Crusoe Reportedly Raises More Than $3 Billion at a $30 Billion Valuation

Crusoe is no longer a flaring-gas experiment with a clever pitch. It is a $30 billion AI infrastructure company — at least on paper.People familiar with the matter told Bloomberg and Reuters that the Denver-based data-center builder has raised more than $3 billion in a new round that values the firm at about $30 billion on a post-money basis. That is roughly three times the $10 billion-plus mark set by its October 2025 Series E.The talks had been public since early July. The close, reported this week, matches the target investors described then. Lead names and full terms have not been formally detailed by the company. The valuation includes the new capital.The same reporting said Crusoe signed a five-year cloud contract with trading firm Jane Street worth about $13 billion. Jane Street would be Crusoe’s highest-profile cloud customer to date. Sources said that deal helped pull demand into the fundraising.Together, the round and the contract are a statement about where AI money is going: not only into models, but into power, land, buildings, and rented GPUs.From Gas Flares to GigawattsCrusoe started in 2018 converting stranded natural gas into electricity for compute, first for crypto, then for AI. The origin story still matters because it explains the company’s claim to own more of the stack than a typical “neocloud.”A neocloud sells GPU time. Crusoe wants to sell campuses: generation, land, prefabricated halls, and cloud software. It opened a factory this year to build the modules those halls are assembled from. The point of that factory is speed. AI customers do not want a five-year construction saga. They want megawatts that turn on.By June the company said it had about 4.9 gigawatts of capacity under contract and more than 40 gigawatts in its project pipeline. Those figures come from Crusoe, not an independent audit. Investors are still treating them as the backbone of the $30 billion price.Named customers in earlier reporting include Meta, Oracle, Microsoft, and OpenAI. Meta has been linked to roughly 1.6 gigawatts across sites in Childress, Texas, and Warrenton, Missouri. A large OpenAI-related cluster has been described as a flagship. Jane Street now adds a finance-world anchor that is not a hyperscaler logo.The Valuation JumpThe October 2025 Series E raised about $1.375 billion at more than $10 billion, co-led by Valor Equity Partners and Mubadala Capital. The roster around that round included Nvidia, Founders Fund, Fidelity, Salesforce Ventures, Tiger Global, T. Rowe Price, Franklin Templeton, and Blue Owl.A later employee share sale was reported around a $13 billion mark. Private-market trading services showed even hotter marks through the summer as AI infrastructure became its own asset class.A step from $10 billion to $30 billion in under a year is the kind of markup software companies used to get. Crusoe is not a software company. It pours concrete, strings power, and finances GPU halls. The multiple is a bet that contracted watts are as scarce as model weights.It is also a bet that the next public listing in this category will not be another model lab. Pre-IPO language has followed Crusoe for months. Rounds this size often sit 12 to 24 months in front of a filing. Nothing requires that timeline. The structure signals that bankers are already in the room.Why Investors Keep Writing ChecksThe AI buildout has a bottleneck problem. Chips matter. Power matters more. Permits, interconnects, transformers, and water matter most of all.Hyperscalers cannot build everything themselves fast enough. That created a class of specialists — CoreWeave, Crusoe, Nscale, Nebius, and others — that finance campuses against long contracts and then rent the capacity. Lenders have started treating GPUs and offtake agreements like collateral on a toll road.Crusoe’s pitch is that owning more of the energy and construction layer makes those contracts safer. If you control the site and the power story, you are less exposed to someone else’s delayed substation. That is the theory. Execution is the risk.A reported setback on a large Wyoming project, after a customer balked at cost and timing, is the reminder. Pipeline slides do not equal energized megawatts. One missed campus does not kill a $30 billion round. A pattern of missed campuses would.The Jane Street contract is meant to answer that doubt. A five-year, $13 billion cloud deal is not a letter of intent on a desert parcel. It is a customer paying for clusters. If the reporting holds, it is also a reason new investors could underwrite the triple in valuation.The Neocloud Arms RaceCrusoe is not raising in a vacuum. Peers have stacked equity and debt at a pace that would have looked absurd two years ago. Multi-billion-dollar facilities against GPU fleets are now normal. The competition is for turbines, land, and interconnection queues as much as for venture capital.That race inflates prices and hides fragility. Contracted capacity can slip. Customers can re-trade. A model-lab slowdown can leave halls half full. High valuations assume that AI training and inference demand stays in a straight line up.They also assume interest rates and construction costs do not eat the spread between what a customer pays per GPU-hour and what it costs to keep the lights on. Crusoe’s energy-first identity is supposed to protect that spread. It only does if the power actually shows up on schedule.What $3 Billion BuysA raise this large is not for branding. It is for steel, modules, substations, and the working capital to hold GPUs before a customer’s first invoice.The module factory is the tell. Prefab is how Crusoe claims it can outrun traditional data-center builders. Scaling that plant, and the sites it feeds in Texas, Missouri, and elsewhere, is the obvious use of proceeds.Some of the money will also sit as a war chest for the next contract cycle. In this market, the company that can say yes to 400 megawatts next quarter wins the meeting. The company that has to syndicate for six months loses it.Existing backers often roll into rounds like this. Nvidia’s presence in prior financings is not incidental. A chipmaker that wants more places to plug in GPUs has a reason to keep a builder alive and expanding.The Open QuestionsThree questions will decide whether $30 billion was a clearing price or a peak.First: how much of the 4.9 gigawatts is live versus promised. Paper watts are cheap. Energized watts are not.Second: customer concentration. Meta, Oracle, OpenAI, and now Jane Street are impressive logos. They are also a short list. One delayed campus or one renegotiated offtake moves the model.Third: path to public markets. A $30 billion private mark creates an IPO math problem. The company will need to show not only backlog but margins that look like infrastructure, not like a construction bet.Until Crusoe publishes its own confirmation — amount, investors, and whether the Jane Street figure is booked the way the sources describe — the round remains “reported.” That is still enough to reset the tape for every other private AI infrastructure shop trying to raise this autumn.The market is no longer asking if it should fund data centers. It is asking how high it will pay for the ones that can turn on.