According to CNBC, investor Michael Burry, famous for predicting the 2008 housing crash, says the artificial intelligence market bubble could unravel within about two years. He made the comments on Michael Lewis’s podcast, drawing a direct parallel to the dotcom bubble, where stock prices peaked long before capital expenditure on the underlying tech did. Burry specifically called out Palantir, arguing it does “consulting” around AI rather than working on the core tech, making its high valuation hard to justify; the stock is up 130% in 2025 and over 2,100% in three years. He warned this decline will be drawn-out because regular investors are now heavily exposed via index funds and ETFs concentrated in AI names. Burry advised selling AI-run-up holdings and instead buying healthcare stocks, a sector up only 11% in three years versus the S&P 500’s 68% gain. He also dismissed Bitcoin as a “tulip bulb” enabling criminal activity, as it traded above $92,500.
Burry’s Bubble Playbook
Here’s the thing about Burry’s warning: it’s not just a hunch. He’s pointing to a specific, historical pattern. During the dotcom mania, the stock market top happened when companies were still in the middle of their massive spending sprees on servers and fiber optics. The money was still flowing, but the investor euphoria had already burned out. He sees the same script playing out now with AI. Companies are announcing billions in capital expenditure for data centers and specialized hardware, but the stock prices of the key players might have already hit their zenith. Basically, the crowd is getting excited about the promise while the bills are still coming due. And if the real revenue from AI doesn’t materialize fast enough to justify those bills? That’s when the air comes out of the balloon.
Why This Crash Could Be Different
This is Burry’s scarier point. The dotcom crash was brutal, but it was somewhat contained to the tech sector and the investors directly betting on it. Today’s market is structurally different. So many regular people are invested passively through index funds and ETFs. And those funds are massively overweight the mega-cap tech stocks driving the AI narrative. What happens if Nvidia, Microsoft, or Meta starts to slide? It doesn’t just hit stock-pickers; it drags down the retirement accounts of millions who might not even know they’re invested in AI. That creates a slower, more grinding decline. There’s no easy exit because everyone is in the same crowded trade. How do you “protect yourself” when the entire index is the risk?
Palantir and the Consulting Critique
Burry’s singling out of Palantir is fascinating. His argument that they’re doing “consulting” cuts to the heart of a big debate in AI: who are the real builders versus the middlemen? It’s a way of saying their sky-high valuation—built on a 2,100% three-year surge—is based on services wrapped around AI, not fundamental ownership of the transformative technology itself. It’s a high-margin business, sure, but is it defensible when every other big consultancy and software firm is bolting on AI features? Burry seems to think it’s not, and that the stock should “fall drastically.” He’s essentially calling it a prime example of bubble excess, where the story has wildly outpaced the substance.
The Contrarian Bet: Healthcare
So where is Burry looking instead? The classic contrarian move: the hated sector. Healthcare stocks have lagged the S&P 500 dramatically, up just 11% over three years. They’re “out of favor,” as he says. In a market obsessed with AI growth stories, boring, cash-generative, demographic-driven healthcare businesses are being ignored. This is pure Burry—finding value where no one else is looking. It’s a bet that the AI bubble bursting will cause a rotation into stable, real-world businesses that people always need, regardless of the tech hype cycle. It’s also a much safer harbor if his dire prediction about a long U.S. stock market downturn comes true. Will he be right again? Who knows. But his logic forces you to question the euphoria everyone else is just accepting.
