Symbolic Burst’ Of AI Bubble: Markets Could Have A 'Much Worse' Day Than Yesterday | Ed Zitron

In the fallout from the massive blue wave in the midterm elections, an even bigger tremor has been seen in the stock market this morning. Trillions of dollars have been wiped off the value of major U.S. tech companies, with Elon Musk's net worth falling by more than $40 billion. Louisa Clarence-Smith is U.S. business editor at The Times and joins us now. Louisa, I know markets have stabilised a bit in the last couple of hours, but tell us the scale of the plunge and what's happened since. Yes. Hi, Cathy. So, yesterday, it was a big shock for Wall Street. Tech stocks and as that composite index fell by more than it's ever done in one day since about March 2020 during the COVID pandemic, about $2.5 trillion of value was wiped off the index. And that has been put down to worries that this democratic wave of lawmakers coming in after the US midterms are going to slow down data centre development because they campaigned on that issue. Data centres are very unpopular. So America, as the candidate said, they're going to either pause new data centre development or introduce tighter regulations that could slow down that build out. Yeah. And just tell us a little bit more detail about the biggest losers and what's happened since, whether the stabilisation, you think it's going to continue or did we hit bottom today? Some of the biggest losers were the chip stocks yesterday, companies like NVIDIA, because if we're building fewer data centers, then there's going to be less demand for their chips, for example. So, for example, And they have stabilized a bit today, but they're still down on where they were before the elections. Mm-hmm. Despite all of that, you know, the AI trade, as it's called on Wall Street, is still very strong. There's still a lot of optimism about these companies. You know, AI is not going to go anywhere. It's very hard to put the genie back in the bottle. It's a bit like electricity. It's something we're probably going to live with. That's interesting. So it's not the end of the bubble. Tech is still the only game in town, really. It still is. I mean, still, it's the biggest position of most hedge funds, insurance companies, of people's pensions, investments. So we're all very exposed. The AI trade is still going. It certainly hasn't unraveled. If it was really unraveling, we'd have a lot further to fall. So we'd have a lot to worry about. But I think people are certainly recalibrating. Given the change in power dynamics in Washington, under Donald Trump, there's been complete optimism about AI, been very low regulations. So the prospect of more Democrats in power looking forward to the next presidential election as well, naturally, that's potentially going to change how many data centers are going to be built in the US, how quickly they're going to be built. How much money these tech companies are therefore going to spend on AI infrastructure and what that means for their growth rates going forward. Louisa, thanks very much for joining us. That was Louisa Clarence-Smith, US Business Editor at The Times. Let's hear now from Ed Zitron, who's host of the Better Offline podcast and a known critic of AI. Good evening, Ed. Good evening. How are you doing? Good. So you must be feeling sort of vindicated that, you know, this bubble seems to have burst, perhaps temporarily, though. Not yet. There is no body yet. We've not seen anything die. The thing that's really vindicating me right now is the Anthropic IPO valuation in the space of a month and a half has gone from an expected $2 trillion to now the, I think Bloomberg just reported $1.2 trillion is the expected valuation, and it could drop lower, the journal was saying. This is bad scenes, but the bubble hasn't burst quite yet. No, exactly. And, you know, there seems to be, as we were just hearing there from Louisa, that, you know, it's still the only game in town. And we look at the demand for this kind of, not just from investors, but the demand for computing power provided by data centres. It's running at least three times available capacity. So, the appetite is there, isn't it, despite the unease among the American public that perhaps sparked this rout in terms of the midterms. Well, that's not strictly true. The demand, like 70% to 80% or more of AI compute is taken up by OpenAI and Anthropic. So the amount of demand there is, is because there's only a shred left when you remove these two AI labs. So there is actually probably a compute glut coming in the next year. We're already seeing signs of people having to resell compute that they bought, thinking that they need it. And Microsoft just walked away from renewing a contract with Oracle that I think was worth billions of dollars because the chips were too old. And just to explain what you mean by compute for people who aren't as tech savvy as you are. Yes. So it's when you rent the AI GPUs that NVIDIA sells out from, say, a Microsoft or a CoreWeave or what have you to either run an AI service using inference or to train an AI model. Right. OK. I'm just about following you. I mean, to what extent do you think these midterms were not just a referendum on Donald Trump, but also a referendum on big tech in terms of the, you know, the data centers that people didn't want in their backyards, whether that's for environmental concerns or whether it was a jobs concern? What do you make of it? Well, I think people have been quite tired of the tech industry in the last two years of Trump's administration have just been a griftopia. It's the reason that everyone was building in Texas was because Texas was until. A couple of days ago, blood red and allowing people to just basically build whatever they want in whatever way they want it. But even then, Greg Abbott was moving on from it. But the big thing is, is that there were 21 different municipalities in Ohio that voted against data centers. People don't want these giant monoliths that are built explicitly to take their jobs, even though AI isn't doing that. And just, I mean, they're tired of the tech industry and they're hitting the flush on it. And I think it's great to watch. But the only thing is, though, that especially somewhere in the UK where growth is pretty insipid, people might say, we'll have some of those tech jobs, thank you. That is the only show in town. It's the only place we're going to get growth from and also boost productivity. The thing is, there has been no evidence of any productivity gains from AI in the last four years. Yes, but it's been four years and trillions of dollars. I think it's fair to start asking where it is. And the big thing is, as well, is... Those jobs for building data centers are imported. They are not local jobs. They are not from the people in the areas. They are imported, leave, and then the data centers themselves have 50 to 100 jobs each. It's kind of a scam, the whole we're going to build jobs thing. It doesn't exist with data centers. It's imported talent, leaving a skeleton crew to run it when it gets built, whenever that is. So what happens next then, in your view, with AI? I think at this point, it's a question of whether Anthropic goes public at all. But if they don't go out at anything, if they go below $1 trillion, I mean, it's kind of game over already. They're either going to have to raise less money or raise a lower valuation. And because the FT reports they might even get a junk rating, it's going to be very difficult for the company to keep funding. But OpenAI is the real question. OpenAI right now is raising around about a $1.25 trillion private valuation. This is down from even a month ago when they were raising $30 billion. at a 1.5 trillion. And a private valuation of that size makes an IPO bordering on impossible. But you know, if you're right, and the bubble has burst, and a lot of people doubt that, but, you know, they think that it's still going to be there in some capacity, maybe sort of constrained somewhat, then, you know, that looks like a sort of collapse of the American economy, doesn't it? With terrible consequences for the rest of the world. I fully agree. It's why for years I've been saying we shouldn't let it get this big. And the thing is, the actual economic effects of AI are pretty much only the data centers. The companies themselves do not make that much money. They all lose money. Their actual economic... Output is very small, and most of the money that Anthropic and OpenAI make, 80% of it, comes from AI startups that ultimately lose money and basically resell the tokens at a loss. So the real thing here that scares me is the exposure of retail investors, so regular people's money, in the NASDAQ. I think like 40-something percent of the NASDAQ's value is these AI stocks that will tank even without the bubble bursting. And I've not said it bursts yet. But even without it bursting, a symbolic burst, a market ingestion of the idea that it's burst would be enough to create a much worse day than happened yesterday. And really, there's nothing to stop it. There's no magic tricks left. Well, I wonder if there is a third way, because I was very interested to read a post on X by the OpenAI boss, Sam Altman. It was very pro-human. He was saying, and I quote it, people should not cede control of the future to super intelligence, Donald Trump called AI and that requirement should ultimately constrain AI progress. I wonder whether the industry itself, big tech, will somehow sort of accommodate some of these concerns that you've set out and that there's a more a sort of slower, more benign, more human friendly way of advancing this technology. Well, first of all, we should never trust Sam Altman. He just says whatever he thinks he has to at that given moment. Second of all, this is not priced for a slowdown. This is not priced for a smaller industry. Nvidia has said it will make nearly $700 billion next year in revenue. The only way to do that is with growth and growth and growth. Anthropic's entire valuation is based on the idea that it will make $180 billion in revenue by the end of 2028. So There is no smaller industry. Perhaps in the future, there might be some dregs of open source, but this industry and this bubble has been inflated based on the idea of rapacious, unrepentant, perpetual growth. And without that, there is no AI trade. Ed, thanks very much for joining us. Ed Zittron, host of the Better Offline podcast, which kind of does what it says on the tin.