Don't overthink it. Government exists to protect the wealthy at the expense of the many. The wealthy will be bailed out.
Wilhoit's law.
“Conservatism consists of exactly one proposition, to wit: There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.”
We slobs will pay for the bailout. Big Brother needs these data centers for mass surveillance, either directly or by proxy e.g. Google et.al.)
My guess is what it looks like is an enormous glut of HPC compute suddenly becoming available to science at virtually no cost at precisely the moment we’ve gutted the entire concept of science.
It’s also been harder than ever for scientists in the non-AI space to get more HPC compute with AI data centers causing pricing of components to skyrocket and constraining supply from manufacturers. So there will certainly be a boon in the HPC space one way or the other when the AI bubble pops if pricing and supply returns closer to pre-AI norms.
The problem with these kinds of analysis is that they're surface level. They criticize "number go up" but rely on it to make comparisons to other bubble events. For instance, AI infra buildout is like 2008 because number went up in both.
For instance:
> So the speculative discourse only works as long as investors subsidize the use of the technology. When that subsidy stops, these AI firms have to actually deliver value, or customers won’t buy it.
Are investors subsidizing the technology? There's upfront build out, but Anthropic is profitable and I believe the big labs are profitable on inference.
In comparison consider the real estate bubble in 2008. Why was real estate going up so much? Was there a surge of people coming into the country driving up demand? Were people using these houses? No it was purely inflationary driven by cheap money and financial engineering.
It also relies on arguments that this technology is all speculative, like one day we'll figure out how to use AI and demand will be high. But the demand is already there. Everybody is using AI, revenues are insane, it's the fastest growing product in history. It spans consumers and increasingly businesses.
The comparison to dot-com crash is also superficial. Was dot-com a bubble? I don't know, if you were transported to peak 2000 hype, would you argue "you guys are in a bubble, and the internet impact on the economy will be no greater than that of the fax machine"? No, of course not. In this case the supply outstripped demand. The build out and hype was early. But with AI you're seeing real usage.
It's not contagion. People are using this technology, not because it's cool and (apart from a few examples) because their bosses are forcing them to hit token metrics, but because it's actually useful and people are finding more uses every day.
It's just lazy. There is real risk that the build out is too much. But to make that argument you would have to say that the model intelligence would asymptote or become increasingly expensive such that its not worth it. Or that demand for broad intelligence is capped somehow. But saying we're in a bubble just because number goes up is not a strong argument.
> Are investors subsidizing the technology? There's upfront build out, but Anthropic is profitable and I believe the big labs are profitable on inference.
That's the thing about a bubble though. It's not about if things are profitable; it's if they will produce the profit in the future to support the current stock price.
Companies in 2000 were profitable and to this day still are (ex. Cisco [1] which despite only 26 years of inflation only reached it's dot-com stock price this year).
IIUC, Anthropic is $1T valuation on $10B revenue and $0.5B profit.
Google's has $4T valuation on $400B revenue and $100B profit. Which (dividing by 4 to get same valuation as Anthrophic) is $100B revenue and $25B profit.
IIUC, World GDP is ~$100T so Anthrophic "just" needs to get 0.1% of all economic activity to depend on them. Which to me actually seems like a tall order. Sure Google does more than that but Google spent ~30 years getting into that position.
So, Anthrophic 1/10 the required revenue despite every company in America pushing as hard as possible to use AI. What will it take for them to get it?
> So, Anthrophic 1/10 the required revenue despite every company in America pushing as hard as possible to use AI. What will it take for them to get it?
Anthropic revenue (annualized):
February 2026 $14 billion,
April 2026 $30 billion
May 2026 $47 billion
The market prices future revenue. What other company doubles revenue every few months? Where does it stop is the question. You need a few more doublings and things start making sense.
"Profitable on inference". Isn't that exactly the same a physical business saying "our widgets have a marginal cost of 95 cents to make, and we can sell for a dollar, so we're profitable, as long as you forget we have a $92 kajillion loan on the factory that has to be serviced."
It raises a chain of interesting questions: what if we pulled the plug on the expensive part (the training and associated infrastructure) in pursuit of making it economically viable?
- How much of the audience is using it based on the long-term promise? "It's still imperfect and annoying, but I want to be ready for when it finally turns into Lieutenant Commander Data." If the vendors said "this is what you actually get once the honeymoon ends", would customers still be satisfied with the product and pricing?
- How do you stop the game of economic chicken? If Anthropic said "Fable is the last model we can offer (until we can pay down the costs to get there)", any competitor with a dime of runway left, will spend a cent of it on training and 9 cents advertising "do you want to be stuck with old tech?"
> "Profitable on inference". Isn't that exactly the same a physical business saying "our widgets have a marginal cost of 95 cents to make, and we can sell for a dollar, so we're profitable, as long as you forget we have a $92 kajillion loan on the factory that has to be serviced."
Yes, that's called an investment. That money's already spent. Look at the marginal revenue of many business. What's going to happen? They'll raise prices because legacy costs? And then the people distilling these models will come in w/out the baggage. Cars for instance have a huge up front cost in design and manufacturing capacity and they only sell for 5-20% more than it costs them to make that one unit. It's a competitive industry
My point is that you can't cherry pick a profitable business unit if you don't have a story for how the entire business can operate profitably.
Cars have low margins, but they generally don't rely on an ongoing infusion of investor money to balance the books. The overall venture still has to turn a profit. Nobody is walking into Hyundai HQ and saying "We are going to sell Sonatas for $6,750 each, because if we can do it long enough, some magic will happen and we'll end up back in the green."
TBH, I'm not quite sure what the "some magic will happen" angle is for AI.
Compute gets cheaper, but I suspect the training arms race is running up costs faster still.
We're already seeing hints of price balking (definitely heard people at work saying they're hesitant about Fable due to the costs) so it's unclear if there's headroom there.
TBH, the best answer I can figure right now is that many players are hoping for a competitor's flameout-- the dream of being the last man standing, and then able to dictate market terms.
> Was dot-com a bubble? I don't know, if you were transported to peak 2000 hype, would you argue "you guys are in a bubble, and the internet impact on the economy will be no greater than that of the fax machine"?
I am old enough to remember this. As a young guy who was on BBSes and IRC, it was obvious to me that the Internet was going to be huge. At the same time, it was obvious that the "dot com bubble" was a real problem. We had that story about Allbirds closing up shop and claiming to be an AI company and their stock went up 300% or whatever - those stories were everywhere in 2000. Companies were IPOing like mad, doubling or more as soon as they launched, no one was making any revenue. It was obviously a house of cards. But year after year, this was true and you were an idiot for not jumping on boad and making free money... until the carousel stopped.
There are definite similarities, but now those risks are all concentrated into a few companies with illogical valuations that are consuming the market. Yes, there are a bunch of AI-using companies, but they are much narrower in their impact and not nearly as many IPOs.
The true world-changing stuff is slow and takes a decade to permeate, and it happens regardless of the investment nonsense. Building up fiber networks, sorting out SSL, simplifying hosting, better software dev tools for the web, etc. For AI that looks like changing organizational structures, adding consistent safeguards, task-aligned workflows. There are much fewer physical changes, but the organizational differences are quite pervasive and will take time.
Still, there is no doubt this is a bubble and it will pop. My assumption is that can't happen until all the big players IPO and pension funds are left holding the bag. This has been PE's playbook for decades.
If it's just going to mean cheaper hardware, cheaper access to models and cheaper energy prices, it's going to be all right I think!
I very much doubt the general public will accept a any sort of government bailout - people are completely done with the rug pulls, the only worrying sign is the new cult standard of people that just blindly accept whatever their leader states.
But in a moment of crisis it might reactivate normal people back into voting. Cause if it doesn't, it might be the final nail on some democracies as we know.
So yeah, either not such a big deal or the final nail for some decades.
Don't overthink it. Government exists to protect the wealthy at the expense of the many. The wealthy will be bailed out.
Wilhoit's law.
“Conservatism consists of exactly one proposition, to wit: There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.”
We slobs will pay for the bailout. Big Brother needs these data centers for mass surveillance, either directly or by proxy e.g. Google et.al.)
https://pylimitics.net/wilhoits-law/
It will not pop before the risk is sold to retail investors at a huge profit.
I think open source AI could cause it. I hope local AI takes over but this government will probably ban it.
My guess is what it looks like is an enormous glut of HPC compute suddenly becoming available to science at virtually no cost at precisely the moment we’ve gutted the entire concept of science.
It’s also been harder than ever for scientists in the non-AI space to get more HPC compute with AI data centers causing pricing of components to skyrocket and constraining supply from manufacturers. So there will certainly be a boon in the HPC space one way or the other when the AI bubble pops if pricing and supply returns closer to pre-AI norms.
We may be finding out come Monday premarket
The problem with these kinds of analysis is that they're surface level. They criticize "number go up" but rely on it to make comparisons to other bubble events. For instance, AI infra buildout is like 2008 because number went up in both.
For instance:
> So the speculative discourse only works as long as investors subsidize the use of the technology. When that subsidy stops, these AI firms have to actually deliver value, or customers won’t buy it.
Are investors subsidizing the technology? There's upfront build out, but Anthropic is profitable and I believe the big labs are profitable on inference.
In comparison consider the real estate bubble in 2008. Why was real estate going up so much? Was there a surge of people coming into the country driving up demand? Were people using these houses? No it was purely inflationary driven by cheap money and financial engineering.
It also relies on arguments that this technology is all speculative, like one day we'll figure out how to use AI and demand will be high. But the demand is already there. Everybody is using AI, revenues are insane, it's the fastest growing product in history. It spans consumers and increasingly businesses.
The comparison to dot-com crash is also superficial. Was dot-com a bubble? I don't know, if you were transported to peak 2000 hype, would you argue "you guys are in a bubble, and the internet impact on the economy will be no greater than that of the fax machine"? No, of course not. In this case the supply outstripped demand. The build out and hype was early. But with AI you're seeing real usage.
It's not contagion. People are using this technology, not because it's cool and (apart from a few examples) because their bosses are forcing them to hit token metrics, but because it's actually useful and people are finding more uses every day.
It's just lazy. There is real risk that the build out is too much. But to make that argument you would have to say that the model intelligence would asymptote or become increasingly expensive such that its not worth it. Or that demand for broad intelligence is capped somehow. But saying we're in a bubble just because number goes up is not a strong argument.
> Are investors subsidizing the technology? There's upfront build out, but Anthropic is profitable and I believe the big labs are profitable on inference.
That's the thing about a bubble though. It's not about if things are profitable; it's if they will produce the profit in the future to support the current stock price.
Companies in 2000 were profitable and to this day still are (ex. Cisco [1] which despite only 26 years of inflation only reached it's dot-com stock price this year).
IIUC, Anthropic is $1T valuation on $10B revenue and $0.5B profit.
Google's has $4T valuation on $400B revenue and $100B profit. Which (dividing by 4 to get same valuation as Anthrophic) is $100B revenue and $25B profit.
IIUC, World GDP is ~$100T so Anthrophic "just" needs to get 0.1% of all economic activity to depend on them. Which to me actually seems like a tall order. Sure Google does more than that but Google spent ~30 years getting into that position.
So, Anthrophic 1/10 the required revenue despite every company in America pushing as hard as possible to use AI. What will it take for them to get it?
[1]: https://www.google.com/search?q=cisco+stock&oq=cisco+stock
> So, Anthrophic 1/10 the required revenue despite every company in America pushing as hard as possible to use AI. What will it take for them to get it?
Anthropic revenue (annualized):
February 2026 $14 billion,
April 2026 $30 billion
May 2026 $47 billion
The market prices future revenue. What other company doubles revenue every few months? Where does it stop is the question. You need a few more doublings and things start making sense.
"Profitable on inference". Isn't that exactly the same a physical business saying "our widgets have a marginal cost of 95 cents to make, and we can sell for a dollar, so we're profitable, as long as you forget we have a $92 kajillion loan on the factory that has to be serviced."
It raises a chain of interesting questions: what if we pulled the plug on the expensive part (the training and associated infrastructure) in pursuit of making it economically viable?
- How much of the audience is using it based on the long-term promise? "It's still imperfect and annoying, but I want to be ready for when it finally turns into Lieutenant Commander Data." If the vendors said "this is what you actually get once the honeymoon ends", would customers still be satisfied with the product and pricing? - How do you stop the game of economic chicken? If Anthropic said "Fable is the last model we can offer (until we can pay down the costs to get there)", any competitor with a dime of runway left, will spend a cent of it on training and 9 cents advertising "do you want to be stuck with old tech?"
> "Profitable on inference". Isn't that exactly the same a physical business saying "our widgets have a marginal cost of 95 cents to make, and we can sell for a dollar, so we're profitable, as long as you forget we have a $92 kajillion loan on the factory that has to be serviced."
Yes, that's called an investment. That money's already spent. Look at the marginal revenue of many business. What's going to happen? They'll raise prices because legacy costs? And then the people distilling these models will come in w/out the baggage. Cars for instance have a huge up front cost in design and manufacturing capacity and they only sell for 5-20% more than it costs them to make that one unit. It's a competitive industry
What's your point?
My point is that you can't cherry pick a profitable business unit if you don't have a story for how the entire business can operate profitably.
Cars have low margins, but they generally don't rely on an ongoing infusion of investor money to balance the books. The overall venture still has to turn a profit. Nobody is walking into Hyundai HQ and saying "We are going to sell Sonatas for $6,750 each, because if we can do it long enough, some magic will happen and we'll end up back in the green."
TBH, I'm not quite sure what the "some magic will happen" angle is for AI.
Compute gets cheaper, but I suspect the training arms race is running up costs faster still.
We're already seeing hints of price balking (definitely heard people at work saying they're hesitant about Fable due to the costs) so it's unclear if there's headroom there.
TBH, the best answer I can figure right now is that many players are hoping for a competitor's flameout-- the dream of being the last man standing, and then able to dictate market terms.
> Was dot-com a bubble? I don't know, if you were transported to peak 2000 hype, would you argue "you guys are in a bubble, and the internet impact on the economy will be no greater than that of the fax machine"?
I am old enough to remember this. As a young guy who was on BBSes and IRC, it was obvious to me that the Internet was going to be huge. At the same time, it was obvious that the "dot com bubble" was a real problem. We had that story about Allbirds closing up shop and claiming to be an AI company and their stock went up 300% or whatever - those stories were everywhere in 2000. Companies were IPOing like mad, doubling or more as soon as they launched, no one was making any revenue. It was obviously a house of cards. But year after year, this was true and you were an idiot for not jumping on boad and making free money... until the carousel stopped.
There are definite similarities, but now those risks are all concentrated into a few companies with illogical valuations that are consuming the market. Yes, there are a bunch of AI-using companies, but they are much narrower in their impact and not nearly as many IPOs.
The true world-changing stuff is slow and takes a decade to permeate, and it happens regardless of the investment nonsense. Building up fiber networks, sorting out SSL, simplifying hosting, better software dev tools for the web, etc. For AI that looks like changing organizational structures, adding consistent safeguards, task-aligned workflows. There are much fewer physical changes, but the organizational differences are quite pervasive and will take time.
Still, there is no doubt this is a bubble and it will pop. My assumption is that can't happen until all the big players IPO and pension funds are left holding the bag. This has been PE's playbook for decades.
If it's just going to mean cheaper hardware, cheaper access to models and cheaper energy prices, it's going to be all right I think!
I very much doubt the general public will accept a any sort of government bailout - people are completely done with the rug pulls, the only worrying sign is the new cult standard of people that just blindly accept whatever their leader states.
But in a moment of crisis it might reactivate normal people back into voting. Cause if it doesn't, it might be the final nail on some democracies as we know.
So yeah, either not such a big deal or the final nail for some decades.
Did the last AI bubble even pop? Why would I expect this one, arguably more groundbreaking wave of changes to "pop"?