Prediction: This Artificial Intelligence (AI) Stock Could Be Michael Burry's Next Big Short

Michael Burry has taken bearish option positions (put options disclosed for Nvidia and Palantir in Q3) and publicly criticized frothy AI valuations as the S&P 500 Shiller CAPE nears 40. He highlights extreme multiples — Palantir with a P/S of 113 and P/E of 403 — and flags an accounting risk at Nvidia where GPUs have a 18–24 month useful life while hyperscalers reportedly depreciate related infrastructure over five to six years, potentially inflating earnings. Burry deregistered Scion Asset Management, launched a paid Substack to free his public commentary, and called Tesla “ridiculously overvalued” (Tesla P/S ~16) while criticizing Elon Musk’s pay package, signaling Tesla could be a future short given investor enthusiasm for its AI/robotics optionality.

Analysis

Market structure: If Burry’s thesis gains traction, short-term winners are cash-rich hyperscalers (AMZN, GOOGL, MSFT) that can stretch depreciation and negotiate lower ASPs; losers are high-P/S AI software names (PLTR) and moonshot equity stories (TSLA) whose valuations assume perpetual hypergrowth. GPU designers (NVDA, AMD) retain pricing power but face demand elasticity if capex cycles slow; TSMC and memory suppliers see lumpy order flow. Cross-assets: a tech re-rating would push flows into IG bonds and USD; implied vol across semis/AI names would spike 30–80% relative to last 12-month averages.

Risk assessment: Tail risks include regulatory action on AI/hardware accounting or SEC restatements that force inventory write-downs (high-impact, 3–12 months), or a sudden inventory glut from hyperscalers (operational, 1–3 quarters). Immediate risk (days) is a retail-driven squeeze on large shorts; short-term (weeks–months) hinge on NVDA earnings and hyperscaler disclosures; long-term (>12 months) depends on real-world monetization of robotaxi/Optimus. Hidden dependency: corporate depreciation policies mask true capex elasticity and create sync risk across customers.

Trade implications: Tactical shorts on structurally overvalued names (PLTR, TSLA) and selective long exposure to durable cash-generative cloud providers is optimal. Use duration-limited option structures to express views (3–9 months) because fundamentals may re-rate quickly around earnings, proxy votes, or macro shocks. Rotate 2–5% from high-beta AI names into defensive tech/software with recurring revenue and 20–30% lower multiples.

Key Decisions for Investors

• Establish a 2% notional bearish exposure to PLTR via a 3-month put spread: buy 1 ATM put and sell a put 20% below ATM (roll/close after next quarterly filing or if shares trade above entry +15%). Rationale: P/S of 113 unsustainable; target 30–60% downside within 3–9 months if guidance disappoints.
• Initiate a 1.5% notional tactical bearish position on TSLA using 6-month puts ~25% OTM (size small due to squeeze risk); pair with a 1.5% long in F or GM (equal notional) to express secular auto share rotation. Exit/reevaluate at Tesla proxy vote or if TSLA falls >30% or rises >20%.
• Buy a 6-month NVDA call spread (buy ATM call, sell 30% OTM) sized 1–2% notional to capture continued datacenter demand while capping premium; add another tranche only if NVDA falls ≥15% from today or after next earnings beat. Monitor hyperscaler capex commentary within 30–45 days.
• Reduce aggregate portfolio exposure to high-P/S AI/semis by 3–5% and allocate that to 3–12 month Treasury bills or IG corporates (target 0.5–2% yield pick-up) until CAPE retreats below 35 or macro growth signals stabilize. Rationale: valuation shock protection if broader tech re-rating unfolds.