Michael Burry is back. The famed “Big Short” investor is in the spotlight for placing a colossal, contrarian bet that pits his Scion Asset Management fund directly against the most celebrated artificial intelligence companies: Palantir (PLTR) and Nvidia (NVDA). He is taking the loneliest position against the AI gold rush.
Burry’s claim is audacious: these AI darlings are dangerously overvalued. The heart of his argument boils down to a fundamental accounting disagreement. He alleges that other companies are overstating the useful life of AI chips by stretching depreciation schedules beyond the typical two to three years. Burry argues this practice artificially inflates valuations and masks looming capital risks.
He targets Palantir and Nvidia because their fortunes are tightly tied to AI infrastructure investment, making them vulnerable to a sudden retrenchment. Unlike hyperscale giants Meta Platforms (META) and Microsoft (MSFT), which have other diversified businesses, these firms are prime candidates for his depreciation thesis.
This myopic view misses the larger story. The AI chip market is not cycling through short bursts of hardware upgrades. It is rising on a secular growth boom fueled by surging enterprise and sovereign AI adoption. At its GTC developer conference in October, Nvidia announced visibility into $500 billion in future chip sales. This robust, half-trillion-dollar demand isn’t about optimistic accounting; it’s about real-world need.
Nvidia and Palantir possess durable competitive advantages that justify their lofty valuations. Nvidia’s CUDA platform is a powerful ecosystem that locks in software developers. Palantir has expanding, long-term government contracts worldwide. This is a predictable revenue stream. Burry’s narrow focus on depreciation overlooks the immense strategic value embedded in these businesses.
Moreover, the idea of accelerated chip obsolescence is countered by data center reality. Older AI chips are rarely scrapped after a couple of years. Instead, they are repurposed for inference workloads or less demanding tasks. For example, A100, H100, and newer H200 chips comfortably coexist within Microsoft’s data centers. The notion that Nvidia’s powerful processors become worthless quickly has no basis in practice.
Burry isn’t asking the most basic question: What do executives at the hyperscalers know that he doesn’t? There must be a powerful reason for their across-the-board increased spending spree. The likely answer is that AI hardware is being purchased not just for operational use, but to sustain leadership in AI innovation.
Unwittingly, Burry has positioned Scion against AI’s accelerating growth, the protective moat of CUDA, and the strategic nature of capital spending at some of the world’s best-managed firms. But that isn’t even the biggest flaw.
We know he established his large put positions before Nvidia’s bullish GTC reveal in October, as Scion’s SEC filing showed the positions at the end of September. This means his wager against AI’s winners is not just a bold contrarian call, but one that is demonstrably early—or outright wrong—on both timing and fundamentals. In a market where disruption is the norm, investors are well served to remember that stories of decline can come too soon.
For more Markman straight to you inbox every morning check out our Substack. (It’s Free)