The Big Short’s Michael Burry Predicts a Possible 1987-Style Drop and is Betting Against AI Giants Nvidia and Palantir. Should You Follow?

Michael Burry reiterated short positions in Nvidia, Palantir, Micron Technology and the iShares Semiconductor ETF, warning that the market rally could culminate in a 1987-style decline and calling AI-safety-related development pauses a cover for slowing growth. The article notes elevated AI-infrastructure spending, inflation, Iran-related geopolitical concerns and a Shiller CAPE ratio exceeded only during the dot-com bubble as near-term risks. However, it maintains that AI demand and related earnings growth remain strong, favoring selective long-term ownership despite potentially weaker short-term returns.

Analysis

This is primarily a positioning/volatility signal rather than new fundamental information. A high-profile short thesis can widen dispersion within AI: NVDA and MU remain exposed to any evidence that hyperscaler capex is being deferred, while PLTR is more vulnerable to multiple compression because a greater share of its valuation depends on sustaining exceptional growth well beyond the next 12 months. The relevant transmission mechanism is not an abstract “AI bubble,” but whether cloud providers convert infrastructure spend into incremental revenue fast enough to protect their own capex budgets.

Over the next days to weeks, public reiteration of bearish positioning may create reflexive downside and lift implied volatility, particularly in the semiconductor complex. That is not independently corroborated evidence of demand deterioration: the thesis is falsified near term if NVDA supply constraints persist and management commentary from hyperscalers continues to indicate accelerating AI capex. Conversely, a single major cloud customer citing power, deployment, or monetization bottlenecks could force a 1-3 month de-rating across SOXX/SMH, with MU likely offering higher beta because memory pricing is especially sensitive to inventory expectations.

The more interesting 6-18 month question is whether model-development restraint translates into lower compute intensity. Safety-related pauses could reduce frontier-training demand at the margin, but inference deployment and enterprise adoption can offset that effect; investors should not treat the two as interchangeable. Consensus may be too focused on headline GPU demand and insufficiently focused on the rate of return on deployed clusters—an issue that would first surface in hyperscaler free-cash-flow guidance, not in commentary from AI vendors.

PLTR is the cleanest valuation-risk expression but also has the highest squeeze risk if commercial growth continues to surprise upward. Prefer hedged exposure over outright broad AI shorts: index-level shorts can be expensive if earnings delivery remains intact, while the likely outcome is greater divergence between infrastructure suppliers with confirmed orders and software names priced for uninterrupted growth.

Key Decisions for Investors

• Do not initiate a blanket NVDA/MU short on this article alone; set a watch trigger around next hyperscaler earnings. A material reduction in aggregate AI capex guidance or explicit deployment delays would justify a 1-3 month long SMH puts / short SOXX position.
• For existing AI longs, buy 2-3 month SOXX put spreads rather than liquidating core exposure: this targets a capex-guidance shock while limiting carry if demand data remain firm. Roll hedges only if semiconductor implied volatility is below its post-earnings range.
• Pair trade candidate: long NVDA / short PLTR over 1-3 months, sized beta-neutral. NVDA has more observable backlog and supply constraints; PLTR has greater long-duration multiple sensitivity. Exit if PLTR commercial growth accelerates or NVDA signals order digestion.
• Monitor MU spot-memory pricing, inventory commentary, and hyperscaler free-cash-flow forecasts. A synchronized weakening in memory pricing and capex guidance is the confirmation needed for a broader semiconductor underweight; absent that, bearish narratives are likely to remain episodic volatility catalysts.