Hyperliquid:
https://app.hyperliquid.xyz/trade/xyz:ANT
this link refers to HYPE-USDC pair
I think you’re incorrect?
The XYZ refers to xyz.trade perp contract.
Their spec for ANT is here: https://docs.trade.xyz/perpetuals/specifications-and-schedul...
Interesting, now it shows correct data! Im on Firefox/Win11, my first request landed on the HYPE-USDC pair. Strange.
Say what you will about their financials, but this IPO is largely going to be driven by how much raw demand there is for AI. And that demand is absolutely insatiable, at least from where we’re currently standing.
The recent cybersecurity incidents will only increase that perception. It appears to be weaponizable, whether or not that perception is grounded in any kind of reality. Whether it exposes vulns, fixes them, or all of the above, it just translates to more demand.
I think this will go up for quite a while before the world realizes the limitations. At least until their first earnings report. That said, I’m not a gambler.
> this IPO is largely going to be driven by how much raw demand there is for AI
Institutional investors are going to be looking at profitability, not "raw demand". Demand can go through the roof and companies like Anthropic and OpenAI still do badly if token/task prices are dropping fast and/or they are are not competitive on pricing.
Individual investors may be buying based on "AI is the future", with no regard as to profitability, and will likely be burnt as a result.
Short-term post-IPO stock performance is no indication of anything - part of the job of the underwriters is to "stabilize" (manipulate) the price, so it takes a while for organic supply & demand forces to prevail, and one might guess that supply will go up considerably once insiders are allowed to sell.
So who’s shorting? Genuinely curious.
(I will be buying shares.)
Well, all the underwriters are shorting. Cause of how the Greenshoe Option works
I don’t think there is much chance of this stock needing price support.
I mean, even though it's overvalued, it's clearly going to go up with all the speculation and the propaganda that everyone has been subjected to
looking at the past few months of Nvidia and neocloud stocks... they are running out of steam on the circular financing and fake demand.
This might be the latest to cash anything in.
I'll give it a few days, but I'm planning to buy puts.
If you want to go short, do not build your position today.
Apt username
thanks :)
Options should open up on Monday.
https://forbesmiddleeast.com/money/markets/cboe-could-list-a...
I will consider a grabbing a small position. The chances it expires worthless are really high, but also the volatility could allow for securing options that may become extremely valuable.
I would trade volatility (either direction) more than any kind of bearish thesis right now.
I suppose betting on volatility is the safest thing to bet on right now. At least on the short term.
The entire AI space is currently performing like Tesla. Unbelievable momentum, confidence, and capital. Day traders are going to have a field day with this, but one would need to have balls of steel to hold puts overnight.
Blue horseshoe loves Anthropic
I'm long but I wouldn't rule out going short if the price chart looks mountaintop like.
I am very curious how a publicly traded AI lab will use model releases as stock pumps. For example, if you are in the lead, do you sit on a new model until the stock price falters, and then release?
Possibly not that directly. Because training is expensive and that affects your quarterly profits and that affects your stock price. If you can reduce how often you need to train a new model you lower your costs.
Not if the stock has priced in a certain cadence of model releases. Same as the expectation of a new iPhone each year.
Stock price will be directly linked to having the best model (which could give OAI an advantage as they aren't competing with shareholder interests - yet)
Why would this be different than any other publicly traded company?
What do we think the stable state market cap for anthropic is? $200B? 2T? I struggle to believe this is worth like 5% of total US GDP.
Their run rate is $85B and they've been growing 1000% year-over-year so even $200B is pretty unlikely. It depends on if you trust their profitability.
What do you think of Apple at $5T, Google and Nvidia at $3.5T, SpaceX at $2T, or Meta at $1.5T?
They have 18B fully diluted shares. That's about $5/share revenue, and the ipo price per share is $83. The math implies a P/S of 17.5. SpaceX is 60, Palantir is 55, Cloudflare is 40, and Snowflake is 18.
The law of large numbers aside, and ignoring the insanity of the rest of the market when using valuation by comparables, the numbers aren't absurd.
Edited post. 200M was a typo brain still cannot fathom the scale of these numbers.
SpaceX is still >1T and close to 2T.
Valuations are completely detached from reality, it's pointless to try to make sense of them.
At least SpaceX has something nobody have been able to copy yet for a decade. So it makes sense, they have a moat.
> At least SpaceX has something nobody have been able to copy yet for a decade
Satellite internet?
Data centers in space, by the end of the year
"has something" refer to something that started in the past. You may argue they have promises.
They have demonstrated a fully reusable orbital launcher a few days ago, with at least 50 tons to LEO capacity: https://www.space.com/space-exploration/launches-spacecraft/...
(The first stage from this flight was not caught, but they demonstrated that during a previous flight & even re-flown one.)
but they're not valued 2T because of rockets or satellite internet, they're valued 2T because they have xAI inside them.
Which is by itself insane. They just caught a fucking orbital rocket on its launch tower - after delivering 50 tons worth of satellites!
Stock price shooting up ? No - but some new shiny AI bullshit shows up and people immediately scramble & stock go up and down at random...
Referring to the company that is projecting around 0.3% of US GDP in annualized revenue by the end of this year, having started at 0.02% at the beginning of 2026? That seems pretty reasonable factoring in that revenue growth rate.
Adding in that only a quarter of US businesses are using AI right now, that the company has set its sights on fundamental biology research and robotics, the company seems undervalued relative to the upside case.
>I struggle to believe this is worth like 5% of total US GDP.
Actual question: Why? America is mostly a service-based economy, and a lot of service-based jobs seem extremely vulnerable to automation. Do you think that the possibility of massive AI job displacement is overrated, or do you think that competition between AI companies will stop Anthropic from capturing that much of the market share?
Competition. For any service-based jobs, service automated -> service become commodity -> competition on price -> Anthropic cannot capture margin.
comparing a stock (ANT) to a flow (GDP) is a common finance mistake. You shouldn't compare a stock to a flow, they are measuring different things. https://en.wikipedia.org/wiki/Stock_and_flow
> However, one may meaningfully take ratios of stocks and flows, or multiply or divide them.
Right out of the section comparing stocks and flows
The OP was clearly comparing the GDP of the US to the market cap of ANT rather than taking a ratio for comparison to other ratios.
Literally Buffet did this, are you a better investor?
I increasingly see this style of responding more and more and wow, it's like it's meant to be as triggering as possible. I'd call it fallacymaxxing.
They either get RSI and take over the economy (in which case their shares are almost certainly worth nothing) or they don't and their shares go to ~0.
Their 'traditional business' path makes no sense:
1) they can never stop training or they'll be selling a commoditised (or just worse) product, so it looks very difficult to ever turn any kind of profit.
2) they don't have huge consumer awareness penetration and big business already trusts Microsoft, Google, Amazon way more so in a commodity world the big players win.
If revenue grows faster than training costs they’ll be able to amortize training eventually.
The real money here is in b2b. Consumers don’t pay for digital products. Even the software companies whose products are consumer facing mostly make their money by selling ads which is a b2b business.
How many new billionaires will the country have tomorrow?
Like 1.5T could make as many as 1,500 billionaires. It's certainly much more concentrated than that. But it is still insane to think about.
Does seem like billion is rapidly becoming the new million
In two dialects of English a billion is a thousand million and a million million, worlds apart, and nobody cares because they're both equally inconceivable
In some other languages such as Finnish, biljoona is a million millions, whereas a thousand millions is miljardi. I think German is the same.
This is a constant source of confusion, since because of this, the next big number triljoona is not the same as a trillion.
the comment you're replying to makes the same point. British English officially started using the "American" short-form system in 1974:
https://en.wikipedia.org/wiki/Billion#History
Interesting,I always wondered what the origin of this was.
This will have effects on the economy for the medium and low class, again
And it'll all trickle down and soon the US will have solved poverty, I predict! I've been wrong before though.
> Like 1.5T could make as many as 1,500 billionaires
I highly doubt that, given the founders themselves are "only" going to be worth something like $20 billion at that valuation. The shares have been massively diluted, and the vast majority of the gains are going to Google, Amazon and a large list of VC firms. The people getting rich are mostly already rich.
This will crash the markets. Mark my words.
if you know when you could be the richest person on earth
Anyone know why you would be limited to purchasing two shares of the IPO? (My sister tells me this—she's climbing aboard the ANT train and told me she was limited to two shares. Wut?)
It's oversubbed, people are getting a fraction of the number they requested to buy.
That's probably just for the initial auction at $86/share? Once that's done there will be plenty of liquidity to pick it up at market price, whether that's higher or lower.
The IPO is 10x oversubscribed, which means there is a ton more demand for shares than the number that will be released today. Brokerages allocate shares from their limited pool, and generally prefer their more valued customers.
I asked for thousands, they gave me 143....
are they reminding me of something we all used to use back in the day when msg was 140 char limited?
<3
I think they allocate certain numbers of shares to different brokers. Maybe your sister's broker's allocation was small and split amongst many customers? I applied for $20k worth at IB and got 43 shares worth ~$4k.
I stopped buying IPOs a while back, and have had zero regrets. I don’t think I’ve seen a single one stay above the IPO price during the first year. Nearly all, especially for companies in growth stages, hit a wall where they’re still growing and burning cash… it causes the market to turn on them, and the stock drops to anywhere from 40-90%.
That’s not to say there aren’t a lot of winners even 1-2 years later but I’ll happily be sitting this one out (and wait for a drop or two). The market gives some pretty sweet deals for those who are patient.
Uber stock gained 65% since its IPO. Airbnb stock gained 29% since its IPO. Snowflake is up 16% since its IPO.
That's quite non-impressive when you remind yourself that these IPOs happened a lot of time (6+ years) ago, and that their stocks are highly volatile.
It seems to me the VC system got quite efficient to accurately compute the total addressable market well enough and inject their money to grow until that whole TAM is covered enough.
Then the VCs release the now horizontal stock to the public markets.
That being said, there are also counter examples, like Datadog (up 362%) or Crowdstrike (up 1377%), both in the same IPO class as Airbnb and Uber.
Don’t forget Cloudflare that has increased by 1906% since 2019
Airbnb is flat vs inflation if you look at its first public sale price.
It’s only the IPO price available to institutional investors where it ends up being profitable.
You’re missing my point, Uber is a great example of what I’m trying to say… after IPO it shed 50% of its value within six months and it wasn’t until 18 months later that it was back to IPO pricing and 24 months before it was above it. Buying it at IPO, you’d still be up, but you’d be up a lot more if you just waited six months or a year.
The same with Datadog, Robinhood, Facebook, Coinbase, Crowdstrike, and I’m sure I could go on. Even ARM was generally flat for six months after IPO… the only real winner I’ve seen recently was Reddit (which dropped a bit but was briefly flat before going up).
My point is buying the IPO instead of waiting will probably have you overpaying for it in the short term (1-2 years), long term it doesn’t matter (if the stock/company is making money). Most big tech IPOs have a sell-off, sometime within the first year or so, and it has been nothing but beneficial for me to wait.
I absolutely want to buy Anthropic’s stock, I’m just willing to wager I’ll get it at a hefty discount sometime in the next year or so… from what I’ve seen their revenue is nowhere near their valuation and we’ll probably see the market recognize that (even briefly).
You are absolutely right.
IPOs statisticly do mostly bad within first 12 - 18 month of trading; this is mainly due to institutionals & insiders selling their stuff.
kind of a sad state of affairs
whoosh
This could cause an overall correction in the market I think. It's likely people will take money out of NVDA and other AI stocks to buy Anthropic's IPO. This could send NVDA and other AI stocks further downward and maybe even cause more panic selling. So the market caps of those companies decrease. Then that value is lost when Anthropic's IPO share price falls from insider selling, lack of interest, etc. In fact, if OpenAI has an IPO, Anthropic shares could be dumped to buy OpenAI. It could cause an overall decline in the market.
Yeah, I’m honestly more worried about a massive decline at this point than I am major upside. The “AI boom” has been the only thing really propping up the economy and I’m not sure it has enough gas at this point.
Longer term, 3-4 years I think it’ll be fine. I’m definitely bullish on AI in general but I definitely expect a rough 6-18 months to start sometime before 2028 (and it may have already started thanks to Jensen’s shenanigans).
They are only floating about 70b for public trading. This is very small figure compared to nvidia's market cap and really minuscule smaller compared to the overall market size. People are moving money out of nvidia for other reasons, but it's most certainly not because of this ipo.
Yeah.. the models themselves might not seem to have much moat but the teams that already set their workflows with claude don't easily switch. So I'm thinking the rate at which users maintain their subscriptions and developer habits are the real moat
I am worried, their business is more like webhosting and less like something that is unique. I rely extremely in AI. But there is Anthropic, Mistral, Claude, Deepseek...
It would not really be a problem if Anthropic is gone tomorrow.
What? Claude is Anthropic's.
Sorry. Meant GPT.
If I wasn't personally convinced that they're deliberately nerfing their models, I'd buy the heck out of this. But I believe they are. And I also believe they are plateau-ing and do not yet know how to seriously advance the abilities of their models.
As a result, I think this stock (which has no voting right btw) is a major gamble and unlikely to pay off before they run out of money.
More importantly, I think Dario is dishonest, and trying to thwart open models which are soon going to be just as good as his models, but without the spying on what you're doing.
how can they both be deliberately nerfing and also don't know how to advance their models? unless I'm misunderstanding
> I think Dario is dishonest, and trying to thwart open models
agreed there
I guess even without knowing how to tune my car to make it faster, I can still unplug one of the spark plugs to make it go slower
> how can they both be deliberately nerfing and also don't know how to advance their models
It's a fair question! Let me elaborate:
I think the best internal model development is basically plateauing and they don't know how to make large improvements now. Pareto has past, but it's worse. The next 20% will take more than 80% of the energy they've put in, and this IPO is an attempt to try and get there. I don't think they have good chances.
In parallel, the models they are shipping to _most_ customers are being nerfed for one or more of a long list of reasons you can read in my posts elsewhere.
I actually think the models are mostly great! But they are trying to save that tiny bit of extra for themselves since they think it's their only way to get ahead compared to others competing in pure software markets against them, which they want to own. They don't want to just sell you pure tokens, they want to build companies.
> ...being nerfed for one or more of a long list of reasons you can read in my posts elsewhere
not interested in reading your posts elsewhere, and this is the critical part of the argument
Before releasing a new “version” of their model they start serving a quantised version of the existing model which performs much worse. Then they release the “new model” which is just the old model in its original state.
maybe, but that would also help their competitors' benchmarks
I have the same feeling, the original Fable and current Fable are same thing to me, 6 months later, with some egregious nerfing in-between.
What market cap is this?
1.5 trillion
Which is some $300 billion below what they were floating early this year.
How are investors not terrified that in what is supposedly the biggest boom ever, the leading company is downsizing its estimate of what it is worth?
Because investors no longer use common sense
Isn't it below 300 billion becuase they are terrified? Or at least not as exuberant?
so many reasons not to take this at face value - greenshoe, lockup periods, 5% float. consider the latter alone - in this math, the 95% unsold are considered to have the same value as the rarified 5%. that defies economics.
About 1.1T market cap will exist tomorrow prior to opening bell, out of a potential 1.5T that includes .4T stuck in options, convertible notes and such.
But it’ll likely open above 83 when trading actually begins. We may see upwards of 2T if it goes over 100 a share
Where did you get these numbers?
China will eat this, they are too expensive.
They already have all the funds they need to finance their investment. This IPO feels like it's exclusively made for the existing stakeholders to cash out and not to raise money for future projects, which is usually the main aim of a public offering. Furthermore they have a great technology, but not great products and not a mature distribution system. As commodification continues, players like Google who can embed the technology in their products will gain ground.
Retail investors will be left holding the bag. This is the ultimate cuck tranche.
Here's something I don't understand that I don't see people talking much about:
Anthropic has raised about $141B.
But their S-1 also says they have $132B in cash on the books. So they've basically not spent hardly any of what they've raised.
Yes, they have huge spending obligations over the coming decade, but I don't understand how this fits the narrative that they're massively unprofitable and just shoveling investor cash into a furnace to stay afloat.
Regardless, they will soon endure crippling legislation and litigation. They will be forced to stall their ambitions and the timing for it all to work will collapse.
Lots of opportunity to make money with this until then, but it’s coming.
I would not hold my breath. Politicians have a history of coddling corporations at the expense of everyone else.
No doubt but many of the people being disrupted here are the voting bloc that these politicians need not only as voters but as donors.
As for the lawyers there’s essentially an endless tap of litigation to try here. AI harms people the more and more it’s made ubiquitous. Someone has to be liable.
Would all this AI wealth completely crush ppl like me on the sidelines?
It's like in a gold rush. But not everyone will get gold.
I think in a few years many of these AI companies will no longer exist; and the survivors will be bigger than they are as of right now.
Fanciful.
There's no voting rights, and no foreseeable profit sharing. As with most tech stocks, you're speculating on Pokémon cards with more liquidity.
It is not quite as bad as Meta. Meta concentrates shareholder and board-election control with one guy (Zuckerberg), whereas Anthropic divides power between its founders and a trust intended to protect its long-term AI mission.
Neither structure gives ordinary public shareholders control proportional to the money they invest. This should probably be made illegal.
> Neither structure gives ordinary public shareholders control proportional to the money they invest. This should probably be made illegal.
I would prefer it if people were still allowed to invest in such companies in the public markets, but perhaps we can make it so that inclusion into an index requires "1 share = 1 vote". That way, at least retirement funds are protected.
I would prefer the exact opposite: non-voting shares in indexes protected only by law. I don't want my index manager to act on my behalf. That's the exact opposite of why I buy index funds.
Meta is up 1,522% since IPO. The reality is founders care way more about the long-term health of the company than the typical outside board member that is just farming prestige and owns minimal equity. Do you really think the former president of Taco Bell International knows more about how to run Nvidia than Jensen Huang?
Meta would be even higher had Zuckerberg not wasted time, money and talent on that Metaverse nonsense.
People said the same thing about Instagram when he acquired it. I’m not so sure you can have one without the other.
Yes he probably does because he knows not to commit treason and fuck spies. When’s the last time you heard the president of Taco Bell being black bagged at the airport? When’s the last time you heard of them getting fired for spying?
If there were an embargo, maybe Taco Boss would smuggle US prime beef into Mexico, who knows.
For a moment I thought you're talking about the crypto market. :)
They used their most advanced models and it replied "in 2026, the best way to make money is a good old fashioned scam!"
At this point I am 99% convinced that all major providers have agi equivalent performance when running as a 'swarm'. Simultaneously they all hamstring their AI to try to prevent having their own AI being used to build someone elses AI.
Codex is completely unusable at all for anything but small tasks when handheld by a person. Orchestration is simply unusable. Anthropic falls apart and refuses a critical set of things that are small enough to block you from getting it to build out real work. Like basic ci/cd server setups and using login creds provided on web, etc. Conveniently setup in a way that doesn't at all prevent a hacker group from at scale pwning people/corps but does prevent normal workflows. Want it to build a full stealth toolset for control of android phone over adb bridge? No problem. Now can you use it for anything? No I won't log in for you. Pressing the fucking login button is just something I'm morally against. etc. Basic room temp IQ computer tasks. Like they are trying to prevent someone from automating your average corporate workflow while it can do everything else trivially. Why? Because 80% of their revenue is from Corps letting employees burn $$$ tokens for every day tasks like proofreading an email.
I know this for a fact because of custom harness work in a corp with Microsoft EntraID SSO over a year ago where all of these tasks were easily performed by the old Sonnet model and I trivially automated every workflow.
And, strangely, when I do something that should make everything much worse, simply just tell Claude to use Codex as sub agents, performance gets drastically better, presumably because teams from both corps didn't have the chance to specifically fine tune their models to be hamstrung when run in this configuration.
It is obviously not about safety and has never been. These models still will easily find vulnerabilities. Just get some code you want to exploit and throw $$$$ inference at it to 'find all bugs confirmed via test case suite' etc. Then throw inference at targeted exploit scripts.
I just hope this doesn't end in my retirement fund
> I just hope this doesn't end in my retirement fund
Don’t worry. It will.
So far the pension funds have been major investors in everything AI, every fund is in it hard and deep now. It absolutely will be in your retirement fund, but don't worry your unlikely to be able to get to retirement unless its really soon.
I'd love to know how exposed we are in Europe to a potential crash.
I'm getting 2008 vibes.
The blast radius is quite large, but appears less contagious than what happened in 2008.
Lots of folks that bet big on AI are going to lose their shirts, but this doesn’t look like the “banks are failing every weekend and the global financial system is on the verge of collapse” situation we had in 2008.
No? Look at the underwriters of Anthropic shares. Mostly banks
It’s not really a liability for them. Certainly not an existential one.
I'm getting 2000 vibes and not even joking. It's almost the same. I'm pulling out of any and all market investments except maybe BRK.B
Technically it’s not as bad. We don’t have the same AI industry so fewer companies to go bang, and traditional “legacy” companies will be havens for investors.
However… when America sneezes, europe catches a cold.
The biggest problem is that poor economic management (in the UK, austerity under Osborne and the Covid response) has eaten our margin. We are not starting from a position of strength.
Our VC culture is much more conservative, which is good and bad. It's partly why we lag so far behind in new technology, but it partly insulates us against high leverage default cascades. However the greatest risk is macro economic. Europe sells a LOT of products and services to the US. When they're hit by economic downturns, that affects us to a large degree. So the right question is: how exposed to a crash is the US economy?
Unlike 2008, the vast majority of capex has been cash, not debt. So this looks nothing like 2008. It also looks nothing like 2000. Growth is extremely high and real and organic. Real customers are paying for real products and services at an unprecedented rate. The risk will come from somewhere else entirely. Possibly something like a major technological advancement in AI which renders most data centers obsolete or unneeded. Or makes intelligence cheap and abundant for everyone, rendering Anthropic and OpenAI worthless. If that occurs it would cause a contraction in investment in the space, but the US economy remains fundamentally strong.
To be honest, I don't see serious crash risks right now related to AI. The companies taking the biggest risks have enormous revenue, profit, and cash reserves. They won't fail. I think the largest risk is actually inflation. Many governments are going to have to increase rates over time, and this has a mediating effect on economic activity. It slows. Not just AI, but all sectors. That could cause a recession.
Models are popping from left and right. Is that couple months of lead trillion dollars more valuable?
Are you willing to bet they can't stay ahead?
For actual science done by scientists they're far behind open models now.
The people I know using this model are perpetually hamstrung by slowness, pedantry, misplaced assumptions and appeals to authority. You can't move forward if you're just trying to norm your training data.
They've created this extremely slow, overwrought system with lots of guardrails and token limits designed only to create addiction. Why would you use that when you can have a marginally less "smart" system that's 50x faster and 100x cheaper? Who cares if the model one shots something. It needs to be checked anyway, by test in a machine or a lab.
What models are scientists using that they gain so much more marginal value from than Claude-class models?
Who are the pedantic, authority-appealing users who are using this model?
What makes you believe they’re norming their training data to that group?
Cloud models have zero value if the findings of scientist are absorbed by these same cloud models and "by coincidence" then published before them. It is impossible to get ZDR as individual that you can trust.
Statistically how many companies that were first to any market the eventual dominating leader of that industry? To me it would still be like picking a single number off a roulette wheel because it is a very slightly larger slot to bet on. Maybe it will pay off, it might even be statistically the best choice, but it is still a slim chance overall that it will be the winner.
I am willing to bet 1:1 that a model from any other lab will beat Fable 5.2 within 3 months in any metric of your choice.
I am also willing to bet for their next frontier model that a model will beat it within 3 months of its release.
They sell trust, and they sell "the best one".
The specific models and capabilities are a sideshow.
And hence, the money flows all to the top despite viable alternatives...
(a separate market of workloads is contracted out to Fireworks, Baseten et al. and a third segment DIY it with rented or owned GPUs. But those customers are different buyer archetypes little overlap with those buying trust.)
Why trust them though? They're a cult.
Well then they have failed miserably. They cannot be trusted, with all that anti-user behavior, fearmongering and all that stuff.
Models are useless if you don't have the infrastructure deals to run them at scale. I know people that use cheap Chinese models and get good results from them. But they are paying a lot for self hosting those models. The models are cheap but the infrastructure isn't.
The reason I think Anthropic's position isn't that strong is that most of the money they raised will go into making other companies very rich. Because they don't actually own a lot of the infrastructure they need. OpenAI has similar issues. The companies that actually provide the data centers (i.e. hyperscalers), chips, gas turbines, etc. Will be on the receiving end. And of course some of those hyperscalers have their own ambitions on tapping into the same markets. The SpaceX pivot to data centers in recent years is particularly interesting. Anthropic is already paying them a lot.
The reason I'm fairly optimistic overall about the economy is all the spending that will result from these investments in all that physical stuff represents an enormous cash injection in the economy. The investment bubble may or may not burst but it won't wipe out the infrastructure. If Anthropic collapses, that infrastructure will just be gobbled up by whomever takes over.
I wonder if China will get their head out of their asses and start manufacturing chips to construct electronics and infrastructure to run AI.
Oh, you mean they have already started on it??
They also has the most energy production. They will go past anyone else at this rate.
I am just waiting for the time when all of us (already or soon) out-of-a-job folks will start reaping the "AI abundance" (that "AI leaders" can't get tired talking about, including our own Mr Huang) while some folks already make billions and millions in these trillion dollar IPOs (soon more to come).
Because if that doesn't happen then I have a feeling the world will proceed something like this: AIs paying for things to other AIs (as a circular gesture; possibly different from the circular finance of today) while some rich folks will sit in their safe island citadels or in NZ (etc) nuclear bunkers controlling the arms, ammunition, surveillance, medicine, finances, along with pretty much everything else of the world while the rest of humanity, the majority – who couldn't tap into that abundance, will finish each other off one by one – or maybe worse, they will be live to witness it all. I try really hard not to picture this post apocalyptic scenario but either all this IPO, moat, and whatnot are hogwash, or that is happening :)
When/where can I start buying put options on this, about 1-3 years out, and out of the money? There's no way this company is sustainable.
You are their exit liquidity
Google IPO back in 2004: you look around, everyone was using Google.
Netflix back in 2010: you look around, everyone was using it.
Anthropic today: look around..
Got a few shares so I can help with the great redistribution of wealth once all these people start cashing out and tanking it. Have fun!
Here's how I find out the Nasdaq doesn't require 4 character symbols.
This reads more like exit liquidity than an IPO.
Companies used to IPO when small and needing money for growth.
Will this have SPCX price action? Or will it just drill to the center of the Earth like CBRS?
$83 but open at $150 Im sure
Saving this for when trading opens
I saved it too. They were off by a bit…
Very possible we see both $150 and $50 tomorrow.
The extinction train continues to gain momentum.
I just hope the very blatant Anthropic astroturfing subsides after the IPO. It's getting tiresome.
It's a joke? No, I checked the calendar and it's not April the 1st.
Some people will be left without a part of their hard earned money.
If you’re retail good luck…
:DDD These agonizing you should not buy wailings every time a fucking IPO is on the horizon.
My take on this is that Anthropic must be closed until they've sorted out their security problems. They and OpenAI have claimed repeatedly that AI poses catastrophic risks and that it's not safe.
Sorry but if a company insists over months that their products are unsafe and even hint they could lead to the destruction of humanity, then I don't understand why authorities don't act and stop the company until they can demonstrate their products are safe.
and its not clear if security will ever be 'solved'
Lol lmao even
I'll be reinvesting some of what I pulled out of NVDA but going wait a good 30 mins after opening bell. Anyone else doing the same?
What happened to Ed Zitron’s predictions
All the questions he asks are still relevant, and we still don't have answers. This is a gamble and he might be right about many things in the end.
really? all?