Michael Burry Is Betting Against These 4 AI Stocks. Is He Right?

Michael Burry is betting against Nvidia, Palantir, Tesla and Micron, citing concerns ranging from Nvidia’s chip depreciation and financing relationships to high valuations and the prospect of a downturn in memory. Palantir trades at $201, or 171 times trailing earnings and 84 times trailing sales; Tesla is down 14% this year after net income fell 60% last year and trades at a 359 P/E. Micron is up 220% year to date, while the article advises caution on its cyclical outlook; its assessment of the stocks is mixed rather than uniformly bearish.

Analysis

The common exposure is not “AI” broadly but the durability and cash conversion of AI capex. If customers’ accelerator economics disappoint, second-order pressure runs from cloud utilization and financing capacity to hardware orders; CoreWeave’s relationship with NVIDIA makes its funding and customer concentration especially relevant as a demand-quality check. This is a scenario, not evidence that reported demand is artificial. NVIDIA’s depreciation concern primarily challenges customers’ return on invested capital and replacement assumptions; it does not by itself establish an accounting problem at NVIDIA. Watch hyperscaler capex alongside utilization, useful-life disclosures, and cash generation.

The names have different clocks. PLTR is most exposed to multiple compression if growth or commercial execution merely decelerates; its valuation can fall without the business failing. MU is a cyclical supply-response thesis: capacity additions and HBM pricing matter more over 6–18 months than near-term shortage headlines. Conversely, constrained supply and firm pricing can extend the cycle. TSLA’s valuation depends on future businesses as well as auto economics; speculative product milestones are not a substitute for measurable revenue, margins, or deliveries.

Near term, a prominent short thesis can amplify volatility but does not establish timing or expected return; put buyers also face premium decay and implied-volatility risk. The contrarian risk is that Burry’s analogies compress distinct businesses into one bubble trade: AI demand may persist even if returns and winners shift. Falsifiers for the bearish case include sustained customer returns/cash conversion, resilient accelerator orders, continued memory pricing strength, and PLTR growth that supports its valuation.

Key Decisions for Investors

• Avoid treating the disclosed puts as a standalone signal. For a bearish NVIDIA view, prefer defined-risk put spreads or a smaller position; add only if hyperscaler capex or utilization weakens, or customer disclosures indicate deteriorating returns. Reassess if capex, cash generation, and accelerator demand remain durable.
• PLTR: valuation makes it a candidate for a small, defined-risk put spread rather than an outright short, timed around earnings or a growth/contracting inflection. The thesis is falsified by sustained strong commercial growth and execution sufficient to support elevated expectations; verify current valuation and options pricing before entry.
• MU: do not short solely on the shortage narrative. Track HBM and DRAM contract pricing, supplier capacity plans, and inventory; consider bearish exposure only on evidence supply is catching up and pricing rolls over. Strong pricing or delayed capacity would invalidate the timing thesis.
• TSLA: no compelling short from valuation alone. Require deterioration in deliveries, automotive economics, or cash flow before adding bearish exposure; treat robotaxi/robotics and other product claims as unproven until they produce verifiable financial contribution.