Crusoe $30B: Jane Street Signs $13B GPU Cloud Deal

Jane Street — a quant trading firm, not an AI lab — just committed $13 billion to Crusoe's GPU cloud over five years. That deal was the primary reason Crusoe closed a $3 billion Series F at a $30 billion valuation on September 3, 2026. The companies building and buying the world's AI infrastructure are not who you expected. Crusoe, the energy-first GPU cloud that supplies computing power to OpenAI, Microsoft, and Meta, raised over $3 billion in its Series F, led by Atreides Management and Valor Equity Partners, with Mubadala Capital participating. The round valued the company at roughly $30 billion — triple its $10 billion Series E valuation from just ten months ago. That 3x re-rate in under a year is not a fluke. Crusoe is building a 1.2-gigawatt data center cluster for OpenAI, a 900-megawatt AI factory campus in Abilene, Texas, for Microsoft, and two facilities for Meta in Texas and Missouri. But the deal that moved investors most was none of those. Bloomberg confirmed on September 3 that Crusoe signed a five-year contract with Jane Street valued at roughly $13 billion. The contract covers GPU clusters and the AI infrastructure needed for training and inference, delivered via Crusoe's cloud platform. Jane Street is not a software company. It does not sell AI products. It is one of the world's largest quantitative trading firms, and it just locked in more GPU capacity than most AI labs will ever use. This is Crusoe's highest-profile customer yet, and it signals a structural shift in who drives AI infrastructure demand. This is not Jane Street's first move here. Earlier in 2026, the firm committed $6 billion to CoreWeave's cloud and invested $1 billion in CoreWeave equity. Add the Crusoe contract and other commitments, and Jane Street's total AI infrastructure spend crosses $20 billion. The firm posted $16.1 billion in trading revenue in Q1 2026 alone — double the prior year — with AI-powered strategies named as a key driver. Quant trading has always been compute-intensive, but the scale of what Jane Street is building goes well beyond market-data processing. The firm runs two distinct GPU workloads. The first handles sub-100-nanosecond latency tasks: ingesting market data and generating responses before most systems can react. The second handles large ML models for research, pricing, and pattern recognition across enormous datasets. Jane Street already operates tens of thousands of GPUs, including a liquid-cooled cluster of 4,032 GPUs in its Texas facility. It is targeting hundreds of thousands. The firm is also weighing a proprietary 200-megawatt data center — meaning the cloud contracts are a floor, not a ceiling. The throughline is competitive. In quantitative finance, the firm that trains models fastest captures the edge. GPU access is not a cost of doing business; it is the business. When a single trading firm locks $13 billion of GPU capacity into a five-year contract, that capacity exits the spot market for the duration. Crusoe is increasingly a committed-capacity provider, not a flexible cloud. The same pattern holds at CoreWeave, where long-term enterprise deals now dominate allocation. Specialized GPU clouds like Crusoe and CoreWeave still undercut hyperscalers on price — H100 rates range from roughly $1.49 to $6.98 per hour depending on provider, with AWS charging nearly five times what Lambda Labs does for equivalent silicon. But flexible capacity is tightening as enterprise contracts absorb more of it. Developers running shorter workloads or needing spot access may find that hyperscalers maintain more consistent availability despite the price premium. The GPU cloud market is bifurcating: dedicated enterprise capacity on one side, spot and reserved instances on the other. The narrative that AI labs and hyperscalers drive GPU demand is increasingly incomplete. Jane Street's $20 billion in cloud commitments puts it alongside OpenAI and Microsoft as a GPU buyer — despite selling no AI products. Financial services, healthcare, and defense are large enough to sign enterprise compute contracts, and they are starting to. Crusoe's $30 billion valuation reflects this reality. The company built its business on energy-first data centers, earned the trust of the major AI labs, and is now signing deals with entities the technology press rarely covers as buyers. The GPU landlord is scaling fast, and the tenants are not who you thought. Share