Leopold Aschenbrenner, the 25-year-old former OpenAI superalignment researcher who left the company in April 2023 and published the widely-circulated "Situational Awareness" memo predicting AGI timelines, has disclosed short positions against Nvidia and Oracle through his hedge fund's first 13F filing. The positions, filed with the SEC on May 15, represent $47 million in notional exposure split across put options and short equity stakes.
The filing shows $31 million in Nvidia exposure through January 2026 put options struck at $115, roughly 18% below current trading levels, and $16 million in Oracle short equity across 420,000 shares. Aschenbrenner's fund, launched in September 2024 with $180 million from Founders Fund and Sequoia Capital, maintains 26% of its disclosed book in these bearish positions. The Nvidia puts carry a delta of approximately -0.32, suggesting the position is structured for tail-risk protection rather than directional conviction. Oracle's short represents 8.9% of the fund's long portfolio, concentrated enough to signal thesis-driven positioning.
The short thesis appears rooted in Aschenbrenner's public writings on AI infrastructure buildout timelines and capital efficiency. His September 2024 essay "The Inevitable Correction" argued that datacenter capex would compress 40-60% by late 2025 as training efficiency improvements outpace raw compute demand, specifically citing Nvidia's H100 and H200 cycles as over-indexed to near-term hype. Oracle's cloud infrastructure division, which signed $98 billion in remaining performance obligations last quarter, draws scrutiny for its enterprise AI positioning against hyperscaler competition. The timing matters: Nvidia trades at 38x forward earnings with $53 billion in datacenter revenue last quarter, while Oracle's cloud growth decelerated to 21% year-over-year in Q4 2024, down from 29% the prior quarter.
This marks the first significant public short from an AI insider with Aschenbrenner's technical credibility and industry access. His 13F positions will now draw quarterly scrutiny from allocators watching for conviction changes, particularly given his dual reputation as both AI accelerationist and infrastructure skeptic. The derivatives positioning suggests he expects volatility before resolution, with the January 2026 expiry allowing for multiple earnings cycles and potential hardware refresh announcements.
Allocators should monitor Nvidia's October datacenter revenue guidance and Oracle's August cloud bookings disclosure for inflection signals. Aschenbrenner's fund typically discloses position changes within 48 hours of material thesis updates via investor letters, creating a real-time feedback loop between his technical analysis and market positioning. The next 13F, due August 14, will reveal whether these shorts expanded during Q2's 12% Nvidia rally or were trimmed into strength.
The short arrives as $840 billion in public AI infrastructure capex commitments enter their second year, with 73% of hyperscaler spending directed at Nvidia-architected clusters. Aschenbrenner's fund now owns the most visible bearish position against that consensus from someone who helped architect OpenAI's scaling roadmaps through GPT-4.