Michael Burry is getting more bearish on Nvidia, Palantir and Micron: here's why

Michael Burry replaced his outright bearish positions in Nvidia, Palantir and Micron with put options extending into 2027. He said research accelerated his timeline for an AI downturn; the article gives no position sizes or market reaction.

Analysis

The signal is more useful as a reminder that AI downside theses have different transmission channels than as evidence that a downturn is imminent. NVDA is most exposed to a slowdown in accelerator demand or customer spending; MU adds memory pricing and supply-cycle sensitivity; PLTR depends more on converting AI interest into durable customer adoption and revenue. A broad AI de-rating could therefore obscure meaningful dispersion, while a common factor unwind could pressure all three regardless of company-specific execution.

Long-dated puts can express a view with defined maximum loss, but the buyer also pays for time and volatility; a public position change does not establish its size, cost, or expected return. Near term, the main risk is sentiment-driven volatility and expensive downside protection, not confirmation of weaker fundamentals. Over 1–3 months, scrutinize company guidance and evidence of customer spending, memory pricing, and commercial adoption. Over 6–18 months, the key question is whether AI investment produces returns sufficient to sustain spending. The thesis weakens if those indicators remain resilient; it strengthens with broad spending cuts or deteriorating company outlooks. Without option pricing and valuation data, there is no basis to call the puts attractive or the shares overvalued.

Key Decisions for Investors

• Do not copy the trade on the basis of an investor’s disclosed positioning alone. Before considering any bearish options exposure, verify current implied volatility, skew, expiration, liquidity, and the position’s actual size; favor defined-risk structures over uncapped short exposure.
• Keep the three names separate in risk attribution: monitor NVDA for customer spending and demand guidance, MU for memory pricing and supply conditions, and PLTR for evidence that adoption is translating into recurring commercial business. Avoid treating a single AI sentiment move as equivalent fundamental evidence across all three.
• For an AI-heavy portfolio, review concentration and stress-test a correlated multiple contraction. If a hedge is needed, compare the cost of single-name protection with broad technology exposure; do not add puts when implied volatility makes the hedge cost unattractive.
• Reassess the bearish thesis on the next company updates and relevant demand indicators. A sustained deterioration in spending, pricing, or adoption would support further downside protection; resilient guidance and operating evidence would argue against extending the trade.