This Thanksgiving's real drama may be Michael Burry versus Nvidia

Michael Burry has built a >$1 billion position in bearish put options against Nvidia and Palantir while publicly campaigning via a newly launched Substack (90,000 subscribers) after deregistering his firm, arguing that Nvidia’s employee stock-based compensation and alleged slow depreciation of GPUs mask overstated economics. Nvidia pushed back in a seven-page analyst memo, disputing Burry’s $112.5 billion compensation figure (saying the relevant buyback is $91 billion) and denying accounting impropriety; nevertheless, Burry’s claims pose reputational and confidence risk to a company with a roughly $4.5 trillion market cap whose stock has risen twelvefold since early 2023, creating potential for a self-fulfilling selloff if investors are persuaded.

Analysis

Market structure: Burry’s campaign raises single-name funding and sentiment risk for NVDA and any firms whose valuation is tied to continued hypergrowth (PLTR, niche AI infra names); winners are long-volatility players, short-biased funds, and diversified semiconductor ETFs if money rotates away from a single megacap. Concentration risk means pricing power could compress quickly if orders fall — a 20–40% retracement in NVDA would likely cascade to suppliers (memory, board makers) and hyperscaler capex plans. Cross-asset: expect spikes in NVDA implied vol, higher correlations across semis, transient equity risk-off that can pull long-term yields down and push USD marginally higher as carry unwinds.

Risk assessment: Tail scenarios include (1) a Burry-led credibility cascade triggering a >30% NVDA drawdown inside 60 days, (2) an accounting/regulatory probe into stock-comp policies or depreciation practices, and (3) a forced deleveraging by dealers financing AI customers; each would meaningfully widen credit spreads and index vol. Near-term (days–weeks) risk is sentiment-driven; medium (3–6 months) depends on dealer financing disclosures and earnings; long-term (>12 months) depends on real demand for generative AI hardware vs. secondhand GPU markets. Hidden dependency: hyperscalers’ capex cadence and resale market for used GPUs are critical second-order levers.

Trade implications: Tactical hedges — small, defined-risk bearish positions on NVDA via 3–9 month put spreads (buy 20% OTM, sell 40% OTM) sized 1–3% portfolio notional — favor skewed downside protection over naked shorting. Pair trade: short NVDA vs long SOXX (or SMH) to capture de-concentration; entry if NVDA premium to SOXX >15% on trailing 30‑day performance. For NVDA holders, harvest premium by selling 1-month covered calls 10–15% OTM and buy 3‑month 10% OTM protective puts for 30–50% of position.

Key Decisions for Investors

• Establish a defined-risk bearish position on NVDA equal to 1–3% of portfolio notional via 6–9 month put spreads (buy 20% OTM, sell 40% OTM) to cap max loss while capturing downside if sentiment-driven unwind occurs; close if NVDA falls >30% or implied vol normalizes below 60%.
• If long NVDA, implement covered-call income: sell 1-month calls 10–15% OTM on 25–50% of holdings and buy 3-month puts 10% OTM on the remaining 50% to limit 1–2% daily-gamma risk; reassess after next earnings and any buyback/disclosure changes.
• Execute a relative-value pair: short NVDA notional 1% and go long SOXX (or SMH) notional 1% to reduce single-name risk while maintaining semiconductor exposure; rebalance if the NVDA/SOXX performance gap widens >15% or narrows <5%.
• Trim concentrated AI/mega-cap exposure: if NVDA >5% of portfolio, reduce to <=3% and redeploy proceeds into enterprise software (MSFT) and telecom infra/defensive semicap names like CSCO and LRCX; revisit after 1 quarter or if NVDA drawdown exceeds 30%.
• Monitor these triggers over next 30–90 days and act: Burry Substack subscriber growth >200k in 30 days, NVDA short interest rising >6% of float, or company disclosure on depreciation/buybacks materially revised — any of which should increase hedge size by 50%.