I mean, sure we're in a bubble, but the trick is to call it, with that old Keynes quote about the market staying irrational longer than you can stay liquid.
But also:
> "So, you're saying a third of the stock market is tied up in seven AI companies that have no way to become profitable and that this is a bubble that's going to burst and take the whole economy with it?"
> I said, "Yes, that's right."
that is something different in this case isn't it? those seven companies making up a third of the market do not need to become profitable, they are insanely profitable. Mostly they invest a lot in AI but if that doesn't pay out, all but NVidia have their day job to go back to.
The difference to the dot com bubble is that the unprofitable companies are privately held and the public companies are extremely profitable and have finally found something to soak up their ridiculous profits other than stock buybacks. How a crash affects anyone other than high-net-worth individuals with money tied up in VC funds is not explained.
As a corporate finance and valuation geek, Ill warn you now: dont try and time mood and momentum. Thats what is driving much of the valuations being thrown around.
If this blows up big time and it is found that the Big Tech firms were operating on lies and false hope, there will be consequences - in the form of shareholders demanding cash returned and setting limits on the cash balance held by Google et al. Apple has stayed smart staying out of this nonsense and not doing M&A.
Investing in projects with negative NPV destroys the wealth of shareholders.
So I do think we're in a bubble, but I also remember when all the discussion around here was around Uber, and I read many, many hot takes about how they were vastly unprofitable, had no real business model, could never be profitable, and only existed because investors were pumping in money and as soon as they stopped, Uber would be dead. Well, it's now ten years later, Uber still exists, and last year they made $43.9bn in revenue and net income of $9.8bn.
Oh dear, we are definitely in a bubble, it's just not in the way of total burst.
Back when everybody got into website building, Microsoft released a software called FrontPage, a WYSIWYG HTML editor that could help you build a website, and some of it's backend features too. With the software you can create a website completed with home, newspages and guestbooks, with ease, compare to writing "raw" code.
Now days however, almost all of us are still writing HTML and backend code manually. Why? I believe it's because the tool is too slow to fit in a quick-moving modern world. It takes Microsoft weeks of work just to come out with something that poorly mimics what was invented by an actual web dev in an afternoon.
Humans are adoptive, tools are not. Some times, tools can better humans in productivity, sometime it can't.
AI is still founding it's use cases. Maybe it's good at acting like a cheap, stupid and spying secretary for everyone, and maybe it can write some code for you, but if you ask it to "write me a YouTube", it just can't help you.
Problem is, real boss/user would demand "write me a YouTube" or "build a Fortnite" or "help me make some money". The fact that you have to write a detailed prompt and then debug it's output, is the exact reason why it's not productive. The reality that it can only help you writing code instead of building an actually usable product based on a simple sentence such as "the company has decided to move to online retail, you need to build a system to enable that" is a proof of LLM's shortcomings.
So, AI has limits, and people are finding out. After that, the bubble will shrink to fit it's actual value.
This is underselling the Uber story to a degree. The original sell for Uber was that their total addressable market was the entire auto industry because people will start preferring taxis over driving. They are still trying to achieve that with similar stories now pushed to sell robotaxis.
Uber was undercutting traditional taxis either through driver incentive or cheaper pricing. Many hot takes were around the sustainability of this business model without VC money. In many places this turned out to be true. Driver incentives are way down and Uber pricing is way up.
That said, this is also conflating one company with an industry. Uber might have survived but how many ride sharing companies have survived in total? How many markets have Uber left because it couldn’t sustain?
In a bubble the destruction is often that some big companies get destroyed and others survive. For every pets.com there is one Amazon. That doesn’t mean Amazon is good example to say naysayers during the dot bubble were wrong.
Simplifying Uber's story to "pricing or more drivers" misses the most important part.
Uber was undercutting traditional taxis because, at least in the US, the traditional taxis was horrible user experience. No phone app, no way to give feedback on driver, horrible cars, unpredictable charges... This was because taxis had monopoly in most cities, so they really did not care about customers.
The times when Uber was super-cheap have long passed, but I still never plan to ride regular taxis. It's Waymo (when availiable) or Lyft for me.
Uber was unprofitable and when it ceased to be unprofitable ceased to be better.
They did managed to offload price on weaker actors party by simply ugnoring laws and hoping it will work for them. It did, but it was not exactly some grand inspiring victory and more of success of "some dont have to follow the law" corruption.
I disagree with Doctorow's conversation with a student at Cornell. You can prevent further misallocation of funds by agitating against "AI" usage in general. If you are at Cornell, organize meetings, protests etc. against the dehumanization and decreased job prospects.
As a student, you have much more freedom to protest than as an employee, and that is where the resistance must come from.
We also need to take into account that while there is a bubble, most of the insane amounts of investment that were seen in headlines have not materialized.
Nvidia will crash, Tesla will crash (Optimus robot nonsense) but Microsoft and Google should be fine. If there is a bailout, protest again. preferably in the physical space and focusing on economic topics rather than culture wars (which is what the politicians want you to focus on).
Something of a logical leap here: if LLMs aren't capable of replacing workers and it's all lies, then what company is going to engage in mass layoffs without seeing results first? Sounds like companies that deserve to go away.
> Plan for a future where you can buy GPUs for ten cents on the dollar, where there's a buyer's market for hiring skilled applied statisticians, and where there's a ton of extremely promising open source models that have barely been optimized and have vast potential for improvement.
This actually sounds like a kinda cool outcome as long as you aren’t an applied statistician.
> but an AI salesman can 100% convince your boss to fire you and replace you with an AI that can't do your job, and when the bubble bursts, the money-hemorrhaging "foundation models" will be shut off and we'll lose the AI that can't do your job, and you will be long gone, retrained or retired or "discouraged" and out of the labor market, and no one will do your job.
Even if the big AI companies turn off their APIs, people will still be able to run local models as well as some other, new business spun up to run them as SaaS.
Isn't the training most of the cost? In which case the current models could have a very long lifetime even if new models are never trained again. They'll go gradually out of date, but for many purposes will still be useful. If they can pull new info from the web they may stay relevant for decades. It's only if running the chatbots is not cost effective that everything halts and my understanding is that the cost of that is lower relatively. Even now, older models are still being used. Also, performance optimizations seem likely to soon reduce the need for data center build out and reduce costs. Seems too soon to say where this is all going. Who even knows if the GPU chips will improve dramatically or if something else (more AI optimized processor architectures) will replace them? It's true that right now it looks like a bubble, but the future is still very much in flux, and the value of the models already created may not disappear overnight.
There is still a demand for these tools. I know they are useful to me. Do they make me more productive as a software engineer? Probably not, at least not significantly. But they're still useful, especially for little tools and one-off scripts which are not intended to become production code anywhere.
I also just enjoy using them for bouncing ideas off of them and doing sanity checks on all sorts of topics, personal and work-related. Sometimes they spark a better idea that I may not have had otherwise. I will still be using them after the bubble bursts.
That being said, I'm also fine if all the current AI companies implode and I'm just running an OSS model locally.
The whole thing was lost on the mag 7 being unprofitable and that he was so sick of talking about AI he decided to take his shot at making money writing a book about AI.
What percent of consumption will go to ai ? For me probably atleast 10%. What percent of investment will go to ai ? For me another 10% probably. I mean some of it will come from less consumption and investment in other things.
> Plan for a future where you can buy GPUs for ten cents on the dollar, where there's a buyer's market for hiring skilled applied statisticians, and where there's a ton of extremely promising open source models that have barely been optimized and have vast potential for improvement
This doesn’t square entirely with the earlier claim that AI companies have (and will continue to have) “dogshit unit economics”.
If you have a bunch of cheap “applied statistician” labor (kind of a reductive take, btw), cheap GPUs, and powerful open source models, it is a near certainty that companies would achieve favorable unit economics by optimizing existing models to run much more efficiently on existing GPUs.
I happily pay $20/month for Google One to use Gemini 2.5 Pro. I don’t really need it to be a whole lot better. It’s a great product. If they can deliver inference of that level with positive margin (and keep it ad free), it’s a viable business.
Investors will likely lose billions, if not trillions, but I don’t think the industry is inherently unprofitable - I just don’t think anybody has been incentivized to optimize for cost yet. Why would you, when investors continue to throw money at you to scale?
I'm 52. I experienced the dot-com bubble very up close. I was in the Raleigh-Durham area for most of it. There were hundreds of startups all over the area. Companies like Nortel were booming. IBM was booming. By 2003 it was all gone--Nortel was a shell of its former self, IBM laid off huge numbers of workers. There was suddenly a glut of office space all over. We moved out in 2006, but even around then there was still a glut of office space! I don't know if it ever recovered because it was so overbuilt.
I remember having lunch with a guy who was stubbornly holding on to his Nortel stock, which was worth mere pennies by like 2005 or so. They not only lost their jobs, they lost their 401Ks which were all in company stock. Anyways, this guy was sure it was going to bounce back. I saw in like 2008 where Nortel finally closed its doors and the stock was de-listed at $0. His dream was dead. I never worked for equity after that time period.
The enormous build out of data centers reminds me of that time period. Yeah, it's all going to collapse.
I largely agree. I don't think AI is ultimately useless, but I think it's about 10% as useful as the broader market seems to think it is. That said, every time I see another article like this, I think to myself "well it's not going to come crashing down any time soon". This is the nature of bubbles - they only collapse when nearly everyone is finally convinced they never will. Right now, stories about the AI bubble collapsing are everywhere, which means the time hasn't come yet.
I have an idea that the market may actually start to react positively to bad job numbers, as that could be taken as a signal that companies are shedding people to replace them with AI (even if that's not the actual reason for the bad numbers). If job numbers started suddenly improving and the unemployment rate dropping, it could be taken to mean that AI is not going to replace everyone after all.
>AI is the asbestos we are shoveling into the walls of our society and our descendants will be digging it out for generations
Seems a bit pessimistic. AGI may not be here next year to keep the bubble going but will probably happen in the next decade or two and do much of the stuff advertised. It's like the dotcom bubble - much of commerce, banking and the like did move to the internet but not till a while after the financial bubble burst.
"So, you're saying a third of the stock market is tied up in seven AI companies that have no way to become profitable and that this is a bubble that's going to burst and take the whole economy with it?"
> This isn't like the early days of the web, or Amazon, or any of those other big winners that lost money before becoming profitable. Those were all propositions with excellent "unit economics" – they got cheaper with every successive technological generation, and the more customers they added, the more profitable they became. AI companies have – in the memorable phraseology of Ed Zitron – "dogshit unit-economics." Each generation of AI has been vastly more expensive than the previous one, and each new AI customer makes the AI companies lose more money...
See, I think this is wrong. The unit economics of LLMs are great, and more than that, they have a fuckton of users with obvious paths to funding for those users that aren't paying per unit (https://www.snellman.net/blog/archive/2025-06-02-llms-are-ch...). The problem is the ludicrous up front over infestment, none of which was actually necessary to get to useful foundation models, as we saw with DeepSeek.
Great points but timing it can be very hard. It can last many more years because this time they have a thing called "money printer". When crash happens, they will use it.
Yes it prints whatever amount they want, even trillions. Magically(!)
> Further: the topline growth that AI companies are selling comes from replacing most workers with AI, and re-tasking the surviving workers as AI babysitters ("humans in the loop"), which won't work. Finally: AI cannot do your job, but an AI salesman can 100% convince your boss to fire you and replace you with an AI that can't do your job
This hits home. A lot of the supposed claims of improvements due to AI that I see are not really supported by measurements in actual companies. Or they could have been just some regular automation 10 years ago, except requiring less code.
If anything I see a tendency of companies, and especially AI companies, to want developers and other workers to work 996 in exchange for magic beans (shares) or some other crazy stupid grift.
53 comments
[ 2.5 ms ] story [ 88.1 ms ] threadBut also: > "So, you're saying a third of the stock market is tied up in seven AI companies that have no way to become profitable and that this is a bubble that's going to burst and take the whole economy with it?"
> I said, "Yes, that's right."
that is something different in this case isn't it? those seven companies making up a third of the market do not need to become profitable, they are insanely profitable. Mostly they invest a lot in AI but if that doesn't pay out, all but NVidia have their day job to go back to.
Real estate and crypto on the other hand...
As a corporate finance and valuation geek, Ill warn you now: dont try and time mood and momentum. Thats what is driving much of the valuations being thrown around.
If this blows up big time and it is found that the Big Tech firms were operating on lies and false hope, there will be consequences - in the form of shareholders demanding cash returned and setting limits on the cash balance held by Google et al. Apple has stayed smart staying out of this nonsense and not doing M&A.
Investing in projects with negative NPV destroys the wealth of shareholders.
If you know of a better place on the internet LMK!
Back when everybody got into website building, Microsoft released a software called FrontPage, a WYSIWYG HTML editor that could help you build a website, and some of it's backend features too. With the software you can create a website completed with home, newspages and guestbooks, with ease, compare to writing "raw" code.
Now days however, almost all of us are still writing HTML and backend code manually. Why? I believe it's because the tool is too slow to fit in a quick-moving modern world. It takes Microsoft weeks of work just to come out with something that poorly mimics what was invented by an actual web dev in an afternoon.
Humans are adoptive, tools are not. Some times, tools can better humans in productivity, sometime it can't.
AI is still founding it's use cases. Maybe it's good at acting like a cheap, stupid and spying secretary for everyone, and maybe it can write some code for you, but if you ask it to "write me a YouTube", it just can't help you.
Problem is, real boss/user would demand "write me a YouTube" or "build a Fortnite" or "help me make some money". The fact that you have to write a detailed prompt and then debug it's output, is the exact reason why it's not productive. The reality that it can only help you writing code instead of building an actually usable product based on a simple sentence such as "the company has decided to move to online retail, you need to build a system to enable that" is a proof of LLM's shortcomings.
So, AI has limits, and people are finding out. After that, the bubble will shrink to fit it's actual value.
Uber was undercutting traditional taxis either through driver incentive or cheaper pricing. Many hot takes were around the sustainability of this business model without VC money. In many places this turned out to be true. Driver incentives are way down and Uber pricing is way up.
That said, this is also conflating one company with an industry. Uber might have survived but how many ride sharing companies have survived in total? How many markets have Uber left because it couldn’t sustain?
In a bubble the destruction is often that some big companies get destroyed and others survive. For every pets.com there is one Amazon. That doesn’t mean Amazon is good example to say naysayers during the dot bubble were wrong.
Uber was undercutting traditional taxis because, at least in the US, the traditional taxis was horrible user experience. No phone app, no way to give feedback on driver, horrible cars, unpredictable charges... This was because taxis had monopoly in most cities, so they really did not care about customers.
The times when Uber was super-cheap have long passed, but I still never plan to ride regular taxis. It's Waymo (when availiable) or Lyft for me.
They did managed to offload price on weaker actors party by simply ugnoring laws and hoping it will work for them. It did, but it was not exactly some grand inspiring victory and more of success of "some dont have to follow the law" corruption.
Uber sunk overall, until profitability, less than $100 billion over nearly 2 decades.
By analogy (which is basically anecdotal evidence but with cognitive rhyme) we should have profitable LLMs in about 320 years.
It makes one tempted to take the sky is falling as a buy signal.
As a student, you have much more freedom to protest than as an employee, and that is where the resistance must come from.
We also need to take into account that while there is a bubble, most of the insane amounts of investment that were seen in headlines have not materialized.
Nvidia will crash, Tesla will crash (Optimus robot nonsense) but Microsoft and Google should be fine. If there is a bailout, protest again. preferably in the physical space and focusing on economic topics rather than culture wars (which is what the politicians want you to focus on).
This actually sounds like a kinda cool outcome as long as you aren’t an applied statistician.
Even if the big AI companies turn off their APIs, people will still be able to run local models as well as some other, new business spun up to run them as SaaS.
I also just enjoy using them for bouncing ideas off of them and doing sanity checks on all sorts of topics, personal and work-related. Sometimes they spark a better idea that I may not have had otherwise. I will still be using them after the bubble bursts.
That being said, I'm also fine if all the current AI companies implode and I'm just running an OSS model locally.
So many AI hucksters these days
This doesn’t square entirely with the earlier claim that AI companies have (and will continue to have) “dogshit unit economics”.
If you have a bunch of cheap “applied statistician” labor (kind of a reductive take, btw), cheap GPUs, and powerful open source models, it is a near certainty that companies would achieve favorable unit economics by optimizing existing models to run much more efficiently on existing GPUs.
I happily pay $20/month for Google One to use Gemini 2.5 Pro. I don’t really need it to be a whole lot better. It’s a great product. If they can deliver inference of that level with positive margin (and keep it ad free), it’s a viable business.
Investors will likely lose billions, if not trillions, but I don’t think the industry is inherently unprofitable - I just don’t think anybody has been incentivized to optimize for cost yet. Why would you, when investors continue to throw money at you to scale?
So you think those $20/month are generating profits?
Because Google is burning its own cash.
I remember having lunch with a guy who was stubbornly holding on to his Nortel stock, which was worth mere pennies by like 2005 or so. They not only lost their jobs, they lost their 401Ks which were all in company stock. Anyways, this guy was sure it was going to bounce back. I saw in like 2008 where Nortel finally closed its doors and the stock was de-listed at $0. His dream was dead. I never worked for equity after that time period.
The enormous build out of data centers reminds me of that time period. Yeah, it's all going to collapse.
I have an idea that the market may actually start to react positively to bad job numbers, as that could be taken as a signal that companies are shedding people to replace them with AI (even if that's not the actual reason for the bad numbers). If job numbers started suddenly improving and the unemployment rate dropping, it could be taken to mean that AI is not going to replace everyone after all.
> No of course there isn't enough capital for all of this. Having said that, there is enough capital to do this for a at least a little while longer.
https://www.wheresyoured.at/openai-onetrillion/
As for money side - think it’ll come. There is obvious utility (but not autonomy) and the economics of it will find their equilibrium. They always do
Seems a bit pessimistic. AGI may not be here next year to keep the bubble going but will probably happen in the next decade or two and do much of the stuff advertised. It's like the dotcom bubble - much of commerce, banking and the like did move to the internet but not till a while after the financial bubble burst.
I said, "Yes, that's right."
Which companies are those?
See, I think this is wrong. The unit economics of LLMs are great, and more than that, they have a fuckton of users with obvious paths to funding for those users that aren't paying per unit (https://www.snellman.net/blog/archive/2025-06-02-llms-are-ch...). The problem is the ludicrous up front over infestment, none of which was actually necessary to get to useful foundation models, as we saw with DeepSeek.
Yes it prints whatever amount they want, even trillions. Magically(!)
This hits home. A lot of the supposed claims of improvements due to AI that I see are not really supported by measurements in actual companies. Or they could have been just some regular automation 10 years ago, except requiring less code.
If anything I see a tendency of companies, and especially AI companies, to want developers and other workers to work 996 in exchange for magic beans (shares) or some other crazy stupid grift.