Leopold Aschenbrenner’s Massive Bet on AI Infrastructure

Leopold Aschenbrenner simply took Situational Consciousness LP from $255M to $13.7B in solely six quarters. His Q1 2026 submitting dropped and one new transfer stands out as particularly sharp. Full breakdown of what he’s doing with AI infrastructure shares right here 👇

The next visitor submit comes from BitcoinMiningStock.io, a public markets intelligence platform delivering knowledge on corporations uncovered to bitcoin mining, synthetic intelligence, and crypto treasury methods. Initially revealed on Might 26, 2026, by Cindy Feng.

I used to be not conscious of Situational Consciousness LP till lately. Earlier final week, CleanSpark rallied whereas majority of bitcoin-mining friends have been buying and selling within the pink. The hypothesis on X attributed the transfer to Leopold Aschenbrenner growing his place considerably within the newest 13F. That bought me curious sufficient to learn the submitting correctly. What I discovered turned out to be extra fascinating than the CleanSpark hypothesis.

Who Leopold Aschenbrenner is, and what Situational Consciousness is

When Leopold Aschenbrenner revealed Situational Consciousness in mid-2024. The 165-page essay argued that synthetic normal intelligence was a lot nearer than the general public consensus believed, that compute and energy infrastructure can be the binding constraint, and that the buildout to assist it will be the defining funding story of the late 2020s.

He launched a hedge fund of the identical title shortly after. Six 13F filings later, Situational Consciousness LP is learn religiously by a significant slice of AI buyers. The fund will get referred to as the Cathie Wooden of AI infrastructure, although the observe report is brief, simply six quarters, and it has by no means been examined in an actual drawdown.

Do you have to copy his trades? That may be a good query. Let’s take an in depth look.

Six quarters of progress

Situational Consciousness LP first appeared on the 13F radar in late 2024 with a small $255M e-book. It has grown virtually each quarter since.

The fund crossed $1B in Q1 2025 and $5B in This fall 2025. Q1 2026 is the largest leap but, greater than doubling to $13.7B. However the lengthy inventory plus name portion of the e-book barely moved (from $5.5B to $5.2B). The complete progress in reported 13F notional is a model new $8.5B put-option e-book on the semiconductor advanced. Notably, 5 earlier quarters Aschenbrenner had no significant put publicity in any respect (only a $9M Infosys put in This fall 2025). Q1 2026 is the primary time he’s submitting an actual choices e-book.

What the Q1 2026 e-book really seems like

Listed here are the most important reported positions, color-coded by sort. Blue is lengthy frequent inventory, inexperienced is name choices, pink is put choices.

The one greatest line in all the fund is not a inventory choose, it’s a $2.04B placed on the VanEck Semi ETF (SMH). 5 of the following six traces are additionally places (NVDA, ORCL, AVGO, AMD, MU). The most important bullish place is Bloom Powerat $879M of frequent inventory, with CoreWeave and SanDisk shut behind.

A number of particulars to flag. NVDA is a model new place– $1.57B placed on 8.99 million shares with no offsetting lengthy. Oracle, Broadcom, AMD, and ASML are additionally contemporary places of roughly $0.5B to $1B every. Micron and Taiwan Semi carry three-legged constructions (put, name, and a small lengthy on the identical title), which seems much less directional and extra like long- volatility positioning. Intel went the opposite approach: This fall held a $747M bullish name on 20 million shares, Q1 holds a $159M put and a 202K-share residual lengthy. Similar title, reverse place, in 1 / 4’s time.

Studying the put overlay

That is the a part of the submitting that deserves essentially the most consideration, as a result of the pure studying might be Aschenbrenner turned bearish on the AI semi advanced. Is that the case although?

There are eleven names the place the fund owns places. On 9 of them, the put publicity dwarfs any offsetting lengthy place by anyplace from ~18x to ~3,150x.

If these places are standalone purchased places, the bearish learn might be proper. However the 13F can’t affirm it. Many particulars like whether or not the places are purchased or offered, or what strikes and expirations are concerned are lacking within the submitting. Any of these modifications the learn materially. Consequently, bearish might be a powerful risk, however not a confirmed name.

Micron and Taiwan Semi are the outliers. Each carry roughly matched places and calls alongside a small lengthy, which seems extra like long- volatility positioning than a directional view.

Intel, however, is the cleanest reversal within the submitting. This fall held a $747M name on 20.24M INTC shares; Q1 holds a $159M put plus a 202K-share lengthy stub. The construction of the e-book round Intel has clearly flipped, even when the directional intent of the brand new put line is open.

Anyway, a reminder earlier than we transfer on: 13F studies choices at notional worth of the underlying shares, not premium paid. The precise money outlay on the $8.5B put e-book needs to be a lot smaller.

The place the conviction exhibits up

The opposite aspect of the submitting, the place provides, is the place the bullish thesis nonetheless speaks clearly. The bitcoin miners are the place Aschenbrenner leaned in hardest, and the rating of the will increase is informative by itself.

The larger proportion provides sit on the smaller-cap miners, whereas the bigger names that already carried the HPC-pivot premium (IREN, Utilized Digital) bought smaller however nonetheless bought significant provides. That sample is per a catch-up learn, the place the cheaper miners shut a few of the valuation hole to the names already buying and selling on AI-data-center expectations. It isn’t a wholesale rotation, the bigger names weren’t trimmed to fund the smaller ones, however it’s a transparent vote that the AI-power and HPC-hosting story is broader than the 4 or 5 names retail tends to deal with.

Three model new lengthy positions level in the identical path: T1 Power ($44M, a US photo voltaic and battery integrator), SharonAI ($18M, Neocloud supplier), and HIVE Digital ($6M). Small in absolute {dollars}, however every suits a broader AI knowledge middle and energy theme fairly than a one-off wager.

What bought lower

Eight names left the e-book solely between This fall and Q1. And there’s no single clear sample of the businesses bought lower. Consequently, any learn on motive right here is speculative.

Nonetheless, the eight exits freed up ~$1.1B of capital, which may have been used to fund a few of new positions.

Last ideas

My trustworthy learn of this submitting is two-part. Aschenbrenner remains to be lengthy the AI infrastructure: Bloom Power, SanDisk, CoreWeave, IREN, Core Scientific on the present e-book, plus meaningfully bigger crypto-miner positions (CleanSpark up 7x, Bitfarms up almost 3x, Riot and Bitdeer each near doubled), and three model new lengthy traces in T1 Power, SharonAI, and HIVE Digital. The bullish thesis on energy, knowledge facilities, and HPC-pivot miners is alive and arguably stronger than it was in This fall.

On the identical time, the fund has constructed a big put-option e-book with the SMH ETF, Nvidia, Oracle and Broadcom as the most important expressions. The places dwarf any offsetting longs on these names. This may be simply learn as Aschenbrenner is bearish on the AI semi advanced. Nonetheless, the 13F doesn’t really affirm directional intent. The submitting doesn’t disclose quick places, or what strikes and expirations are concerned. A assured bearish learn requires extra data than the 13F alone offers.

For an on a regular basis investor, the sensible takeaway is two-sided. The bullish place provides (the smaller-cap miner strikes, the brand new lengthy traces, the maintained core longs) carry clear and mirror-able sign. The put e-book carries a sign too, however a softer one: Aschenbrenner has now constructed a big sufficient choices place on the chip aspect of the commerce that it materially shapes the chance profile of his e-book. Deal with it as a flag value watching fairly than a confirmed view, and positively not a short-NVDA advice.

With that each one stated, it’s value mentioning that the present 13F is dated March 31, so positions might have moved since. The following 13F (due mid-August) will inform us whether or not the put e-book grows, shrinks, or rotates. That submitting will inform us extra clues on Leopold Aschenbrenner’s up to date thesis.