Key Takeaways
- • Leopold Aschenbrenner, a 24-year-old former OpenAI researcher, grew an initial $225 million investment into $5.5 billion in approximately 12 months through his hedge fund Situational Awareness LP, representing a 25x return.
- • Aschenbrenner's investment thesis targets AI infrastructure bottlenecks—specifically power generation, GPU capacity, storage, and data centers—rather than traditional AI software companies, focusing on companies like Bloom Energy, CleanSpark, and Riot Platforms.
- • Bitcoin miners now control approximately 27 gigawatts of power capacity and are engaged in $90 billion in AI-related deals, positioning them as critical infrastructure providers for the emerging superintelligence race.
- • Core Scientific and other major mining operators have signed long-term contracts (12+ years) to repurpose mining hardware and facilities for AI compute workloads, creating a diversification strategy following the April 2024 Bitcoin halving event that reduced block rewards.
- • Situational Awareness LP outperformed the broader stock market by 47 percentage points in its first six months of operation, demonstrating institutional-grade returns from concentrated exposure to AI infrastructure enablers rather than direct AI application companies.
Summary
Leopold Aschenbrenner's contrarian investment strategy represents a significant shift in how institutional capital is approaching artificial intelligence infrastructure. The 24-year-old former OpenAI researcher founded Situational Awareness LP with an initial $225 million capital base and achieved a 25x multiplier—growing the fund to $5.5 billion in roughly one year. This performance substantially outpaced broader equity markets, with the fund returning 47 percentage points above market benchmarks during its first six months of operation.
Aschenbrenner's core investment thesis diverges sharply from conventional AI exposure. Rather than investing in software companies developing large language models or AI applications, he identified that all participants in the superintelligence race face identical constraints: electrical power generation, GPU procurement, data center capacity, and storage infrastructure. This "picks and shovels" approach mirrors historical Gold Rush economics—profiting from those supplying tools rather than those seeking treasure. His portfolio includes Bloom Energy (fuel cells for data centers), CleanSpark (high-performance computing facilities), and Riot Platforms (data center and crypto mining infrastructure).
Bitcoin miners have emerged as unexpectedly critical players in the AI infrastructure ecosystem. Collectively, mining operators now control approximately 27 gigawatts of installed power capacity and are engaged in approximately $90 billion of AI-related infrastructure deals. This concentration of electrical resources and computing hardware makes them attractive counterparties for AI compute requirements. Core Scientific exemplifies this diversification, having signed a 12-year contract to repurpose mining operations for AI workloads. This strategic shift gained momentum following April 2024's Bitcoin halving, which reduced mining block rewards and incentivized operators to monetize existing infrastructure through higher-margin AI compute services.
The sustainability of this thesis depends on continued capital intensity in frontier AI development. If large AI developers face genuine bottlenecks in power and compute provisioning—rather than software innovation or algorithmic breakthroughs—then infrastructure providers command structural pricing power. Aschenbrenner's early returns suggest institutional investors are increasingly credible of this constraint-based analysis, particularly given his visibility as an OpenAI researcher and public advocate for AI safety considerations.
Aschenbrenner's core investment thesis diverges sharply from conventional AI exposure. Rather than investing in software companies developing large language models or AI applications, he identified that all participants in the superintelligence race face identical constraints: electrical power generation, GPU procurement, data center capacity, and storage infrastructure. This "picks and shovels" approach mirrors historical Gold Rush economics—profiting from those supplying tools rather than those seeking treasure. His portfolio includes Bloom Energy (fuel cells for data centers), CleanSpark (high-performance computing facilities), and Riot Platforms (data center and crypto mining infrastructure).
Bitcoin miners have emerged as unexpectedly critical players in the AI infrastructure ecosystem. Collectively, mining operators now control approximately 27 gigawatts of installed power capacity and are engaged in approximately $90 billion of AI-related infrastructure deals. This concentration of electrical resources and computing hardware makes them attractive counterparties for AI compute requirements. Core Scientific exemplifies this diversification, having signed a 12-year contract to repurpose mining operations for AI workloads. This strategic shift gained momentum following April 2024's Bitcoin halving, which reduced mining block rewards and incentivized operators to monetize existing infrastructure through higher-margin AI compute services.
The sustainability of this thesis depends on continued capital intensity in frontier AI development. If large AI developers face genuine bottlenecks in power and compute provisioning—rather than software innovation or algorithmic breakthroughs—then infrastructure providers command structural pricing power. Aschenbrenner's early returns suggest institutional investors are increasingly credible of this constraint-based analysis, particularly given his visibility as an OpenAI researcher and public advocate for AI safety considerations.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.