Bezos Calls the AI Bubble
Jeff Bezos made news today, putting his imprimatur on the idea that we’re in an AI bubble, although I think the fact that he’s now sporting a light beard was just as interesting. But that’s just me. Bezos’ declaration—at an Italian tech conference—shows that the idea of a bubble is no longer much of a debate. In fact, it’s pretty much conventional wisdom. Last week, Bloomberg reported that hedge fund manager David Einhorn had warned of the chances “that a tremendous amount of capital destruction is going to come through this cycle” of AI spending.
Our story last week about Mira Murati’s Thinking Machines Lab seems to epitomize, for some observers, the bubble-like nature of things. Bloomberg columnist Matt Levine was not the only observer to quote the story’s citation of an investor describing the “absurd pitch meeting” where Murati said, “We’re doing an AI company with the best AI people, but we can’t answer any questions.” As Bezos said today in Italy, when “people get very excited as they are today about artificial intelligence, for example, is that every experiment gets funded, every company gets funded...the good ideas and the bad ideas. Investors have a hard time in the middle of this excitement distinguishing between the good ideas and the bad ideas and so that’s also probably happening today.”
If the question is no longer whether we’re in a bubble, then the real question now is when it will burst and who will survive. On the first point, that’s very hard to say. You can imagine that the failure of one of the new AI data center firms that have cropped up could trigger a correction, given how interconnected these firms are to investors, lenders and AI firms themselves. Monday’s bankruptcy filing of First Brands Group, a heavily indebted auto parts company, has raised some alarm bells on Wall Street and is the kind of thing that—happening in AI land—would send ripples everywhere.
Who’s best positioned to survive? That would be companies like chip giant Taiwan Semiconductor Manufacturing Co. that are certainly benefiting from AI but would still have a thriving business without it. Nvidia, similarly, would surely come through a correction just fine (albeit with a much lower stock price). The worst positioned are the huge number of AI startups that lack a differentiated product or sustainable business and need to keep raising money to keep the doors open. Consider that aside from OpenAI and Anthropic, only about a dozen AI startups developing models or applications generate more than $100 million in annualized revenue, we reported in August. The list of startups that don’t reach that threshold is long.