How AI Pioneer Leopold Aschenbrenner Is Betting Big on the Semiconductor Future

Leopold Aschenbrenner, the former OpenAI researcher turned AI strategist, is making bold moves in the semiconductor sector through his $2 billion hedge fund, Situational Awareness. Rather than viewing the chip industry as a monolithic bet on artificial intelligence, Aschenbrenner’s strategy reveals a nuanced approach: he’s simultaneously hedging against most of the sector while aggressively backing two major players he believes will dominate the AI era.

The stakes are high. As AI continues to reshape technology landscapes, companies that fail to adapt face obsolescence, while those positioned at the intersection of AI development and national infrastructure could see exponential growth. Aschenbrenner’s recent SEC filings offer a window into this calculated strategy and what it reveals about his outlook on AI’s winners and losers.

The $2 Billion Bet: Hedging Against the Broader Chip Sector

Situational Awareness’s most striking position might be what it’s betting against. According to the fund’s latest 13F disclosure, Aschenbrenner maintained 20,441 put contracts on the VanEck Semiconductor ETF (NASDAQ: SMH), effectively shorting the entire semiconductor industry. This massive position represented 27% of the fund’s public holdings at the end of Q2.

The math behind this trade tells an interesting story. Because the ETF is cap-weighted, with Nvidia alone comprising over 20% of its portfolio, Aschenbrenner’s put position amounts to a hedge against the sector’s heavyweights. Put options give investors the right to sell securities at predetermined prices—they gain value when underlying stocks fall. In portfolio terms, this acts as insurance, protecting gains elsewhere even if chip stocks broadly decline.

But this isn’t a full capitulation on semiconductors. The hedge actually highlights where Aschenbrenner sees real value. His fund maintains substantial positions in Intel call options and Broadcom stock—together accounting for 37% of the publicly traded portfolio. This selective positioning suggests that while he’s wary of the broader industry narrative, he’s identified specific companies he believes are structurally positioned to win in the AI-driven semiconductor landscape.

Intel’s National Security Play: Why Aschenbrenner Sees Opportunity

Aschenbrenner’s bullish Intel position centers on a thesis that goes beyond typical semiconductor cycles: national security and AI superintelligence development. He established a significant call option position in the company during the first quarter, betting on price appreciation through leveraged contracts.

On the surface, Intel’s position seems precarious. GPUs and specialized chips have displaced traditional CPUs in AI data center environments, cutting into Intel’s core market. Yet Aschenbrenner’s argument carries weight. Intel remains the only U.S.-based company capable of operating a leading-edge semiconductor foundry—a capability increasingly viewed as critical infrastructure for AI development.

This thesis gained concrete validation in August when the U.S. government committed to acquiring a 9.9% stake in Intel, with additional warrant provisions for an additional 5%. The investment, coupled with Intel’s commitment to deploy $100 billion in domestic manufacturing capacity, reflects growing recognition that semiconductor sovereignty matters. The foundry business, which Intel CEO Lip-Bu Tan had previously suggested might be discontinued without major contracts, now appears strategically vital to national AI competitiveness.

Custom Silicon and Specialized AI Chips: The Broadcom Thesis

Aschenbrenner initiated his Broadcom position in Q1 and added to it in Q2, building what appears to be conviction in a different semiconductor narrative. His thesis focuses on specialization: as the decade progresses, AI infrastructure will increasingly rely on custom-designed silicon rather than off-the-shelf processors.

This plays directly to Broadcom’s strengths. The company serves as a key partner for building custom AI accelerators used by hyperscale cloud operators. Broadcom’s customer roster includes four major hyperscalers, with one significant new addition—believed to be OpenAI—bringing $10 billion in committed contracts for the coming year. Management’s guidance suggests AI chip revenues will accelerate through late 2025 and into 2026, driven by market share gains from both established and new customers.

Unlike Intel’s national security angle, Broadcom’s story is market-driven. As AI models grow larger and more specialized, cloud providers increasingly prefer tailored hardware over generic solutions. Broadcom’s position in this emerging ecosystem reflects a straightforward bet on industry structural change.

What This Strategy Signals for AI Investors

The juxtaposition of Aschenbrenner’s positions—hedging the broad industry while backing specific winners—reveals important insights about navigating AI’s infrastructure build-out. The strategy assumes that not all semiconductor plays will succeed equally. Generic exposure to the sector faces risks from overcapacity, margin compression, and shifting technology preferences. But companies solving specific problems—whether national security manufacturing capability or custom AI silicon—operate from positions of defensible advantage.

Of course, these positions were disclosed at the end of Q2. Market movements since that time have lifted both Intel and Broadcom shares, making current entry points less attractive to those looking for bargains. That’s likely another reason for the large protective put position on the broader ETF: Aschenbrenner is protecting gains in a sector where valuations have risen significantly.

For investors evaluating AI semiconductor exposure, the lesson seems clear. Rather than making a generalized bet on chip industry growth, Aschenbrenner’s approach emphasizes thesis-driven selectivity—backing companies with specific structural advantages while hedging systematic risks. Whether that discipline ultimately proves prescient will depend on whether his thesis about AI superintelligence development, national security priorities, and specialized hardware architecture actually plays out as he envisions.