Crusoe Raises $3.9B to Fuel Gigawatt Data Centers and Truck-In AI Factories
News Summary
Crusoe, the AI infrastructure company that began life as a cryptocurrency miner, has raised $3.9 billion in a Series F funding round, pushing its valuation to $30.9 billion. The round, announced on September 17, 2026 (Eastern Time), will fund two very different kinds of buildouts at once: sprawling gigawatt-scale data center campuses for training the largest AI models, and small, truck-transportable "Spark" units that can be deployed near almost any power source in a matter of weeks.
The Funding Round
The $3.9 billion Series F was co-led by Atreides Management, Mubadala Capital, and Valor Equity Partners, with participation from Founders Fund, GIC, Nvidia, Qatar Investment Authority, Radical Ventures, TPG, and more than two dozen other investors. The round arrives roughly ten months after Crusoe's October 2025 Series E, which raised $1.38 billion at a $10 billion valuation — meaning the company's valuation has roughly tripled in less than a year.
Three new members joined Crusoe's board as part of the round: Thomas Seifert, chief financial officer of Cloudflare; Bill Stein, partner and chief investment officer at Primary Digital Infrastructure; and JB Straubel, founder and CEO of battery recycling company Redwood Materials, a Tesla board member, and a Crusoe investor since 2021.
Founded in 2018, Crusoe originally mined cryptocurrency using otherwise-wasted natural gas flared at oil fields. As demand for AI computing power surged, the company pivoted almost entirely to building and operating data center infrastructure for AI training and inference, a shift that has made it one of the most closely watched infrastructure players in the current AI buildout.
Two Very Different Kinds of "AI Factories"
Crusoe's growth strategy rests on two complementary infrastructure formats.
The first is the massive, campus-scale data center, exemplified by its site in Abilene, Texas. Crusoe has built a 900-megawatt facility there to support Microsoft's AI infrastructure needs, and the company has said the Abilene campus is projected to reach a total capacity of 2.1 gigawatts as it continues to expand. The Abilene site has also hosted training work for OpenAI. Large campuses like this one are best suited to training frontier AI models, a process that benefits from having enormous numbers of chips clustered together in one location.
The second format is Spark, a line of modular, truck-transportable data center units that Crusoe manufactures in-house. According to CEO Chase Lochmiller, inference workloads — running an already-trained AI model to answer queries or generate content — typically require far fewer chips than training does, so concentrating that work in an Abilene-scale campus can be unnecessarily costly. Spark units are designed to be built in a factory, trucked to a site with available power, and connected within weeks rather than the years typically required for traditional data center construction. Crusoe is building a manufacturing facility in Denver that the company expects will eventually be capable of producing up to one gigawatt of Spark capacity per year. Some Spark units are already operating in Reno, Nevada, drawing power from a microgrid that combines solar panels with repurposed electric-vehicle batteries supplied through Redwood Materials, the company led by new Crusoe board member JB Straubel.
Business Model and Recent Momentum
Crusoe generates revenue in three main ways: leasing data center space to customers who bring their own GPUs, renting out GPUs that Crusoe itself owns, and selling inference compute capacity for running AI models on behalf of customers. Its customer roster includes Meta, Microsoft, Oracle, and OpenAI.
The company has reported roughly 6 gigawatts of total contracted data center capacity, of which about 1 gigawatt is currently operational, with the remainder under construction or in development. Its managed inference product, launched in late 2025, has reportedly grown from near-zero revenue at the start of 2026 to more than $100 million in contracted annual recurring revenue within a matter of months.
Crusoe also recently signed a five-year, $13 billion cloud infrastructure contract with quantitative trading firm Jane Street to supply GPU and AI computing capacity, one of the largest disclosed deals of its kind involving a financial services customer rather than an AI lab.
What Comes Next
CEO Chase Lochmiller has framed the company's strategy as controlling the AI supply chain end to end, "from electrons to tokens" — meaning everything from securing power sources to delivering finished AI outputs to customers. With the new capital, Crusoe plans to continue expanding both ends of its infrastructure portfolio: scaling up gigawatt-class training campuses like Abilene while ramping Spark production in Denver to serve smaller, more distributed inference workloads.
Crusoe's leadership has also held early discussions with major investment banks, including Goldman Sachs and Morgan Stanley, about a potential future initial public offering, though no formal timeline has been set. The scale of the new round — and the roster of sovereign wealth funds, chipmakers, and infrastructure investors behind it — underscores how central power availability and manufacturing speed have become to the broader AI industry's ability to keep pace with demand for computing capacity.