AI-native hedge funds influence AI infrastructure stocks

The signal here is not a single hedge-fund trade but a shift in who the market treats as informationally credible about AI infrastructure. A fund whose edge is modeling compute scaling and inference economics took a multibillion-dollar position in a neocloud, and the stock reacted as if the disclosure itself were a fundamental catalyst. For data and ML teams, that is a useful read on where capital expects the next bottleneck-and the next spend-to concentrate.

A SEC Schedule 13G discloses that Situational Awareness LP and affiliated reporting persons, including former OpenAI researcher Leopold Aschenbrenner and Carl Shulman, hold 12,410,060 Class A ordinary shares of Nebius Group (NBIS), about 5.6% of the company as of March 31, 2026. CNBC and Benzinga reported the shares rose more than 11% around the after-close filing on May 27, 2026; The Motley Fool valued the position at roughly $2.6 billion and noted an acquisition price near $197.73 per share. (Note: some early secondary write-ups cited a 5.5% stake; the figure in the SEC filing is 5.6%.)

Situational Awareness is positioned long the physical layer of AI build-outs, with reported exposure to competing neoclouds Iren and CoreWeave as well. The thesis is straightforward and technical: if frontier training and inference keep scaling, the binding constraints are compute capacity, power, and data-center supply, and the companies that own those assets capture durable demand. The market's quick repricing of NBIS suggests participants read a stake from an investor with that background as confirmation of the underlying compute-demand curve, not merely a bet on one ticker.

A 13G signals a passive holding, not an activist campaign, and neither Nebius nor the fund framed the position as an endorsement. The broader pattern worth tracking is informational: traditional equity signals came from sell-side research and large institutional flows, whereas this class of signal derives credibility from proximity to frontier AI research. That can amplify volatility in infrastructure names on disclosure events that carry no operating news.

Track further 13D/13G changes from AI-native funds, the price and volume response in listed AI-infrastructure names (neoclouds, accelerator vendors, power and cooling suppliers), and whether such filings increasingly serve as crosschecks on private-round narratives about compute demand. For teams inside public companies, expect investor questions to lean technical-compute economics, model-efficiency roadmaps, and scalability-rather than purely financial metrics.