Overview
- Burry published a detailed analysis arguing that Amazon, Meta, Alphabet, Microsoft and Oracle face roughly $3 trillion in AI‑related purchase obligations, leases and guarantees, and he highlights nearly $900 billion at Alphabet and about $700 billion at Meta.
- He contends companies are overstating the useful life of expensive Nvidia GPUs, which he says could understate industry depreciation by about $176 billion from 2026–2028.
- Nvidia formally disputed Burry’s chip‑life math in a seven‑page memo that defends a four‑to‑six‑year depreciation window, and Micron told investors memory demand should stay tight through 2028.
- Burry has tightened his timeframe and moved from outright shorts into leveraged put options on key AI and semiconductor names that expire through mid‑to‑late 2027, signaling he expects a correction in the nearer term.
- His thesis rests on complex financing links among hyperscalers, chipmakers and private lenders; if AI revenue or financing weakens, it could force write‑downs, pressure credit backers and shift capital away from other parts of the economy, while Burry is simultaneously buying select non‑AI retail names as hedges.