On May 18, 2026, Situational Awareness LP filed its Q1 2026 Form 13F.

The fund’s nominal exposure to U.S. equities and options expanded from $5.52 billion at the end of 2025 to $13.677 billion, a 148% increase in a single quarter.

What drew market attention was not the scale, but the structure: over 60% of the new nominal exposure was concentrated entirely in put options on the semiconductor sector.

What Was Done in Q1

The put options covered nine underlying assets: VanEck Semiconductor ETF (SMH), NVIDIA, Broadcom, Oracle, AMD, Micron, TSMC, ASML, and Intel.

SMH had the largest nominal put exposure at $2.04 billion, followed by NVIDIA with $1.56 billion. Micron and TSMC were both long and short in options—indicating volatility bets rather than one-sided shorting.

It should be noted that Form 13F only discloses nominal option values, not net short positions. These put positions could represent active shorting, or they could be hedging long equity positions—making intent unclear from the filing alone.

Without additional data, the full strategy remains opaque.

In terms of equities, the fund continued increasing exposure to compute infrastructure.

CoreWeave shares rose from 6.1 million to 7.18 million; IREN and Applied Digital were also increased in position.

Expansion in the mining sector was most pronounced: Bitfarms (now rebranded as Keel Infrastructure) increased from 6.9 million to 19.88 million shares; CleanSpark grew from 1.64 million to 12.28 million shares; Riot Platforms rose from 6.17 million to 11.5 million shares.

Bloom Energy reduced holdings by 3.59 million shares but still retained approximately $879 million in market value, along with 408,500 call options—indicating profit-taking, not a strategic shift.

Exit actions were concentrated in optical communications.

Lumentum and Coherent were fully exited—the former had represented 8.68% of the portfolio last quarter, now zero.

Intel’s move deserves special attention: last quarter it held ~20 million call options, which were completely liquidated this quarter, while new put positions were established.

This was not a simple closeout—it marked a complete reversal in direction, from bullish to bearish.

Where the Bottleneck Is, There the Capital Follows

The logic behind this 13F filing reflects a concrete supply-demand assessment: the constraints on AI expansion are shifting.

For the past two years, the primary bottleneck has been GPU scarcity—leading markets to continuously price in NVIDIA, HBM memory, advanced process nodes, and optical communications. During this phase, the semiconductor sector enjoyed significant valuation premiums.

But as compute clusters scale into tens or even hundreds of thousands of GPUs, new constraints are emerging.

Grid interconnection applications in the U.S. currently exceed 2 TW, with average wait times exceeding five years; transformer production capacity is limited, and new data center construction cycles span years. While chips can be ramped up indefinitely, the power, land, and construction capacity needed to sustain them cannot keep pace.

Under this framework, shorting semiconductors isn’t a bet against AI’s success—it’s a recognition that chip valuations have already priced in expectations, and value is now migrating downstream to physical infrastructure.

Purchasing puts on SMH and NVIDIA serves as a hedge against potential valuation correction in the chip space; maintaining positions in CoreWeave, mining transition plays, and Bloom Energy reflects a bet on real bottlenecks in power and colocation capacity.

CoreWeave’s own actions confirm this thesis: call options dropped from 10.81 million to 1.81 million, while common stock increased from 6.1 million to 7.18 million shares.

The directional intent remained unchanged—only the risk profile shifted from high-leverage options to lower-volatility equities, reducing portfolio volatility.

From $225M to $13.677B

The fund was founded in September 2024, with its first 13F disclosing approximately $225 million in U.S. exposure. By year-end 2025, this had grown to $5.52 billion; as of March 31, 2026, nominal exposure reached $13.677 billion.

In the first half of 2025, the fund achieved returns of around 47%, compared to just ~6% for the S&P 500—beating the index by roughly 12.5 percentage points annually.

Before launching the fund, the 24-year-old German published a 165-page report titled *Situational Awareness: The Decade Ahead*, outlining his AGI timeline and the view that power and compute infrastructure would become the dominant bottleneck. Early capital came from Nat Friedman, Daniel Gross, and Stripe co-founders Patrick and John Collison.

The significance of this quarterly report lies in translating a previously narrative-driven thesis into an actual, executable portfolio structure.

Chips are merely the entry point—the true determinant of AI expansion speed is whether electricity can be connected, data centers can be built, and grid interconnection approvals can be obtained within five years.

If this thesis holds, the keywords of AI investing over the past two years were GPUs and models; in the coming years, they may well be power, land, and construction timelines.

Original: ChainCatcher