Crusoe Raises $3B After Jane Street Signs $13B AI Cloud Deal
On September 3, 2026, Bloomberg reported that Crusoe, the AI cloud and data center operator, had raised over $3 billion at a roughly $30 billion valuation in a Series F co-led by Atreides Management and Valor Equity Partners, with Mubadala Capital, the asset management arm of Abu Dhabi’s sovereign wealth fund, also participating. The round closed less than a week after Bloomberg separately reported that Jane Street Group, the quantitative trading firm, had signed a five-year AI cloud contract with Crusoe valued at around $13 billion, making it the largest disclosed single-customer AI compute contract on record. Reuters confirmed the Jane Street deal and the $30 billion valuation on the same day. TechCrunch confirmed both figures and the co-lead investors.
For enterprise AI teams, the story sits at the intersection of two questions that have been building since 2024: who actually controls the compute, and who is actually buying it? The Jane Street deal provides the sharpest answer so far to both.
From Flared Gas to Frontier Compute
Chase Lochmiller and Cully Cavness founded Crusoe in 2018 to capture natural gas that oil companies were burning off as waste and convert it into power for portable data centers. The business model was initially built around Bitcoin mining, which Crusoe sold off to NYDIG in early 2025. By then, AI had made the company’s real asset obvious: the ability to site large compute clusters in places where energy is abundant and inexpensive, without waiting for grid permitting.
Crusoe now designs and operates full AI campuses rather than leasing space. Its facility in Abilene, Texas, built in partnership with OpenAI and Oracle, is the first phase of the Stargate project, the $500 billion US AI infrastructure initiative announced by the Trump administration in January 2025.
The company’s growth has been steep even by AI infrastructure standards. Crusoe raised $1.375 billion in a Series E at roughly $10 billion valuation in October 2025, with NVIDIA, Fidelity, and Founders Fund among the participants. The $30 billion figure in September 2026 is a tripling of that mark in under a year.
The Jane Street Deal Changes the Scale
The most significant fact in the Crusoe story is not the $3 billion raise. It is the $13 billion contract that preceded it.
Jane Street, headquartered in New York, is one of the world’s most profitable proprietary trading firms. It does not sell AI products. It uses AI to trade securities. The distinction matters because it reframes who the customers of AI compute infrastructure actually are.
As The Next Web reported, Jane Street has now contracted an estimated $21.5 billion in AI compute across three providers: a $6 billion cloud contract with CoreWeave (plus a $1 billion equity stake in the same company), the $13 billion contract with Crusoe, and a $1.5 billion investment in rival neocloud FluidStack. A firm that sells nothing AI-related has assembled more committed AI compute spend than most frontier AI labs.
The reason is the nature of quantitative trading. Jane Street’s pricing models, risk systems, and signal research run on the same hardware that trains large language models. At Jane Street’s scale, dedicated GPU clusters and early access to new chip generations function as a direct competitive advantage: the firm that retrains its models daily on the latest hardware generates better signals than one waiting for shared cloud capacity to free up. Locking in five-year contracts is not a speculative bet on AI. It is operational risk management.
The contract was also pledged as loan collateral before it was made public, according to Bloomberg. This pattern, where contracted revenue is used to secure the debt that buys the hardware to fulfill the contract, is now standard in AI infrastructure financing. The Crusoe deal is one of the clearest public examples of it.
| Provider | Contract or investment | Amount |
|---|---|---|
| CoreWeave | Cloud contract | $6.0B |
| CoreWeave | Equity stake | $1.0B |
| Crusoe | Cloud contract (5-year) | $13.0B |
| FluidStack | Investment | $1.5B |
| Total | $21.5B |
Jane Street’s disclosed AI compute commitments as of September 2026. Sources: Bloomberg, The Next Web, AI Market Watch.
What Non-Lab Buyers Mean for Enterprise AI Strategy
The Jane Street pattern is not yet common, but it signals where large enterprise AI compute strategy is heading.
For years, the assumption was that AI infrastructure customers were primarily the frontier labs (OpenAI, Anthropic, Google, Meta) and secondarily the hyperscalers building for those labs. The Jane Street position challenges that model in two ways.
First, it shows that compute-intensive non-AI companies, starting with quant trading but extending to pharmaceutical research, satellite imaging, and financial risk modeling, will become significant AI infrastructure buyers in their own right. The common thread is a need to run AI workloads continuously, on proprietary data, with hardware access that cannot be compromised by shared cloud queues.
Second, it shows that neoclouds such as Crusoe, CoreWeave, Nebius, and Lambda are credible enterprise infrastructure alternatives to AWS, Azure, and Google Cloud for AI-specific workloads. Synergy Research Group puts total neocloud revenue at over $25 billion for 2025 and projects growth to roughly $400 billion by 2031, a 58 percent compound annual growth rate. That forecast depends on the neocloud category continuing to attract customers at Jane Street scale, which appears to be happening.
For enterprise teams evaluating their AI compute strategy, this creates a set of practical considerations:
Contract structure matters. Five-year contracts at $13 billion are not written by procurement teams used to month-to-month cloud billing. They require legal, finance, and infrastructure operations to align on multi-year commitments. The upside is hardware priority and pricing predictability. The risk is concentration: a single counterparty failure becomes a business continuity problem.
Neocloud options are real, not experimental. CoreWeave’s IPO in March 2025 gave enterprise CFOs a publicly traded counterparty to evaluate. Crusoe’s pre-IPO structuring suggests a listing within two years. Lambda has raised $850 million. These are capitalized, contracted companies, not early-stage startups.
Energy and location are constraint layers, not footnotes. Crusoe’s founding premise, that energy abundance matters more than proximity to existing data center hubs, has proven out. Enterprises building AI infrastructure strategy in 2026 should treat power capacity as a first-order planning variable, not something their cloud vendor handles invisibly.
The IPO Signal
The Crusoe raise includes one additional signal: the company has met with Goldman Sachs, Morgan Stanley, JPMorgan, and Bank of America to discuss a near-term listing, according to Axios. The pre-IPO structuring of the Series F reinforces this reading.
If Crusoe lists, it would follow CoreWeave (March 2025) and precede the potential listings of Lambda, Nscale, and others. A cluster of neocloud IPOs gives enterprise buyers public-market visibility into the capitalization and customer concentration of the companies supplying their compute, which should improve procurement and risk assessment.
The gap between contracted revenue and delivered infrastructure capacity is the story to watch. Crusoe has commitments at a scale that requires executing new campuses faster than any single construction cycle allows. Whether the company can deploy capacity as fast as it is being contracted for will determine whether the $30 billion valuation holds in public markets.
Why This Week Is a Turning Point
The Crusoe and Jane Street story does not fit neatly into the standard AI narrative of model launches and capability benchmarks. It sits one layer below: in the physical infrastructure, energy contracts, and long-term financial commitments that make any model launch possible.
The largest disclosed AI cloud contract in history was signed not by an AI lab or a hyperscaler, but by a trading firm. The company that received it was founded on flared gas and now builds infrastructure for OpenAI’s most ambitious compute projects. Both facts are signals that the AI infrastructure market is maturing into something enterprise strategy teams need to plan around, not just observe.
For enterprises working through their own AI compute and vendor strategy, the Crusoe raise clarifies two things. The compute bottleneck is real and will persist long enough for multi-year contracts to be rational. And the set of credible suppliers is larger than the three major hyperscalers. That combination gives enterprise buyers more leverage and more complexity than they had twelve months ago.
Enera works with enterprise teams building AI-native operations at scale. If you are working through your AI infrastructure and vendor strategy, our team is available to help. For context on how leading enterprises are thinking about compute, vendor risk, and the infrastructure layer under their AI workloads, see our earlier coverage of Jane Street’s Etched investment and the transformer ASIC race and our primer on what it means to be AI-native versus AI-aware.