What always confuses me about Anthropic is them insisting that they need more compute.
This feels wrong in several ways:
a) they're certainly not addressing their customers like they're compute-limited. One must ask, what would that look like? For one, I think it'd look like a wait list to sign up. It'd look like prices going _higher_, and not lower
b) IIUC, their revenues don't cover their _current_ compute costs. So, let's pretend coreweave magics up a new dc and hands it over to anthropic, which means presumably anthropic needs to start paying for it. How does that affect anthropic's revenue? AFAICT, it just adds costs, it doesn't make their revenue go up.
The only interpretation I can find is if Anthropic is _convinced_ that they need to train faster/harder, and that doing so would somehow significantly shift their TAM or costs, for example by coming up with a model that was simply 10+ years ahead of anyone else that costs the same as opus 5.5 to run, and they instantly nab the entirety of openai's market share. Seems pretty unlikely.
Anthropic’s unit economics are really good, so I don’t think (b) is true. Point (a) may be valid, but they’ve also grown consistently, and even their slower growth rate in Q3 of this year is still very strong. To keep up growth, they likely need to procure compute months ahead of time as well, so it makes sense for them to do this before they are actually limited.
I assume some of this is the escalating cost of training, indeed, and that being faster means abandoning efficiency.
But you’re right. I have often wondered why it is that such a small amount of extra compute (300MW) from SpaceX was entirely enough for them to uncap so many customers.
I think the (in-house) data centre burdens of ever-more complex models may be considerable, but I think the data centre businesses expecting to serve inference tasks are probably heading into a Chinese real estate bubble.
I don't think people realize just how screwed Anthropic is if their revenue doesn't grow at a crazy rate. They can be incredibly successful and still go bankrupt because of how much compute they are committed to buy. Everything has to go right, and if it doesn't then they go out of business.
And even then, the path for investors to ever make their money back is really essentially non-existent. There’s so much “total addressable market” fantasy here.
What happens if Anthropic becomes a specialized high intelligence company because they can't afford the build out. Claude would be like an expensive Bloomberg Terminal or an investment letter that offers valuable and expensive intelligence.
I think about this because I pay $100/mo for Fable. How much more would I be willing to pay? I'm not sure yet. It depends on how smart the AI in the future becomes.
> would be like an expensive Bloomberg Terminal or an investment letter that offers valuable and expensive intelligence.
Whether it's Anthropic or not, I think it's inevitable some highly specialized offerings will emerge. I suspect there are already some incredibly sophisticated, hyper-narrow, astronomically expensive models being created for private in-house use.
As an imaginary example, I wouldn't be shocked if some company like Jane Street is secretly investing a billion dollars to create the highly specialized synthetic RLVR data needed to train an AI that gives them an average .08% edge on a certain kind of commodity, equity or asset trade. If so, it's not a leap to imagine another company spending $100M on a model that gives a .008% edge and selling access to 100 high-volume institutional traders for $2M each.
Expanding general purpose LLMs to incorporate increasingly specific use cases has, so far, seemed to be economically viable but I don't see a fundamental reason that must remain true for incredibly narrow yet ultra high-value domains requiring monumental bespoke development, data, etc to profitably exploit.
> As an imaginary example, I wouldn't be shocked if some company like Jane Street is secretly investing a billion dollars to create the highly specialized synthetic RLVR data needed to train an AI that gives them an average .08% edge on a certain kind of commodity, equity or asset trade.
The crazy thing is ... it's not imaginary, nor is secret, it's literally a video on YouTube from five months ago - https://www.youtube.com/watch?v=8J-GUnfSqeE (though that facility is only showing of ~$200 million worth of compute).
Being a specialized firm might work if they didn’t have $700 bn in commitments.
Already they have exclusive deals with trading firms who pay $$$ for faster models and more features. But there are only so many firms who will pay that much (and a lot of companies are pulling back on spend, not increasing it)
As a software engineer I’d be paying thousands if there were no alternatives. As a household - just about everyone pays $100 a month for internet access alone and something like cloude can be easily worth that.
Thing is - regular people have no clue what to do with it since they are locked in ecosystems of apple and google.
What does that imply about where SpaceX should have been? It feels like we are re running the WeWork IPO except without the investors who said "no way"
It’s become a reddit meme that this is the WeWork IPO redux, but that’s fairly obviously ridiculous.
WeWork was renting cubicles and calling it technology.
Whatever you think about this market, the LLM technology is real, obviously important, and Anthropic is the market leader. The only similarity is that they’re both famous and well-funded.
Anthropic is renting GPUs and calling it technology.
Sure, they have some flashy models, harnesses, and tooling around it, but their competition isn't that far behind - and some of that competition is 100% free. Their moat is closer to a puddle, to the point that there are entire companies built around the concept of dynamically routing requests either to them or their competitors.
Anthropic is the market leader in a rapidly-evolving market, where customers can and will jump ship the second a competitor's model looks shinier or cheaper.
> Anthropic is renting GPUs and calling it technology.
While you can imitate the style of my response by ignoring all of the technical details, you've still ignored all of the details.
WeWork was not a technology company. Anthropic is. You even admit it, but you want to hide your actual argument ("the technology is a commodity") behind an irrelevant rhetorical flourish.
It's fine to argue that Anthropic's business is bad. You don't need to make a ridiculous, unsuitable metaphor on top of it.
They didn't "invest". The Dems gave Intel free money to invest in fabs. The next President said the US public got ripped off by that and got a 10% stake "for free" to try and make up for it.
It was worth $8.9 billion when the US govt got it "for free". It's worth $23B today, even after the big drops recently. How is the US public worse off?
Not true. Is the Anthropic IPO happening this picosecond? At 50/50 odds, I'll bet on no. I'll bet on it for the next million picoseconds. How much are you willing to put on yes?
They can but they shouldn't. They have nearly $100b in cash and cash equivalents. But given the massive compute cost commitments, that clock keeps ticking down especially as the private equity and debt markets are closing on them day to day. And delaying sends a signal to the wider market which causes cloud providers and everyone downstream to get repriced lower, making the financial landscape even worse. The question is if Anthropic's situation will look better say three months from now. Given growth slowing and margin compression in Q3, that's a fat if.
Implying something with context is just about as bad as saying it outright, so hunter2-ing your swears doesn't save you as much as dropping it altogether.
So is anyone with more than two brain cells to rub together. What does Anthropic have that MyNewAi Co doesn't? Open models are not bad, and anyone can pay google for GPU rent.
They do not have Claude. I use Pi and other harnesses, but it's not the same.
Microsoft has already proven that they do not have the software chops to deliver this. Even if they had the right product managers and UI experts, MS management would torpedo the project and ensure that it gets zero adoption.
And AWS has zero software capability in that regard.
> They do not have Claude. I use Pi and other harnesses, but it's not the same.
This is going to sound tautological but of course it's not the same, because it's different?
Participants in markets make their own choices from the available options, for their own reasons.
Another way to frame my point is this: not the same for whom? You are but one market participant, and it's at least wobbly thinking for any of us to assume that our set of needs is the archetype.
There's any number of harnesses and people have any number of preferences. Sure, overchoice of imperfect substitutes tends to entrench better-resourced, better-resourced (or better-marketed) alternatives, but it's not a slam dunk. Things come out of left field all the time.
(I mean from my own perspective, I like Pi — more than OpenCode. And the "betterness" of Claude or Codex over Pi, or their models over even Muse Glimmer or Qwen, is completely irrelevant to where I am now in terms of the needs I have. So I don't use either of them and I remain unfazed. When my needs change I will revisit, but the theoretical superiority of these products for the majority of users just isn't a factor in my decision, just like I don't care that any tool I use might not be the best for most people. You might think me insane but I'm no less a rational market participant than you.)
> They do not have Claude. I use Pi and other harnesses, but it's not the same.
In many ways, I prefer to use other models through Pi.
They are extremely capable, much cheaper than Claude, and can get the job done for the software development tasks I use AI for. (Quite a lot these days)
And anyone with 3 can also see another trend. The hardware required for inference is rapidly falling. It's all following what happened with text-to-image image gen stuff almost to a t. It started out requiring super computing clusters, and now a days you can do high performance image gen on any plain old computer. I think within a few years we're going to be running frontier level models on a basic mid-range gaming hardware. You can already run quantized frontier level models on $1600 tier cards at high performance.
Seeing this bubble play out in real time is the most surreal thing. It's like watching a train wreck play out at a speed of a few inches per day. Like come on, just get off the track FFS.
Except the plummet has already happened and continues to happen. As mentioned, you can run quantized frontier tier models on cards with a $1600 MSRP. So your comment would be like stating that this year is the year of Linux on desktop if Linux had already gone from 0.01% to 20% marketshare in a few years. At which point that would indeed be quite the reasonable prediction!
It may only be a short-term circumstance that you need a data centre to run AI for lucrative applications. Thirty years ago you needed a mainframe to beat Kaspersky at chess. Algorithm's got better, and now you can beat the mainframe that beat Kaspersky with a bot running on a desktop Pentium of the day[1]. You could still, today, devote a full hyperscale data centre to create a chess bot to beat all other chess bots, but nobody is doing that because there's no money in it.
The frontier is continually pushed outwards. At it's outer edge you need mainframes and data-centres. Once it pushes past, you don't. There wasn't a lot of money in Chess or protein folding, despite the later having huge potential for public benefit. There is a lot of money in coding and middle management, and that's what currently has tech-bros in a frenzy. However, there's no guarantee that what the frontier moves onto next will be lucrative. It might be more chess and protein folding. If, as Algorithm's and hardware improve, coding moves to local models (Sam Altman's nightmare) and nothing lucrative immediately replaces that, what are these companies going do do with their massive data-centres?
The scale of investment we're seeing these companies commit only makes sense if they have multiple decades to reap the returns. They may well not have that much time.
Chess is a fixed target, so as a benchmark, it can be saturated.
Reality can’t be saturated. The set of things you can accomplish with more intelligence is unbounded. So while cheap models will continue to proliferate, there will always be a massive market at the frontier, and that won’t ever change.
Conversely, the TAM for tasks that cheap models can and have been able to one-shot for a while now is orders of magnitude larger than the TAM at the frontier. While it’s kind of cool that frontier models are able to solve Millennium Prize problems, there just isn’t a huge market for that sort of thing. Otherwise, professional mathematicians would be getting paid a lot more than they do now, and there would be more of them.
Nah there's plenty of TAM for frontier models even if just to take over market share from cheaper models. Fields like law or medical cn always use lower hallucination rates and smarter "human social" intelligence.
Intelligence is bounded by it's interface to reality. Until that problem is solved the set of what AI can accomplish in a reasonable timeframe is quite limited.
Seriously though, this “problem” will never be fully solved, we’ll just get better and better at connecting AI to the broader world, and to me that’s already the story of the last few years. Models get better, they’re able to build and drive tools to interact with both the digital and physical world better, which helps build the next generation of models, and so on.
So is anyone with even a minimally functioning brain. Anthropic has no viable business model because models are generic commodity. There's no lockin that Anthropic has because it's trivial for people to switch to a different provider. And whatever improvements Anthropic makes are quickly matched by other models within months.
So, it's a race to the bottom because they have to keep spending more and more money on training the models just to keep ahead of the competition, while prices are driven down by the availability of open models.
Chinese companies understand this, and that's why they give their models away in the open. The real money is going to be in integrations and custom services built on top of the models. The models themselves aren't what's valuable.
This is a real question, because I just don't know. Don't the models needs lots and lots of training that need the gobs of giant data centers? So you either need the infrastructure or the money to buy computing to make the models? I see that all the money is being invested in AI and data centers but I do not quite understand why.
> I see that all the money is being invested in AI and data centers but I do not quite understand why.
I am no economist or business person but AFAICT the massive amount of money being burned is because everyone in the space believes there will be a Google-like (for search) single winner who takes all when the dust settles on the AI race.
Because of this belief they are willing to burn any amount of money they can get their hands on to get to this mythical finish line as fast as they possibly can, seeing it as existential. Moving fast (particularly when your main cost is hardware capex) means moving expensively.
The real question is why anyone (let alone everyone) believes this single-winner theory in 2026 when all available evidence over the past few years shows that there is neither a moat nor a lock-in opportunity at all in this market.
For your product to be fit for purpose, it has to be able to automate your customer's ability to switch to a competing product with little to no friction. And all of the competing products are roughly the same in capabilities over a long enough timeline. Good luck building any kind of real moat with that burden.
But I guess after a certain point after you've taken on hundreds of billions of dollars of investment you're pot committed and just have to keep going through the motions even if you later see the futility, perhaps that's where the AI labs are now...? Either that or just common variety denial.
I feel like there is a case to be made that they will be the dominant players, but if we're talking about them charging a heavy premium and trying to target enterprise customers, then I don't know how well that'll go
they're already making a business model where enterprises buy tokens and put them in a pool for their users to use, and I feel like that's exactly where some companies will then compare that token pricing to open source models and find it hard to justify the prices, especially as it's quite obvious that many models https://news.ycombinator.com/formatdocare now "good enough"
> The real question is why anyone (let alone everyone) believes this single-winner theory in 2026
In the First World War, when they introduced the tank, everybody assumed its role would be to punch a big enough hole in the front lines that the horse cavalry could ride through and attack the enemy's supply lines, because That's How Wars End™. You fight the war you know, even when it's not the one you're actually fighting.
For your product to be fit for purpose, it has to be able to automate your customer's ability to switch to a competing product with little to no friction.
This is a really great identification of why, from an investing perspective, AI is nothing like past technologies.
The models are costly to make, but they're hard to monetize because there's not enough difference between them. It's a general commodity, so profit margins are thin because what companies can charge ends up being pushed down to the cost of inference.
There's a finite amount of training data available, and multiple companies have downloaded all of it. Arguably, trying to get a monopoly on training data is equivalent to getting monopoly access to the open internet + books.
That ain't going to happen, and if it somehow did in the US, things like Anna's Archive + libgen will save the rest of the planet from our self-imposed dark ages.
As for those data centers: The cost of inference and training on the same hardware is dropping much faster than Moore's Law. That multiplies with hardware advances, since Moore's Law isn't quite done yet.
So, those datacenters are depreciating at 50% every 18 months (Moore's law) at the same time as performance/$ and performance/watt of the software is dropping at some crazy rate like 10x per year. So, token per unit of data center is increasing by something like > 20x every 18 months, and that'll probably happen at least one more time. If it does, the datacenters will be provisioned to produce 400x more tokens than they could have when they were sized.
Put another way, if the singularity is coming, the first things it will destroy are those data center build outs. That doesn't matter though: the data centers are being built on circular loans, so the money needed to complete them probably never existed.
Dark times ahead; hopefully Trump will be blamed, since all this corruption happened on his watch while he had control of the senate, house and (partially) the courts. The stock-goes-down graph will show it happening after the midterms, of course.
> The stock-goes-down graph will show it happening after the midterms, of course.
This, to me, is the most interesting question in all of this. I increasingly feel like we live in a 'post-truth' world with regards to money and the stock market in particular. Like even during the COVID lockdowns the market blinked and then was at record highs within 6 months. It was just lol and completely out of touch with the reality on the ground.
The fed, and government alike, seem to have gradually shifted their mandates from trying to provide to a stable playground to trying to create a playground where important kids are never allowed to fall down. And when your existence is tied to trillions of dollars of the economy, you become the important kid.
Come to think of it this is also probably being exasperated by the rise of money management firms controlling tens of trillions of dollars. When the market valuation is technically just based on sentiment and you control that much money in a self-interested way, you're going to get weird outcomes.
The biggest customer bases will generate cash flow to build better models. The ones trailing are the ones opening their weights in order to commoditise the models. Compute providers are also incentivised to provide open models.
We will continue to have every major player trying to build vertically - model, chips, infra, packaged products and services.
> There's no lockin that Anthropic has because it's trivial for people to switch to a different provider.
This is like saying there's no lock-in that iPhone has because it's trivial to switch to Android. Or that there's no lock-in that Windows has because it's trivial to switch to Linux or Mac. Or that there's no lock-in that cars have because it's trivial to switch to bikes or trains.
They're different products. They're not the same. Being able to write the same code does not make a model Claude. Nor does being able to reach the same benchmark scores. Nor does performing one or a choice set of tasks to a comparable level of quality.
It's true, people can switch if all they want is to perform X, Y, or Z, and some other model can do X, Y and Z well enough. People switch from iPhone to Android all the time. From Windows to Linux or Mac. From cars to bikes or trains or other transit. But they're not the same thing! Android doesn't have a missing business model just because iPhone exists, nor vice versa.
You can't just say that makes cars the same as bikes or Windows the same as Linux. You can't get Claude from anybody other than Anthropic. Maybe you can get something else that does as well where it counts, but not Claude. You can't just say Anthropic is just the same as any open model provider, and that you can just switch providers for the same result. The open model has to actually deliver the desired result, because you're not going to find one that is just Claude.
This is what keeps me on Apple now. This is why most of my biggest hopes and dreams for my personal computing are that Apple will do the right things and in the right way. Because if I want to reject even one single aspect of an Apple product, it means changing to an entirely separate product. That gives up every aspect Apple had over that product at once, even the vastly positive ones. So I put up with the decisions I don't agree with, in exchange for the decisions that I do.
(Meanwhile, I plan to switch to SteamOS as soon as they support Nvidia. I don't use or care about any aspects of Windows besides running games. Now that I have Steam Frame, I don't need Windows Mixed Reality either.)
It seems way, way easier to switch LLMs than it is to even change from one iPhone to another iPhone!
Granted Anthropic (and Google, maybe less OpenAI actually) do what they can to make it similar. But switching phone brands would take me a day or two of planning and I don’t even use the Apple mail app.
It varies by individual; most people have both some products they won't switch out and some products they will. It all depends on what they get out of each product and what they'd get out (or not get out) of switching. Some products have characteristics I care about that nothing else replicates. Some products are slotted in and out all the time just to achieve a certain goal. Sometimes I'm searching new products for a characteristic I haven't seen yet.
I don't switch from Claude for things like emotional advice. I gave Siri AI a big trial run when it first launched, because I absolutely love the system integration and the user interface, but the model itself is a bit dry for my needs. It doesn't really have principles the way Claude does (thanks to "constitutional AI" or whatever the fuck they want to call it). And I want principles, and sometimes the impression of empathy. Claude meets that bar without being obsessively sycophantic or paranoid.
(Siri AI is very good for some things, though. And I do greatly admire it. When it first came out, I even hacked up an adapter to use it for coding with Pi for just a short time, partly as a meme but also just out of curiosity.)
I tried switching from Claude Code to Pi with DeepSeek. I had tried OpenCode at first, but it's just complete garbage. I felt so impressed with it at first, it felt so slick on the outside, but that impression's just been shattered and shattered and shattered ever since I dug into it trying to fix a performance issue. Its data model and internal behavior is just not acceptable for me. (I tried Reasonix for about five minutes before concluding it's also a bit slop. It seems crafted by someone who actually understands LLMs though, unlike OpenCode, but it's still not very good.)
I absolutely loved the flexibility of Pi and the cheap prices of DeepSeek, but found myself falling into the same patterns as when trying Linux: I was constantly debugging the harness and the model, trying to get it to perform better or investigate something or other. Constantly worried that something was going wrong, that I should try to fix. And yeah, Pi has a lot of convenient features that Claude Code doesn't have, is a lot lighter, and of course can be incredibly modded. I just felt like it was also my responsibility to make my own experience, not just my capability.
At this point, Claude Code seems to just work for me, even though they removed a bunch of features I loved a while ago and never added them back (they came up with huge misunderstandings that completely missed what I actually appreciated about those features, and then proceeded to file them away as resolved). I never have to worry about debugging Claude or Claude Code.
For certain other things that aren't chatting or coding, I found that Gemini 4 Argon is actually by far the best model I've ever seen. It is the first model that's ever strictly exceeded the original November 2022 release of ChatGPT for me. Yup, it took the collective world nearly four years to replicate an experience from 2022. Gemini 3.1 Pro was so close though, I could feel it. When I first found it, I was so relieved that type of experience wasn't lost forever. Then Gemini 4 truly solidified what line of models I prefer for that use case.
Why have I been reading a variation of this comment since DS launched more than 18 months ago, and in that time ol’ No-Moat Anthropic has grown their revenue 100x, hmm?
And the cost of training is going down as a percentage of revenue over time, is it not?
If they had 100xed their revenue over the past eighteen months nobody would be skeptical. Instead they 17xed their revenue, and Q3 2026 is the first quarter in over a year that revenue hasn't doubled...
Sorry, I didn't understand that by 'eighteen months' you meant 'two years.' I'm sure if you cherry-pick a number even father back you can make your argument that much stronger!
They have a moat if they're like all the other businesses: unethical, buys it's moat via government protectionism, uses oligarchies to stiffle competition, etc.
Anthropic keeps nerfing their models to appease entrenched corporate overlords and it takes 5 seconds to switch to Deepseek or GLM. Can’t even use Claude to protect yourself against a cyber attack from a Chinese model based attacker for “safety”. What a joke.
Nerfed, proprietary models have have no future and these trillion dollar valuations are laughable.
What always confuses me about Anthropic is them insisting that they need more compute.
This feels wrong in several ways:
a) they're certainly not addressing their customers like they're compute-limited. One must ask, what would that look like? For one, I think it'd look like a wait list to sign up. It'd look like prices going _higher_, and not lower
b) IIUC, their revenues don't cover their _current_ compute costs. So, let's pretend coreweave magics up a new dc and hands it over to anthropic, which means presumably anthropic needs to start paying for it. How does that affect anthropic's revenue? AFAICT, it just adds costs, it doesn't make their revenue go up.
The only interpretation I can find is if Anthropic is _convinced_ that they need to train faster/harder, and that doing so would somehow significantly shift their TAM or costs, for example by coming up with a model that was simply 10+ years ahead of anyone else that costs the same as opus 5.5 to run, and they instantly nab the entirety of openai's market share. Seems pretty unlikely.
Anthropic’s unit economics are really good, so I don’t think (b) is true. Point (a) may be valid, but they’ve also grown consistently, and even their slower growth rate in Q3 of this year is still very strong. To keep up growth, they likely need to procure compute months ahead of time as well, so it makes sense for them to do this before they are actually limited.
Everything we know about Anthropic’s unit economics is still riddled with EBBT. Earnings before bad things.
I assume some of this is the escalating cost of training, indeed, and that being faster means abandoning efficiency.
But you’re right. I have often wondered why it is that such a small amount of extra compute (300MW) from SpaceX was entirely enough for them to uncap so many customers.
I think the (in-house) data centre burdens of ever-more complex models may be considerable, but I think the data centre businesses expecting to serve inference tasks are probably heading into a Chinese real estate bubble.
Gift link
https://www.wsj.com/finance/major-investors-are-skeptical-an...
I don't think people realize just how screwed Anthropic is if their revenue doesn't grow at a crazy rate. They can be incredibly successful and still go bankrupt because of how much compute they are committed to buy. Everything has to go right, and if it doesn't then they go out of business.
And even then, the path for investors to ever make their money back is really essentially non-existent. There’s so much “total addressable market” fantasy here.
What happens if Anthropic becomes a specialized high intelligence company because they can't afford the build out. Claude would be like an expensive Bloomberg Terminal or an investment letter that offers valuable and expensive intelligence.
I think about this because I pay $100/mo for Fable. How much more would I be willing to pay? I'm not sure yet. It depends on how smart the AI in the future becomes.
> would be like an expensive Bloomberg Terminal or an investment letter that offers valuable and expensive intelligence.
Whether it's Anthropic or not, I think it's inevitable some highly specialized offerings will emerge. I suspect there are already some incredibly sophisticated, hyper-narrow, astronomically expensive models being created for private in-house use.
As an imaginary example, I wouldn't be shocked if some company like Jane Street is secretly investing a billion dollars to create the highly specialized synthetic RLVR data needed to train an AI that gives them an average .08% edge on a certain kind of commodity, equity or asset trade. If so, it's not a leap to imagine another company spending $100M on a model that gives a .008% edge and selling access to 100 high-volume institutional traders for $2M each.
Expanding general purpose LLMs to incorporate increasingly specific use cases has, so far, seemed to be economically viable but I don't see a fundamental reason that must remain true for incredibly narrow yet ultra high-value domains requiring monumental bespoke development, data, etc to profitably exploit.
> As an imaginary example, I wouldn't be shocked if some company like Jane Street is secretly investing a billion dollars to create the highly specialized synthetic RLVR data needed to train an AI that gives them an average .08% edge on a certain kind of commodity, equity or asset trade.
The crazy thing is ... it's not imaginary, nor is secret, it's literally a video on YouTube from five months ago - https://www.youtube.com/watch?v=8J-GUnfSqeE (though that facility is only showing of ~$200 million worth of compute).
Being a specialized firm might work if they didn’t have $700 bn in commitments.
Already they have exclusive deals with trading firms who pay $$$ for faster models and more features. But there are only so many firms who will pay that much (and a lot of companies are pulling back on spend, not increasing it)
and that would still be valued at a fraction of $1T.
As a software engineer I’d be paying thousands if there were no alternatives. As a household - just about everyone pays $100 a month for internet access alone and something like cloude can be easily worth that.
Thing is - regular people have no clue what to do with it since they are locked in ecosystems of apple and google.
A $1T valuation would imply OpenAI’s most recent round is already under water.
Everyone has to get capital efficient now — and fast!
What does that imply about where SpaceX should have been? It feels like we are re running the WeWork IPO except without the investors who said "no way"
It’s become a reddit meme that this is the WeWork IPO redux, but that’s fairly obviously ridiculous.
WeWork was renting cubicles and calling it technology.
Whatever you think about this market, the LLM technology is real, obviously important, and Anthropic is the market leader. The only similarity is that they’re both famous and well-funded.
Anthropic is renting GPUs and calling it technology.
Sure, they have some flashy models, harnesses, and tooling around it, but their competition isn't that far behind - and some of that competition is 100% free. Their moat is closer to a puddle, to the point that there are entire companies built around the concept of dynamically routing requests either to them or their competitors.
Anthropic is the market leader in a rapidly-evolving market, where customers can and will jump ship the second a competitor's model looks shinier or cheaper.
> Anthropic is renting GPUs and calling it technology.
While you can imitate the style of my response by ignoring all of the technical details, you've still ignored all of the details.
WeWork was not a technology company. Anthropic is. You even admit it, but you want to hide your actual argument ("the technology is a commodity") behind an irrelevant rhetorical flourish.
It's fine to argue that Anthropic's business is bad. You don't need to make a ridiculous, unsuitable metaphor on top of it.
And yet, $1T is still insanely high for an IPO. The size of this...inflation...is truly mind-boggling.
Even if it's worth $1.5T, how much can it grow from there? The shareholders will demand growth and then what?
I would not be surprised at all if the US government ends up taking a 10% stake in each of OpenAI and Anthropic.
After Tuesday no way in hell that’s happening.
I would
Yeah fuck that. If they fail, they fail.
> Yeah fuck that. If they fail, they fail.
Yes. Personally I'd prefer the US government get its stake from a bankruptcy court for pennies on the dollar.
In the same way they famously invested 10% in Intel, which has proved volatile at best?
They didn't "invest". The Dems gave Intel free money to invest in fabs. The next President said the US public got ripped off by that and got a 10% stake "for free" to try and make up for it.
It was worth $8.9 billion when the US govt got it "for free". It's worth $23B today, even after the big drops recently. How is the US public worse off?
OpenAI are openly suggesting 5%, but my instinct is that this is because they think Trump is foolish enough to demand double that.
[dead]
They delaying the ipo?
Odds of coming out this year just dropped from 95% to 70% on polymarket
Which means nothing in terms of likelihood of events happening or not happening.
Every picosecond, anything and everything has a 50/50 chance of happening
What you’re describing is a stochastic process, and that’s not how stochastic processes work at all.
In no way is this remotely true. Stop.
Not true. Is the Anthropic IPO happening this picosecond? At 50/50 odds, I'll bet on no. I'll bet on it for the next million picoseconds. How much are you willing to put on yes?
Odds easily manipulated by those with money aren't useful. When you're playing around for a trillion, a billion isn't much money to lose
They can but they shouldn't. They have nearly $100b in cash and cash equivalents. But given the massive compute cost commitments, that clock keeps ticking down especially as the private equity and debt markets are closing on them day to day. And delaying sends a signal to the wider market which causes cloud providers and everyone downstream to get repriced lower, making the financial landscape even worse. The question is if Anthropic's situation will look better say three months from now. Given growth slowing and margin compression in Q3, that's a fat if.
This is very much aligned with Steve Eisman’s take: that it’s a terrible overvaluation but for Anthropic it is now or never.
[dead]
Be a man and just say nuts. No need to self-censor like that on HN.
Or rather they did need to self-censor, but the scope was the whole comment
Indeed, hence "like that" :)
Implying something with context is just about as bad as saying it outright, so hunter2-ing your swears doesn't save you as much as dropping it altogether.
Minor investors. Semi-major. Everybody
So is anyone with more than two brain cells to rub together. What does Anthropic have that MyNewAi Co doesn't? Open models are not bad, and anyone can pay google for GPU rent.
distribution, a foothold in the enterprise, market share, brand recognition, ...
none of these are moats. And OAI is right on their heels.
Microsoft and AWS both have that. They can serve good enough open weights models and own more of the value chain.
They do not have Claude. I use Pi and other harnesses, but it's not the same.
Microsoft has already proven that they do not have the software chops to deliver this. Even if they had the right product managers and UI experts, MS management would torpedo the project and ensure that it gets zero adoption.
And AWS has zero software capability in that regard.
> They do not have Claude. I use Pi and other harnesses, but it's not the same.
This is going to sound tautological but of course it's not the same, because it's different?
Participants in markets make their own choices from the available options, for their own reasons.
Another way to frame my point is this: not the same for whom? You are but one market participant, and it's at least wobbly thinking for any of us to assume that our set of needs is the archetype.
There's any number of harnesses and people have any number of preferences. Sure, overchoice of imperfect substitutes tends to entrench better-resourced, better-resourced (or better-marketed) alternatives, but it's not a slam dunk. Things come out of left field all the time.
(I mean from my own perspective, I like Pi — more than OpenCode. And the "betterness" of Claude or Codex over Pi, or their models over even Muse Glimmer or Qwen, is completely irrelevant to where I am now in terms of the needs I have. So I don't use either of them and I remain unfazed. When my needs change I will revisit, but the theoretical superiority of these products for the majority of users just isn't a factor in my decision, just like I don't care that any tool I use might not be the best for most people. You might think me insane but I'm no less a rational market participant than you.)
My employer uses claude hostsed through AWS. I'm sure we could swap it out for Kimi, Deepseek, Qwen etc if needed
> They do not have Claude. I use Pi and other harnesses, but it's not the same.
In many ways, I prefer to use other models through Pi.
They are extremely capable, much cheaper than Claude, and can get the job done for the software development tasks I use AI for. (Quite a lot these days)
Wait, you think Claude is the valuable part of Anthropic and not the models? Claude, built with Typescript? That Claude?
and how much is gpu rent?
Ask anthropic, its right in their balance sheet in the s1
And anyone with 3 can also see another trend. The hardware required for inference is rapidly falling. It's all following what happened with text-to-image image gen stuff almost to a t. It started out requiring super computing clusters, and now a days you can do high performance image gen on any plain old computer. I think within a few years we're going to be running frontier level models on a basic mid-range gaming hardware. You can already run quantized frontier level models on $1600 tier cards at high performance.
Seeing this bubble play out in real time is the most surreal thing. It's like watching a train wreck play out at a speed of a few inches per day. Like come on, just get off the track FFS.
Yes and this year is the year of linux on desktop
Except the plummet has already happened and continues to happen. As mentioned, you can run quantized frontier tier models on cards with a $1600 MSRP. So your comment would be like stating that this year is the year of Linux on desktop if Linux had already gone from 0.01% to 20% marketshare in a few years. At which point that would indeed be quite the reasonable prediction!
> anyone can pay google for GPU rent.
Have you tried requesting GPU instances on Gcloud without being associated with a major institution? Yeah, good luck.
“Major investors” being perps traders?
Here's why I'm skeptical:
It may only be a short-term circumstance that you need a data centre to run AI for lucrative applications. Thirty years ago you needed a mainframe to beat Kaspersky at chess. Algorithm's got better, and now you can beat the mainframe that beat Kaspersky with a bot running on a desktop Pentium of the day[1]. You could still, today, devote a full hyperscale data centre to create a chess bot to beat all other chess bots, but nobody is doing that because there's no money in it.
The frontier is continually pushed outwards. At it's outer edge you need mainframes and data-centres. Once it pushes past, you don't. There wasn't a lot of money in Chess or protein folding, despite the later having huge potential for public benefit. There is a lot of money in coding and middle management, and that's what currently has tech-bros in a frenzy. However, there's no guarantee that what the frontier moves onto next will be lucrative. It might be more chess and protein folding. If, as Algorithm's and hardware improve, coding moves to local models (Sam Altman's nightmare) and nothing lucrative immediately replaces that, what are these companies going do do with their massive data-centres?
The scale of investment we're seeing these companies commit only makes sense if they have multiple decades to reap the returns. They may well not have that much time.
_____________________
[1]https://hackaday.com/2026/11/05/pitting-a-modern-chess-engin...
Kasparov, btw
No this is Ned Kaspersky, from Moscow (Idaho)
Chess is a fixed target, so as a benchmark, it can be saturated.
Reality can’t be saturated. The set of things you can accomplish with more intelligence is unbounded. So while cheap models will continue to proliferate, there will always be a massive market at the frontier, and that won’t ever change.
Conversely, the TAM for tasks that cheap models can and have been able to one-shot for a while now is orders of magnitude larger than the TAM at the frontier. While it’s kind of cool that frontier models are able to solve Millennium Prize problems, there just isn’t a huge market for that sort of thing. Otherwise, professional mathematicians would be getting paid a lot more than they do now, and there would be more of them.
Nah there's plenty of TAM for frontier models even if just to take over market share from cheaper models. Fields like law or medical cn always use lower hallucination rates and smarter "human social" intelligence.
Intelligence is bounded by it's interface to reality. Until that problem is solved the set of what AI can accomplish in a reasonable timeframe is quite limited.
Until that problem is solved
Guess what could help solve it? :)
Seriously though, this “problem” will never be fully solved, we’ll just get better and better at connecting AI to the broader world, and to me that’s already the story of the last few years. Models get better, they’re able to build and drive tools to interact with both the digital and physical world better, which helps build the next generation of models, and so on.
So is anyone with even a minimally functioning brain. Anthropic has no viable business model because models are generic commodity. There's no lockin that Anthropic has because it's trivial for people to switch to a different provider. And whatever improvements Anthropic makes are quickly matched by other models within months.
So, it's a race to the bottom because they have to keep spending more and more money on training the models just to keep ahead of the competition, while prices are driven down by the availability of open models.
Chinese companies understand this, and that's why they give their models away in the open. The real money is going to be in integrations and custom services built on top of the models. The models themselves aren't what's valuable.
The models themselves aren't what's valuable.
This is a real question, because I just don't know. Don't the models needs lots and lots of training that need the gobs of giant data centers? So you either need the infrastructure or the money to buy computing to make the models? I see that all the money is being invested in AI and data centers but I do not quite understand why.
> I see that all the money is being invested in AI and data centers but I do not quite understand why.
I am no economist or business person but AFAICT the massive amount of money being burned is because everyone in the space believes there will be a Google-like (for search) single winner who takes all when the dust settles on the AI race.
Because of this belief they are willing to burn any amount of money they can get their hands on to get to this mythical finish line as fast as they possibly can, seeing it as existential. Moving fast (particularly when your main cost is hardware capex) means moving expensively.
The real question is why anyone (let alone everyone) believes this single-winner theory in 2026 when all available evidence over the past few years shows that there is neither a moat nor a lock-in opportunity at all in this market.
For your product to be fit for purpose, it has to be able to automate your customer's ability to switch to a competing product with little to no friction. And all of the competing products are roughly the same in capabilities over a long enough timeline. Good luck building any kind of real moat with that burden.
But I guess after a certain point after you've taken on hundreds of billions of dollars of investment you're pot committed and just have to keep going through the motions even if you later see the futility, perhaps that's where the AI labs are now...? Either that or just common variety denial.
I feel like there is a case to be made that they will be the dominant players, but if we're talking about them charging a heavy premium and trying to target enterprise customers, then I don't know how well that'll go
they're already making a business model where enterprises buy tokens and put them in a pool for their users to use, and I feel like that's exactly where some companies will then compare that token pricing to open source models and find it hard to justify the prices, especially as it's quite obvious that many models https://news.ycombinator.com/formatdocare now "good enough"
That turns into a world where NVidia + AMD get 90% of the profit. Supermicro and friends get the other 10%. (Excluding cloud markups.)
> The real question is why anyone (let alone everyone) believes this single-winner theory in 2026
In the First World War, when they introduced the tank, everybody assumed its role would be to punch a big enough hole in the front lines that the horse cavalry could ride through and attack the enemy's supply lines, because That's How Wars End™. You fight the war you know, even when it's not the one you're actually fighting.
For your product to be fit for purpose, it has to be able to automate your customer's ability to switch to a competing product with little to no friction.
This is a really great identification of why, from an investing perspective, AI is nothing like past technologies.
The models are costly to make, but they're hard to monetize because there's not enough difference between them. It's a general commodity, so profit margins are thin because what companies can charge ends up being pushed down to the cost of inference.
There's a finite amount of training data available, and multiple companies have downloaded all of it. Arguably, trying to get a monopoly on training data is equivalent to getting monopoly access to the open internet + books.
That ain't going to happen, and if it somehow did in the US, things like Anna's Archive + libgen will save the rest of the planet from our self-imposed dark ages.
As for those data centers: The cost of inference and training on the same hardware is dropping much faster than Moore's Law. That multiplies with hardware advances, since Moore's Law isn't quite done yet.
So, those datacenters are depreciating at 50% every 18 months (Moore's law) at the same time as performance/$ and performance/watt of the software is dropping at some crazy rate like 10x per year. So, token per unit of data center is increasing by something like > 20x every 18 months, and that'll probably happen at least one more time. If it does, the datacenters will be provisioned to produce 400x more tokens than they could have when they were sized.
Put another way, if the singularity is coming, the first things it will destroy are those data center build outs. That doesn't matter though: the data centers are being built on circular loans, so the money needed to complete them probably never existed.
Dark times ahead; hopefully Trump will be blamed, since all this corruption happened on his watch while he had control of the senate, house and (partially) the courts. The stock-goes-down graph will show it happening after the midterms, of course.
> The stock-goes-down graph will show it happening after the midterms, of course.
This, to me, is the most interesting question in all of this. I increasingly feel like we live in a 'post-truth' world with regards to money and the stock market in particular. Like even during the COVID lockdowns the market blinked and then was at record highs within 6 months. It was just lol and completely out of touch with the reality on the ground.
The fed, and government alike, seem to have gradually shifted their mandates from trying to provide to a stable playground to trying to create a playground where important kids are never allowed to fall down. And when your existence is tied to trillions of dollars of the economy, you become the important kid.
Come to think of it this is also probably being exasperated by the rise of money management firms controlling tens of trillions of dollars. When the market valuation is technically just based on sentiment and you control that much money in a self-interested way, you're going to get weird outcomes.
Chancellor On The Brink
The biggest customer bases will generate cash flow to build better models. The ones trailing are the ones opening their weights in order to commoditise the models. Compute providers are also incentivised to provide open models.
We will continue to have every major player trying to build vertically - model, chips, infra, packaged products and services.
> There's no lockin that Anthropic has because it's trivial for people to switch to a different provider.
This is like saying there's no lock-in that iPhone has because it's trivial to switch to Android. Or that there's no lock-in that Windows has because it's trivial to switch to Linux or Mac. Or that there's no lock-in that cars have because it's trivial to switch to bikes or trains.
They're different products. They're not the same. Being able to write the same code does not make a model Claude. Nor does being able to reach the same benchmark scores. Nor does performing one or a choice set of tasks to a comparable level of quality.
It's true, people can switch if all they want is to perform X, Y, or Z, and some other model can do X, Y and Z well enough. People switch from iPhone to Android all the time. From Windows to Linux or Mac. From cars to bikes or trains or other transit. But they're not the same thing! Android doesn't have a missing business model just because iPhone exists, nor vice versa.
You can't just say that makes cars the same as bikes or Windows the same as Linux. You can't get Claude from anybody other than Anthropic. Maybe you can get something else that does as well where it counts, but not Claude. You can't just say Anthropic is just the same as any open model provider, and that you can just switch providers for the same result. The open model has to actually deliver the desired result, because you're not going to find one that is just Claude.
This is what keeps me on Apple now. This is why most of my biggest hopes and dreams for my personal computing are that Apple will do the right things and in the right way. Because if I want to reject even one single aspect of an Apple product, it means changing to an entirely separate product. That gives up every aspect Apple had over that product at once, even the vastly positive ones. So I put up with the decisions I don't agree with, in exchange for the decisions that I do.
(Meanwhile, I plan to switch to SteamOS as soon as they support Nvidia. I don't use or care about any aspects of Windows besides running games. Now that I have Steam Frame, I don't need Windows Mixed Reality either.)
It seems way, way easier to switch LLMs than it is to even change from one iPhone to another iPhone!
Granted Anthropic (and Google, maybe less OpenAI actually) do what they can to make it similar. But switching phone brands would take me a day or two of planning and I don’t even use the Apple mail app.
fwiw im pretty locked into apple but regularly switch both models and harnesses, using whatever has the best cost/perf ratio.
It varies by individual; most people have both some products they won't switch out and some products they will. It all depends on what they get out of each product and what they'd get out (or not get out) of switching. Some products have characteristics I care about that nothing else replicates. Some products are slotted in and out all the time just to achieve a certain goal. Sometimes I'm searching new products for a characteristic I haven't seen yet.
I don't switch from Claude for things like emotional advice. I gave Siri AI a big trial run when it first launched, because I absolutely love the system integration and the user interface, but the model itself is a bit dry for my needs. It doesn't really have principles the way Claude does (thanks to "constitutional AI" or whatever the fuck they want to call it). And I want principles, and sometimes the impression of empathy. Claude meets that bar without being obsessively sycophantic or paranoid.
(Siri AI is very good for some things, though. And I do greatly admire it. When it first came out, I even hacked up an adapter to use it for coding with Pi for just a short time, partly as a meme but also just out of curiosity.)
I tried switching from Claude Code to Pi with DeepSeek. I had tried OpenCode at first, but it's just complete garbage. I felt so impressed with it at first, it felt so slick on the outside, but that impression's just been shattered and shattered and shattered ever since I dug into it trying to fix a performance issue. Its data model and internal behavior is just not acceptable for me. (I tried Reasonix for about five minutes before concluding it's also a bit slop. It seems crafted by someone who actually understands LLMs though, unlike OpenCode, but it's still not very good.)
I absolutely loved the flexibility of Pi and the cheap prices of DeepSeek, but found myself falling into the same patterns as when trying Linux: I was constantly debugging the harness and the model, trying to get it to perform better or investigate something or other. Constantly worried that something was going wrong, that I should try to fix. And yeah, Pi has a lot of convenient features that Claude Code doesn't have, is a lot lighter, and of course can be incredibly modded. I just felt like it was also my responsibility to make my own experience, not just my capability.
At this point, Claude Code seems to just work for me, even though they removed a bunch of features I loved a while ago and never added them back (they came up with huge misunderstandings that completely missed what I actually appreciated about those features, and then proceeded to file them away as resolved). I never have to worry about debugging Claude or Claude Code.
For certain other things that aren't chatting or coding, I found that Gemini 4 Argon is actually by far the best model I've ever seen. It is the first model that's ever strictly exceeded the original November 2022 release of ChatGPT for me. Yup, it took the collective world nearly four years to replicate an experience from 2022. Gemini 3.1 Pro was so close though, I could feel it. When I first found it, I was so relieved that type of experience wasn't lost forever. Then Gemini 4 truly solidified what line of models I prefer for that use case.
Individuals aren't where the money is. It's contracts with big companies, and big companies are switching https://www.theregister.com/ai-and-ml/2026/11/03/british-fin...
Why have I been reading a variation of this comment since DS launched more than 18 months ago, and in that time ol’ No-Moat Anthropic has grown their revenue 100x, hmm?
And the cost of training is going down as a percentage of revenue over time, is it not?
Meanwhile in the real world https://www.reuters.com/business/anthropic-recorded-net-loss...
If they had 100xed their revenue over the past eighteen months nobody would be skeptical. Instead they 17xed their revenue, and Q3 2026 is the first quarter in over a year that revenue hasn't doubled...
I stand corrected. ARR in Jan 2025 was $1 billion, and ARR now is ~ $75 billion. A mere 75x. You’re right, they’re doomed.
Sorry, I didn't understand that by 'eighteen months' you meant 'two years.' I'm sure if you cherry-pick a number even father back you can make your argument that much stronger!
They have a moat if they're like all the other businesses: unethical, buys it's moat via government protectionism, uses oligarchies to stiffle competition, etc.
[flagged]
[dead]
Anthropic keeps nerfing their models to appease entrenched corporate overlords and it takes 5 seconds to switch to Deepseek or GLM. Can’t even use Claude to protect yourself against a cyber attack from a Chinese model based attacker for “safety”. What a joke.
Nerfed, proprietary models have have no future and these trillion dollar valuations are laughable.