Overview

Scion Asset Management's Q3 2025 13F filing revealed a complete reversal from its bullish Q2 stance: Michael Burry liquidated 100% of his prior call-option positions and rotated into a concentrated $1.38 billion portfolio built around two things — a large directional short against AI leaders Nvidia and Palantir, and an "anti-bubble" long book in defensive, tangible-asset sectors.

Key Concepts

• 13F Filings — quarterly disclosures of U.S. institutional holdings, filed with a 45-day lag. That lag means a 13F is always a historical snapshot, not a real-time position — by the time it's public, the fund may have already adjusted or closed the trade.
• Circular Financing Thesis — Burry's argument that reported AI revenue is partly a closed loop: an AI leader invests in a startup, the startup uses that same capital to buy the AI leader's own GPUs/cloud services, and the AI leader books the purchase as organic revenue — inflating growth without genuine external demand.
• Naked vs. Hedging Position — Scion's Q1 2025 Nvidia put disclosure included language noting the position "may serve to hedge long positions." The Q3 2025 filing dropped that disclaimer entirely, signaling the short is unhedged and directional rather than a portfolio hedge.

The Portfolio Pivot

• Q2 2025 — bullish, call-heavy portfolio (UnitedHealth, Regeneron, Lululemon, Meta, Estée Lauder), reflecting a "risk-on" reversal from an even earlier near-total liquidation.
• Q3 2025 — all Q2 calls sold; new portfolio concentrated in just eight positions, with Palantir and Nvidia puts alone consuming ~79.5% of the $1.38B reported value. The remainder is long: Pfizer and Halliburton calls, plus contrarian equity stakes in Molina Healthcare, SLM Corp, and a doubled-down Lululemon equity position (while exiting the Lululemon call).

Market Reaction

The filing (November 3) triggered an immediate, AI-specific sell-off on November 4: Palantir fell 8-9.5% despite reporting strong Q3 earnings and raising guidance, and Nvidia fell in sympathy. Palantir's CEO publicly attacked the thesis on CNBC, a reaction that arguably lent the short more credibility rather than less. This is distinct from a separate, unrelated market-wide weak spot in mid-August 2025 driven by a weak labor report and tariff uncertainty — the two events are often conflated but had different causes.

Counter-Arguments

• The "early" problem — a correct bubble call is only useful with correct timing; Burry's own history includes being early (and therefore wrong in practice) on prior public calls.
• The bull case — AI proponents argue the buildout reflects genuine technological transformation, not a self-referential financing loop.
• 13F staleness — by the time the filing is public, 45 days have already passed; the disclosed position may no longer reflect the fund's actual current exposure.

Key Takeaways

• A 13F shows what a fund owned as of quarter-end, not what it owns today — treat any single filing as a lagging, not real-time, signal.
• The presence or absence of specific hedge-disclaimer language in a filing can itself be informative about a position's intent.
• A public figure's concentrated bet moving markets (the "Burry Effect") is a case study in how information asymmetry and reputation, not just position size, can trigger outsized price reactions.
• Distinguishing between correlated events (a stock-specific reaction to a filing) and coincidental ones (a broader macro-driven sell-off weeks earlier) matters for correctly attributing cause.

Related Reading

• The AI Antithesis: Deconstructing Michael Burry's $1.38 Billion Q3 2025 Pivot — full article with the complete Q2-to-Q3 portfolio breakdown and academic research review.
• Full Research Paper