Should Investors Worry About Michael Burry's Big AI Stock Shorts

Michael Burry is short Nvidia, Micron and Palantir, but the article argues his Nvidia thesis is weakened by hyperscaler AI-infrastructure paybacks of roughly 1-3 years and continued spending. It views Micron as the most exposed short because conventional-memory supply normalization could reverse pricing, margins and earnings despite its sub-6x fiscal-2027 forward P/E. Palantir trades at a steep nearly 57x forward sales multiple, but its 85% gross margin and 62% adjusted operating margin support the view that it is a high-quality AI SaaS business rather than a consulting-like model.

Analysis

The relevant Nvidia risk is not asset life but the gap between current GPU scarcity rents and normalized inference pricing. AMZN and GOOG can earn attractive early returns while still cutting incremental purchases once utilization, token prices, or internally designed silicon improve; that would first hit NVDA's shipment growth and networking attach rate rather than installed-base revenue. Over the next 1-3 months, hyperscaler capex guidance remains the decisive catalyst; over 6-18 months, watch whether custom ASIC adoption at AMZN/GOOG shifts AI spend from merchant GPUs to lower-cost internal architectures.

MU has the least favorable asymmetry because its earnings leverage depends on two separate assumptions holding simultaneously: sustained HBM scarcity and conventional DRAM/NAND discipline. Capacity conversion can tighten legacy memory initially, but a supply response from SK Hynix and Samsung in HBM—or a weaker handset/PC recovery releasing conventional supply—would compress blended ASPs and gross margin rapidly. The key falsifier for the bearish view is multi-quarter HBM contract pricing that remains firm while conventional-memory inventory days continue falling; absent that evidence, a low forward P/E is not valuation protection because the denominator is near-cycle earnings.

PLTR's margin profile does not resolve the central valuation issue: at an extreme sales multiple, even modest deceleration in commercial growth, billings, or remaining-performance-obligation conversion can cause multiple compression independent of profitability. The more useful diligence item is whether receivables growth is accompanied by stable DSO, cash collections, and expanding multi-year commitments; deterioration would support the concern that implementation-heavy revenue is being valued as frictionless SaaS. Conversely, a sustained acceleration in commercial net retention and cash conversion would make the short thesis structurally vulnerable.

Consensus is too binary on the prominent short seller. The actionable distinction is between NVDA's near-term demand visibility, MU's cycle-risk exposure, and PLTR's duration/valuation risk: they should not trade as one "AI bubble" basket. A broad AI de-rating driven by rates would likely hurt PLTR most, whereas a capex slowdown would create greater downside beta in NVDA suppliers and memory than in software vendors with contracted government revenue.

Key Decisions for Investors

• Maintain a relative long NVDA / short MU position through the next hyperscaler earnings cycle, sized modestly: NVDA has nearer-term order visibility while MU is more exposed to a reversal in memory pricing. Reassess if AMZN or GOOG reduce AI-capex guidance, or if MU reports sustained HBM pricing and rising gross-margin guidance despite easing conventional-memory supply.
• Avoid initiating a directional PLTR short solely on valuation. Use a 3-6 month put spread only after evidence of commercial billings or cash-collection deceleration; the defined-risk structure is preferable given squeeze risk and the stock's sensitivity to AI contract announcements.
• For semiconductor exposure, favor a barbell of NVDA plus hyperscaler beneficiaries AMZN/GOOG rather than a broad long-memory allocation. AMZN and GOOG retain the option to substitute internally designed silicon, limiting their dependence on merchant GPU economics while preserving AI-service upside.
• Set a pre-earnings watchlist for MU: conventional DRAM/NAND ASP trends, inventory days, HBM supply commitments from SK Hynix/Samsung, and MU's next-quarter gross-margin bridge. A sequential ASP decline or margin guide-down is the catalyst to add to the short; without those datapoints, remain neutral rather than front-run a cycle turn.