Leopold Aschenbrenner’s Situational Awareness LP has disclosed that it bought billions of dollars in shares of a cloud infrastructure company designed for AI workflows.
The News
The latest filing from the U.S. Securities and Exchange Commission (SEC) shows that Aschenbrenner bought 12,410,060 shares of Nebius (NBIS), valued at more than $2.581 billion at Thursday’s closing price of $208.
Aschenbrenner made his name as an investor for his early calls in Intel (INTC), Bloom Energy (BE), and SanDisk (SNDK). Intel is up a whopping 224% year-to-date, while BE and SNDK have surged 229% and 548%, respectively, over the same time frame.
The SEC filing also shows that Aschenbrenner’s Situational Awareness has a 5.6% stake in Nebius.
What It Means for Investors
Nebius has surged to a fresh all-time high in after-hours trading, rallying to as high as $239 following Aschenbrenner’s SEC disclosure.
In simple terms, Nebius builds and rents out computing infrastructure, specifically GPU-powered cloud capacity, to firms that need massive compute for AI training, tuning, inference and deployment. In December, OpenAI CEO Sam Altman said the ChatGPT creator was short on compute and that it could have generated more revenue if it had access to more compute. Meanwhile, Anthropic CEO Dario Amodei said companies are being forced to take risks due to uncertainty in access to compute.
Nebius is positioning itself to solve the compute bottleneck. But that’s not the only investment thesis for the AI infrastructure company.
In March, Nvidia announced that it would invest $2 billion in Nebius and team up with the company on AI factory design and support, AI infrastructure deployment, fleet management and inference. On May 1st, Nebius announced that it had agreed to acquire Eigen AI, a leading inference and model optimization company, in a cash-and-stock deal worth $643 million at the time.
The deals indicate that Nebius is making a strong push toward inference. In CapitalAI Daily’s analysis of Micron, we explained how inference is the next big phase of AI advancement. But in simple terms, inference enables AI models to reason and think through possible solutions to generate the best answers, as opposed to just generating answers based on internet text.
In November of 2025, Deloitte estimated that inference workloads would account for about two-thirds of all compute this year, and the market for inference-optimized chips would soar to $50 billion. While Nebius is not a chip manufacturer, it is developing a system-optimization layer that aims to extract materially better performance from hardware and deliver higher throughput and lower cost per inference without additional engineering overhead.
The system-optimization layer is critical in data center infrastructure, especially as individuals and enterprises adopt AI agents to automate workflows. Earlier this month, Goldman Sachs projected that agentic AI token consumption will witness a 24-fold increase to 120 quadrillion tokens per month by 2030. The more tokens people and businesses generate, the more compute, memory, networking and power the AI economy needs.
Nebius is positioning itself to rent out compute capacity optimized for speed and lower costs. That appears to be the investment thesis for NBIS.
Disclaimer: Opinions expressed at CapitalAI Daily are not investment advice. Investors should do their own due diligence before making any decisions involving securities, cryptocurrencies, or digital assets. Your transfers and trades are at your own risk, and any losses you may incur are your responsibility. CapitalAI Daily does not recommend the buying or selling of any assets, nor is CapitalAI Daily an investment advisor. See our Editorial Standards and Terms of Use.