Is it really a bubble about to burst when literally everyone is talking about AI being in a bubble and maybe bursting soon?
To me, we're clearly not peak AI exuberance. AI agents are just getting started and getting so freaking good. Just the other day, I used Vercel's v0 to build a small business website for a relative in 10 minutes. It looked fantastic and very mobile friendly. I fed the website to ChatGPT5.1 and asked it to improve the marketing text. I fed those improvements back to v0. Finished in 15 minutes. Would have taken me at least one week in the past to do a design, code it, test it for desktop/mobile, write the copy.
The way AI has disrupted software building in 3 short years is astonishing. Yes, code is uniquely great for LLM training due to open source code and documentation but as other industries catch up on LLM training, they will change profoundly too.
Google, Meta, Microsoft, and Amazon will get through easily. They don't have excessive debt. They can afford to lose their investments into AI. Their valuations will take a hit. Nvidia will lose revenue and profits, stock will go down by 60% or more, but it will also survive.
Oracle will likely fail. It funded its AI pivot with debt. The Debt-to-Revenue ratio is 1.77, the Debt-to-Equity ratio D/E is 520, and it has a free cash flow problem.
OpenAI, Anthropic, and others will be bought for cents on the dollar.
Makes one think that this was the plan all along. I think they saw how SVB went down and realize that if they're reckless and irresponsible at a big enough scale they can get the government to just transfer money to them. It's almost like this is their new business model "we're selling exposure to the $XX trillion dollar bailout industry."
Hasn't it been pretty widely acknowledged that AI funding has created a whirlpool of money cycling between a few players- cloud / datacenter hosts / operators (oracle), GPU (nvidia) and model operators (openai).
To pile on, there's hardly a product being developed that doesn't integrate "ai" in some way. I was trying to figure out why my brand new laptop was running slowly, and (among other things) noticed 3 different services running- microsoft copilot, microsoft 365 copilot (not the same as the first, naturally) and the laptop manufacturer's "chat" service. That same day, I had no fewer than 5 other programs all begging me to try their AI integrations.
Job boards for startups are all filled with "using AI" fluff because that's the only thing investors seem to want to put money into.
> He told the BBC that the company owns what he called a “full stack” of technologies, from chips to YouTube data to models and frontier science research. This integrated approach, he suggested, would help the company weather any market turbulence better than competitors.
I guess but is it better for an investor to own 2 shares of Google or 1 share of OpenAI and 1 share of TSMC?
Like I have no doubt that being vertically integrated as a single company has lot of benefits but one can also create a trust that invests vertically as well.
Well if AI goes poof - the equity markets take a really big bad hit. So I would probably move out of equity and into something more concrete and reinvest if you can time the market bottom.
Nvidia earnings tomorrow will be the litmus test if things are going to topple over.
In more traditional industries investors seem to prefer less integration. One example being Siemens, who divested their Healthineers and Energy divisions and all their share values increased by a lot.
Maybe this is because these industries are better understood and there is less risk involved, but I wonder if the current big software companies will take similar paths in the future.
Investors often prefer stocks that are not themselves a portfolio (like AWS and Amazon retail being bundled; or your Siemens example). Lots of people would like to buy into AWS's margins without also buying Amazon retail's margins. But this is different from vertical integration that creates competitive advantage.
I've been trying to grok this idea of - when does a bubble pop. Like in theory if everyone knows it's a bubble, that should cause it to pop, because people should be making their way to the exists, playing music chairs to get their money out early.
But as I try to sort of narrative the ideas behind bubbles and bursts, one thing I realize, is that I think in order for a bubble to burst, people essentially have to want it to burst(or the opposite have to want to not keep it going).
But like Bernie Madoff got caught because he couldn't keep paying dividends in his ponzi scheme, and people started withdrawing money. But in theory, even if everyone knew, if no one withdrew their money (and told the FCC) and he was able to use the current deposits to pay dividends a few years. The ponzi scheme didn't _have_ to end, the bubble didn't have to pop.
So I've been wondering, like if everyone knows AI is a bubble, what has to happen to have it collapse? Like if a price is what people are willing to pay, in order for Tesla to collapse, people have to decide they no longer want to pay $400 for Tesla shares. If they keep paying $400 for tesla shares, then it will continue to be worth $400.
So I've been trying to think, in the most simple terms, what would have to happen to have the AI bubble pop, and basically, as long as people perceive AI companies to have the biggest returns, and they don't want to move their money to another place with higher returns (similar to TSLA bulls) then the bubble won't pop.
And I guess that can keep happening as long as the economy keeps growing. And if circular deals are causing the stock market to keep rising, can they just go on like this forever?
The downside of course being, the starvation of investments in other parts of the economy, and giving up what may be better gains. It's game theory, as long as no one decides to stop playing the game, and say pull out all their money and put it into I dunno, bonds or GME, the music keeps playing?
And there we have the reason for all of these interdependent deals between all these firms, they're all hedging with each other they can keep this set of plates spinning.
Very cool and healthy for the CEO of a company investing massive amounts into a given technology to casually refer to it as a "bubble" at the same time. I guess he softens the statement a bit by calling it "an AI bubble" instead of the "the AI bubble", but it's still interesting to see everyone involved in this economic mess start to acknowledge it.
Unironically agreed that it's good for a CEO to remain relatively level headed and clear eyed.
The comparison made to the dotcom bubble is apt. It was a bubble, but that didn't mean that all the internet and e-commerce ideas were wrong, it was more a matter of investing too much too early. When the AI bubble pops or deflates, progress on AI models will continue on.
Yeah but imagine being one of the dotcom CEOs, knowing it was gonna pop, and being like "hey guys this is a bubble, also, we're pivoting to it as much as possible". The smart ones cashed out like Mark Cuban did.
I think it will pop but not in the way everyone thinks it will pop. There's plenty that's not going to go away / anywhere, but I'm sure lots of startups will fail and close their doors.
What way do you envision it popping? Nvidia has tons of investments on their books in smaller companies. If a couple of them start showing poor earnings, it could cause a death spiral for NVDA because 1) their investment just tanked, and 2) those companies are no longer buying chips from them therefore reducing revenue.
Nvidia also makes up ~7% of the S&P 500 so if their stock price falls substantially, that's a big chunk of capital just... gone for a lot of people.
The company may survive but there are still thousands of layoffs. I just wrote in another comment that in 1999 I was getting 5 recruiter calls a week. In all of 2001 I got zero.
Most bubbles occur due to excessive levels of credit offered too cheaply, resulting in a whole bunch of defaults happening at the same time. All the major AI players have borrowed money to buy GPUs and build data centers and have used Special Purpose Vehicles to do it so the debt doesn't fall on their own balance sheet, probably using a certain amount of stock as collateral. If the SPV defaults, could that trigger a big sell-off?
Every CEO is reading from the same script right now. It might be a bubble but it’s just like the internet, it’s still going to be relevant and it’s just the crap companies and grifters who will die.
73 comments
[ 2.4 ms ] story [ 69.8 ms ] threadNot immune, maybe, but pretty well off if they didn't buy in.
To me, we're clearly not peak AI exuberance. AI agents are just getting started and getting so freaking good. Just the other day, I used Vercel's v0 to build a small business website for a relative in 10 minutes. It looked fantastic and very mobile friendly. I fed the website to ChatGPT5.1 and asked it to improve the marketing text. I fed those improvements back to v0. Finished in 15 minutes. Would have taken me at least one week in the past to do a design, code it, test it for desktop/mobile, write the copy.
The way AI has disrupted software building in 3 short years is astonishing. Yes, code is uniquely great for LLM training due to open source code and documentation but as other industries catch up on LLM training, they will change profoundly too.
Startups and other unprofitable companies however...
But unlike 08 crisis, we're getting a heads up to bring out the lube.
Oracle will likely fail. It funded its AI pivot with debt. The Debt-to-Revenue ratio is 1.77, the Debt-to-Equity ratio D/E is 520, and it has a free cash flow problem.
OpenAI, Anthropic, and others will be bought for cents on the dollar.
And just like 23andme, so will all that data be sold for dimes.
It's actually 520% or 5.2 - still high but 520 would be crazy.
Not sure how the situation is in Europe and Asia, but I would guess about the same.
(I know of a few companies, but they’re tiny tiny minnows compared to the big AI companies listed in the US).
Makes one think that this was the plan all along. I think they saw how SVB went down and realize that if they're reckless and irresponsible at a big enough scale they can get the government to just transfer money to them. It's almost like this is their new business model "we're selling exposure to the $XX trillion dollar bailout industry."
To pile on, there's hardly a product being developed that doesn't integrate "ai" in some way. I was trying to figure out why my brand new laptop was running slowly, and (among other things) noticed 3 different services running- microsoft copilot, microsoft 365 copilot (not the same as the first, naturally) and the laptop manufacturer's "chat" service. That same day, I had no fewer than 5 other programs all begging me to try their AI integrations.
Job boards for startups are all filled with "using AI" fluff because that's the only thing investors seem to want to put money into.
We really are all dirty here.
I guess but is it better for an investor to own 2 shares of Google or 1 share of OpenAI and 1 share of TSMC?
Like I have no doubt that being vertically integrated as a single company has lot of benefits but one can also create a trust that invests vertically as well.
Nvidia earnings tomorrow will be the litmus test if things are going to topple over.
Maybe this is because these industries are better understood and there is less risk involved, but I wonder if the current big software companies will take similar paths in the future.
But as I try to sort of narrative the ideas behind bubbles and bursts, one thing I realize, is that I think in order for a bubble to burst, people essentially have to want it to burst(or the opposite have to want to not keep it going).
But like Bernie Madoff got caught because he couldn't keep paying dividends in his ponzi scheme, and people started withdrawing money. But in theory, even if everyone knew, if no one withdrew their money (and told the FCC) and he was able to use the current deposits to pay dividends a few years. The ponzi scheme didn't _have_ to end, the bubble didn't have to pop.
So I've been wondering, like if everyone knows AI is a bubble, what has to happen to have it collapse? Like if a price is what people are willing to pay, in order for Tesla to collapse, people have to decide they no longer want to pay $400 for Tesla shares. If they keep paying $400 for tesla shares, then it will continue to be worth $400.
So I've been trying to think, in the most simple terms, what would have to happen to have the AI bubble pop, and basically, as long as people perceive AI companies to have the biggest returns, and they don't want to move their money to another place with higher returns (similar to TSLA bulls) then the bubble won't pop.
And I guess that can keep happening as long as the economy keeps growing. And if circular deals are causing the stock market to keep rising, can they just go on like this forever?
The downside of course being, the starvation of investments in other parts of the economy, and giving up what may be better gains. It's game theory, as long as no one decides to stop playing the game, and say pull out all their money and put it into I dunno, bonds or GME, the music keeps playing?
See y'all in the spring!
They can't, not firever. Bubbles pop.
The comparison made to the dotcom bubble is apt. It was a bubble, but that didn't mean that all the internet and e-commerce ideas were wrong, it was more a matter of investing too much too early. When the AI bubble pops or deflates, progress on AI models will continue on.
Nvidia also makes up ~7% of the S&P 500 so if their stock price falls substantially, that's a big chunk of capital just... gone for a lot of people.
Everyone thinks it will pop the way the dotcom bubble popped
> There's plenty that's not going to go away / anywhere, but I'm sure lots of startups will fail and close their doors.
Just like the dotcom bubble pop - IOW, just like the way everyone thinks it will pop.
I wonder who’s writing the script.