Michael Burry Filed His Last 13F The Day PLTR Hit Its All-Time High — Then He Disappeared
On November 3, 2025, Scion Asset Management filed what would be Michael Burry's final 13F with the SEC. That same day, Palantir stock hit an all-time high of $207.
Seven days later, Burry deregistered Scion as an investment adviser. The fund was done.
His swan song? 97% in options. $1.1 billion shorting AI stocks.
The Final Portfolio: 8 Positions, One Conviction
Forget diversification. Burry's Q3 2025 filing was unlike anything he'd ever filed:
- $PLTR puts: $912.1M (66% of portfolio) — betting Palantir crashes
- $NVDA puts: $186.6M (13.5%) — betting NVIDIA crashes
- $PFE calls: $152.9M (11%) — contrarian pharma bet
- $HAL calls: $61.5M (4.5%) — energy bullish
- Everything else: $68M in equity longs. An afterthought.
Two AI shorts = 79.5% of the entire fund. Six of eight positions were derivatives.
Important caveat: 13F reports put options at notional value (strike × shares), not premium paid. The actual capital deployed was "significantly less" — maybe $50-100M on PLTR puts. Still enormous for a fund managing ~$155M in client assets.
The Nuclear 90-Day Pivot
Here's the part that doesn't get enough attention.
In Q2 2025, Burry held 15 diversified equity positions worth $578 million — normal hedge fund stuff:
- UnitedHealth Group ($115M)
- Regeneron ($113M)
- Lululemon ($107M)
- Meta, Estée Lauder, JD.com, Alibaba, ASML...
Then in Q3? He sold every single one.
In 90 days, he liquidated 15 equity longs and replaced them with 6 derivatives laser-focused on one thesis: AI stocks are going to crash.
That's not rebalancing. That's a complete transformation of the fund's identity.
The $176 Billion Phantom Profits Thesis
Why PLTR and NVDA specifically?
Burry's been writing about this on his "Cassandra Unchained" Substack since shutting down Scion. The core argument:
Hyperscalers are artificially inflating profits by extending GPU depreciation schedules.
Companies like Meta, Microsoft, Amazon, and Google are accounting for their AI hardware as if it lasts 5-6 years. Burry argues GPUs become functionally obsolete in 3-4 years given how fast AI is advancing.
The math: this accounting maneuver could understate depreciation by ~$176 billion between 2026 and 2028, inflating Meta's reported profits by 21% and Oracle's by 27% by 2028.
When the catch-up hits — write-downs, impairment charges, accelerated depreciation — earnings revisions will be brutal.
PLTR at 60-70x forward earnings depends on sustained AI spending. NVDA is the direct supplier of the hardware Burry thinks is being overvalued. Both become obvious shorts under this thesis.
Did It Work?
Partially, and briefly.
PLTR dropped 21% from its November 3 ATH by November 25. NVDA fell 20% from its late October peak. Both recovered by year-end.
The NVDA puts reportedly had a $110 strike expiring December 2027. With NVDA trading at $186 at year-end, those puts were deeply out-of-the-money. The PLTR position is less clear publicly.
Burry may have been right about the direction in November. He may have been early on the full thesis. Or — and this is the uncomfortable reality — the man who called the 2008 housing crash filed his last trade as a professional money manager and then walked away before we find out if he was right.
Why This Filing Matters More Than Most
Burry has been right before when everyone thought he was crazy. He was also wrong plenty of times in between.
But the structure of this final filing is remarkable regardless of outcome:
- A professional fund manager put 97% of reported assets in derivatives
- He concentrated 79.5% in two positions
- He filed it on the exact day his biggest short hit its all-time high
- Then he shut the fund down
Whether this was timing genius, thesis conviction, or something else entirely — we probably won't know until 2027 when those NVDA puts expire.
Full analysis with complete portfolio breakdown, the Q2→Q3 comparison, and price action charts: 13F Insight — Scion Q3 2025 Deep Dive
13F Insight tracks institutional 13F filings and breaks down what the smart money is actually doing. Follow for weekly deep dives.
Top comments (0)