Michael Burry Buys Put Options on Nvidia, Micron, and Palantir as "The Big Short" Investor Predicts a 1987-Style Market Crash

Michael Burry has covered short positions and bought put options on Nvidia, Micron, and Palantir, arguing that AI-driven capital spending, leverage, and crowded momentum positioning could precipitate a sharp market correction. His bear case centers on potentially overstated depreciation assumptions for AI hardware and a future memory downcycle as Chinese and new Samsung, SK Hynix, and Micron capacity raises supply. The article counters that major AI beneficiaries have real recurring demand and that hyperscalers including Alphabet, Microsoft, Amazon, Meta, and Oracle have sufficient profitability to sustain infrastructure spending, while cautioning investors against blindly copying Burry's trades.

Analysis

The investable issue is not a celebrity short thesis but whether AI infrastructure spending converts into durable inference revenue before replacement cycles pressure returns on capital. NVDA’s valuation support depends on sustained accelerator mix and customer monetization; the more vulnerable leg is the second-order supplier chain, where MU’s HBM premium can normalize rapidly if qualified supply broadens. Hyperscalers can absorb a period of weak AI ROI, but their eventual capex discipline would transmit first to merchant silicon, memory and networking rather than to their own balance sheets.

Near term (days to weeks), public disclosure of bearish options is more likely to increase single-name implied volatility than alter fundamentals; put notional, strikes, expiry and any offsetting common-stock exposure are missing, so it is not a directional positioning signal. Over 1-3 months, earnings commentary on GPU utilization, HBM contract pricing, lead times and incremental capex is the relevant catalyst set. A synchronized reduction in 2027 capex intentions from MSFT, AMZN, GOOG, META and ORCL would matter materially more than any one buyer’s pause.

The contrarian view is that the consensus bear case may be too concentrated on a demand cliff. If accelerator supply improves, customers may reallocate spend from scarce training capacity toward inference and enterprise deployment, preserving aggregate compute demand while changing winners. That favors diversified platform owners and potentially PLTR only if commercial software growth proves scalable without a matching rise in implementation intensity; otherwise PLTR is exposed to both multiple compression and labor-content skepticism over 6-18 months.

Falsification is straightforward: maintain constructive AI exposure if hyperscalers reaffirm capex while reporting improving AI revenue or utilization, MU sustains HBM pricing/mix through new capacity ramps, and NVDA’s gross-margin trajectory remains resilient. Reduce risk if two or more hyperscalers guide capex below expectations, memory pricing rolls over for consecutive contract periods, or NVDA inventory/customer concentration worsens.

Key Decisions for Investors

• Do not follow the disclosed put activity absent strike, expiry and net exposure data; treat it as a volatility watch item, not a trade signal, over the next 30 days.
• Express a defensive AI rotation over 1-3 months: long MSFT or GOOG / short MU in equal dollar risk. The pair isolates capex-monetization resilience from memory-cycle downside; exit if MU HBM pricing and gross-margin guidance improve while either platform company cuts AI capex.
• For concentrated NVDA exposure, buy 3-6 month downside put spreads rather than reduce all beta immediately. Fund only a partial hedge, targeting protection through the next hyperscaler earnings cycle; unwind if aggregate capex guidance remains intact and NVDA’s margin outlook holds.
• Avoid adding PLTR on sentiment-driven weakness until commercial growth, services mix and incremental margin demonstrate scalable software economics. A positive trigger is accelerating commercial revenue alongside stable or improving operating leverage; failure on either metric supports a relative short versus MSFT.
• Set a portfolio alert for a second consecutive month of weaker memory contract pricing or a broad hyperscaler capex reset. Either would justify increasing the MU underweight and extending hedges to NVDA/semiconductor beta via SOXX puts.