Nvidia just clapped back at Michael Burry. Read the chip titan's response.

Nvidia pushed back on high-profile investor Michael Burry's criticism in a memo to analysts, disputing his claims that stock-based compensation and repurchases reduced owner earnings by 50% and that buybacks totaled $112.5B; Nvidia says it repurchased $91B of shares since 2018 and that Burry incorrectly included RSU taxes. The company also denied comparisons to historical accounting frauds, defended its employee equity practices versus peers, and said its strategic investments represent a small share of revenue and private capital activity. Burry, who closed Scion Asset Management to outside investors and launched a Substack, has continued public critiques as the broader AI trade has cooled amid valuation, circular-dealmaking and GPU depreciation concerns.

Analysis

Market structure: Nvidia’s public rebuttal to Michael Burry shores up narrative control and reduces short-term information asymmetry; its $91B repurchase since 2018 meaningfully compressed free float and increases sensitivity to sentiment-driven flows. Winners: NVDA (pricing power on H100/A100 cycles), hyperscalers and TSMC (capacity tightness); Losers: cyclical GPU renters and smaller AI plays that lack pricing leverage. Cross-asset: a meaningful NVDA correction (>10–15%) would push Nasdaq implied vols higher, tighten equity funding spreads, and likely send Treasuries tighter as risk-off rotates into duration over 48–72 hours.

Risk assessment: Tail risks include export-control escalation (China restrictions), a sudden technical obsolescence of current GPU stacks, or governance/legal actions tied to compensation accounting that could force buyback/restate disclosures; probability medium but impact high. Immediate (days): headline-driven IV spikes and borrow squeezes; short-term (weeks–months): guidance revisions and Burry’s full report; long-term (quarters–years): realization of data-center TAM and sustainable ASPs. Hidden dependencies: NVDA’s economics depend on hyperscaler procurement cadence and TSMC wafer allocation — any shift there is a second-order shock.

Trade implications: Direct: consider establishing a 1–2% long NVDA position on pullbacks of 10–15% within next 4–8 weeks, target hold 6–18 months if data-center revenue growth >30% y/y; hedge 25–50% of that notional with 3-month 5%‑delta puts. Pair trade: long NVDA (1%) vs short AMD (0.6%) to express Nvidia’s margin/architecture premium while reducing gross beta; pair rebalancing monthly. Options: sell 30-day 12–15% OTM covered calls to fund 3-month 5–7% delta puts, or buy 6–12 month call spreads (20%/40% OTM) if you prefer directional upside with limited cost.

Contrarian angles: The market may be over-weighting narrative risk from high-profile critics versus underlying economics — if NVDA’s guidance and gross margins hold, a 10–20% pullback would be an entry not a warning. Historical parallel: dominant cycle leaders (e.g., early Nvidia, MSFT in cloud) saw sharp sentiment corrections but resumed steep revenue multiple expansion when TAM growth proved durable. Unintended consequence: activist scrutiny could force greater disclosure (positive for holders) or temporarily depress buyback signaling; set hard stops at 20% below entry and reassess on next earnings or Burry publication within 60 days.

Key Decisions for Investors

• Establish a 1–2% long NVDA position on any 10–15% pullback within the next 4–8 weeks; target horizon 6–18 months provided data-center revenue growth remains >30% y/y and gross margins hold above current guidance.
• Hedge 25–50% of that NVDA exposure by buying 3-month 5%‑delta puts (tail insurance); alternatively fund hedges by selling 30-day covered calls 12–15% OTM and roll monthly if IV remains elevated.
• Implement a pair trade: long NVDA (1%) vs short AMD (0.6%) to capture relative value of architecture-led pricing power while lowering net portfolio beta; rebalance monthly and cut if NVDA falls >20% or AMD outperforms by >15% in 30 days.
• Avoid naked long exposure to high-multiple AI application names; rotate 3–6% from momentum AI software into semiconductor hardware (NVDA/TSMC/ADRs) over next 1–3 months, emphasizing companies with confirmed wafer allocations and visible backlog.