Michael Burry’s reputation still hangs on a single, legendary bet against subprime mortgages from nearly twenty years ago. It was a brilliant, researched move—but in the years since, he’s developed a habit of predicting crashes that never quite arrive. Now, he’s set his sights on the AI race, claiming Palantir is worth less than $50 and betting that the "hyperscalers" are cooking the books.
Burry’s "Big Short" 2.0 hinges on a technicality: amortisation. He claims tech giants are "cheating" by spreading the cost of AI chips over 5 or 6 years instead of 3, artificially inflating their profits. He’s essentially praying for a market meltdown to prove him right and collect on his puts.
But as Dr. Ankur Crawford pointed out on the Josh Brown podcast (103.13), Burry’s math misses the reality of the hardware.
Crawford’s rebuttal is simple: A chip doesn’t die just because a newer model comes out. While Burry screams "shenanigans," Crawford notes that many of these chips provide high-level compute for 8 years or more. They might start by training the world's biggest AI models, but they spend the rest of their lives efficiently handling "inference" and cloud tasks.
The Bottom Line: Burry is treating cutting-edge tech like a carton of milk with a short expiry date. Crawford sees it for what it is: a long-term utility. If a crash happens, Palantir might dip below $50 along with everything else—but it won't be because Burry "found the fraud." It’ll just be a market cycle. Between the two, Crawford’s view that "the chips are working just fine" feels a lot more grounded than Burry’s desperate search for another 2008.
c Share Investor 2026
*Using google Gemini 3 to tidy up the sentence structure and spelling.
