Key Takeaways
- • Michael Burry has intensified his bearish positioning against semiconductor and AI-related equities, including short positions against Micron Technology, Nvidia, Applied Materials, and a semiconductor ETF, with reported bets totaling approximately $330 million against semiconductor sector declines of 41%.
- • Burry's core thesis centers on alleged accounting manipulation by AI hyperscalers, who he contends are understating depreciation on chips and servers, thereby inflating reported profits and masking deteriorating economics.
- • Burry argues that semiconductor technology lifecycles are substantially shorter than industry accounting assumptions, estimating actual economic useful life at 2-3 years rather than the longer periods currently used for depreciation calculations.
- • Palantir Technologies CEO Alex Karp publicly dismissed Burry's thesis as "batshit crazy," reflecting significant disagreement within the investment community regarding the sustainability of AI infrastructure valuations and semiconductor demand.
- • Burry has characterized the current environment as the "beginning of the end" for the AI trade, signaling his conviction that current valuations and growth expectations embedded in semiconductor and AI-adjacent equities are unsustainable.
Summary
Michael Burry, the contrarian investor known for his prescient bearish positioning ahead of the 2008 financial crisis, has substantially escalated his short bets against the artificial intelligence sector and semiconductor manufacturers as of July 2025. His expanded positioning targets multiple critical infrastructure players including Micron Technology, Nvidia, Applied Materials, Caterpillar, Tesla, and a broad-based semiconductor ETF, with aggregate short positions reportedly valued at approximately $330 million predicated on sector declines of 41%.
Burry's investment thesis rests on a fundamental critique of accounting practices within the AI infrastructure ecosystem. He alleges that major AI hyperscalers—the cloud computing and data center operators driving semiconductor demand—are systematically understating depreciation expenses on computing chips and server hardware. This accounting approach, he contends, artificially inflates reported profitability by extending the assumed useful life of semiconductor equipment beyond its actual economic viability. Specifically, Burry argues that semiconductor technology depreciates far more rapidly than current accounting treatments reflect, with genuine economic life spanning merely 2-3 years as technological obsolescence accelerates.
This thesis challenges a foundational assumption underpinning current AI infrastructure valuations: that substantial capital investments in chips and servers will generate returns over extended periods. If Burry's accounting critique proves correct, AI hyperscalers' returns on capital would be materially lower than currently reported, potentially triggering significant repricing across the semiconductor and AI-dependent technology sectors.
The investment establishment has not uniformly embraced Burry's perspective. Palantir Technologies CEO Alex Karp publicly characterized the short thesis as "batshit crazy," dismissing the notion that semiconductors and AI technology represent attractive short positions. This public rebuke underscores the sharp disagreement regarding AI infrastructure economics and valuations between prominent market participants.
Burry has framed his current positioning as the "beginning of the end" for the AI trade, suggesting his conviction that current market expectations and valuations embedded in semiconductor equities are unsustainable and vulnerable to significant downward repricing.
Burry's investment thesis rests on a fundamental critique of accounting practices within the AI infrastructure ecosystem. He alleges that major AI hyperscalers—the cloud computing and data center operators driving semiconductor demand—are systematically understating depreciation expenses on computing chips and server hardware. This accounting approach, he contends, artificially inflates reported profitability by extending the assumed useful life of semiconductor equipment beyond its actual economic viability. Specifically, Burry argues that semiconductor technology depreciates far more rapidly than current accounting treatments reflect, with genuine economic life spanning merely 2-3 years as technological obsolescence accelerates.
This thesis challenges a foundational assumption underpinning current AI infrastructure valuations: that substantial capital investments in chips and servers will generate returns over extended periods. If Burry's accounting critique proves correct, AI hyperscalers' returns on capital would be materially lower than currently reported, potentially triggering significant repricing across the semiconductor and AI-dependent technology sectors.
The investment establishment has not uniformly embraced Burry's perspective. Palantir Technologies CEO Alex Karp publicly characterized the short thesis as "batshit crazy," dismissing the notion that semiconductors and AI technology represent attractive short positions. This public rebuke underscores the sharp disagreement regarding AI infrastructure economics and valuations between prominent market participants.
Burry has framed his current positioning as the "beginning of the end" for the AI trade, suggesting his conviction that current market expectations and valuations embedded in semiconductor equities are unsustainable and vulnerable to significant downward repricing.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.