Hedge fund

Situational Awareness LP — Q1 2026 13F

Aschenbrenner's fund nearly tripled to $13.7B reported value and layered an $8.5B put-option overlay across the big-cap AI semiconductor complex.

Total reported value

$13.68B

+148% vs Q4 2025 ($5.52B)

Net long market value

$3.86B

28% of reported value

Put notional

$8.46B

62% · brand new hedge book

Call notional

$1.36B

10% of reported value

Position lines

42

vs 29 in Q4 2025 · +24 new, −10 exited

§ 01

The thesis , as filed.

Narrative · investment frame

Situational Awareness LP is the San Francisco hedge fund founded by Leopold Aschenbrenner, whose 2024 essay of the same name argued that scaling laws lead to AGI by 2027 and that the binding constraint on the path is not models or talent but compute, electricity, and the physical buildout to support them. The fund’s Q1 2026 13F filing reads almost like a footnote-by-footnote translation of that argument into public-market positions.

The long book is concentrated in three layers of that thesis. First, power generation and grid hardware : Bloom Energy is the single largest equity position at $878.7M (23% of the long book), joined by a constellation of smaller energy and infrastructure picks — Babcock & Wilcox, T1 Energy, Solaris Energy, Power Solutions, ProPetro. Second, memory and high-bandwidth storage : Sandisk is the second-largest long at $724.4M. Third, the GPU-cloud and crypto-miner-pivoting-to-AI complex : CoreWeave, Core Scientific, IREN, Applied Digital, Riot, CleanSpark, Bitfarms, Bitdeer, Cipher, Hut 8, and a handful of recent-IPO infrastructure plays (WhiteFiber, SharonAI).

Compute, electricity, the buildout to support them — these are the binding constraints. The fund is positioned almost surgically along that supply chain, while shorting the names the market has already priced for the AI win. — interpretation, not direct quote

The big shift in this filing is on the short side. Q4 2025 was a pure-long book of 29 names. Q1 2026 introduces a $8.46B put-option overlay against the consensus AI-semiconductor winners: SMH (the VanEck Semiconductor ETF, the biggest single put at $2.04B), NVDA , AVGO , AMD , MU , TSM , ASML , INTC , plus a notable Oracle put. It is, in effect, an “everyone agrees AI” hedge layered on top of an “but they don’t see the second-order plays” long book.

A note on size. 13F “value” for option positions is the underlying notional at quarter-end, not premium paid. So the headline $13.68B figure is gross-of-options; the long-only delta-one book is closer to $3.86B, and the fund’s actual AUM is presumably smaller still after netting puts against the corresponding cash legs.

§ 02

Portfolio architecture.

Long-vs-hedge composition · sector mix

Reported value breakdown · $13.68B total

Net long market value $3.86B · 28.2% Put options notional $8.46B · 61.9% Call options notional $1.36B · 10.0%

For options, “value” is the value of the underlying shares at quarter-end — not the premium paid for the contracts. The fund’s put book covers a far larger gross notional than its long book, but cost a fraction of that to put on.

Sector mix of the long book · % of $3.86B

AI Datacenter / Crypto Pivot

$1.06B

27.6%

Power / Energy

$1.04B

27.1%

AI Cloud / GPU Hyperscale

$963.3M

25.0%

Memory/Storage

$724.4M

18.8%

Semis (Big Cap)

$59.5M

1.5%

Optical / Networking

$718K

0.0%

§ 03

§ Continue reading

Subscribe.

The rest of this report — the long book, hedge book, call book, quarter-over-quarter movers, and full position ledger — is available to members of reports.enscala.com.

$5/month
cancel anytime
or
$50/year
save 16%

No password. We'll email you a one-time link. More about subscribing →