Leopold Aschenbrenner, founder of the AI-focused investment fund Situational Awareness, recently faced significant financial pressure after banks called in their loans, forcing the fund to sell most of its publicly traded securities to rival Citadel. Although Aschenbrenner had gained recognition for his prescient views on artificial intelligence, his heavy use of leverage ultimately led to a liquidity crunch when AI-related stocks moved in the opposite direction of his expectations.
Situational Awareness, established in 2024 by the OpenAI alumnus, had taken aggressive positions in AI and tech stocks, many funded by debt. As a result, the fund was vulnerable to short-term volatility. Over the past month, the average share price of the 29 US-listed stocks it held dropped by 21 percent, with some large holdings like chipmaker Sandisk experiencing declines of up to 45 percent. This swift downturn forced the fund to liquidate assets to meet debt obligations.
Despite these setbacks, Aschenbrenner maintains that his fund remains up approximately 80 percent for the year overall. Situational Awareness still holds unleveraged positions and maintains its stake in Anthropic, a private AI laboratory not yet public. The fund’s leadership emphasizes having learned from the experience that combining debt with unpredictable markets can create untenable risks. By reducing reliance on leverage, Situational Awareness aims to better withstand fluctuations in valuation without needing to divest core investments hastily.
Financial analysts point to the fund’s experience as a cautionary example for others in the technology and AI sectors. Infrastructure developers, particularly those investing heavily in AI data centers, are reportedly facing similar risks due to substantial debt financing. Industry forecasts estimate around $9 trillion in investments for AI-related data centers by 2030, but concerns are growing about a “temporal mismatch” between loan durations and the rapid changes expected in AI technology.
Potential challenges include chips aging before associated debts are repaid, the emergence of more efficient AI models that could reduce the need for certain data centers, and competitive innovations such as Chinese AI models like Kimi K3 that offer greater efficiency. While the consensus holds that AI adoption will expand extensively, the timing, key beneficiaries, and return profiles remain uncertain, potentially exposing lenders and investors to considerable risks.
Although major financial institutions have largely avoided investing their own balance sheets directly in the AI boom, they continue to facilitate debt financing for others, often collecting fees rather than absorbing full risk. The unfolding situation underscores the lessons for market participants considering how to finance and scale AI technologies amid rapidly shifting conditions and volatile valuations.