Crusoe Raises $1.4 Billion to Build the Power Grid for AI

Crusoe Energy Systems just raised roughly $1.38 to $1.4 billion in a late-October funding round led by Mubadala Capital and Valor Equity Partners. The round included participation from Nvidia, Founders Fund, Fidelity, and Salesforce Ventures, bringing Crusoe’s total equity funding since 2018 to about $3.9 billion. Crusoe’s chief executive, Chase Lochmiller, said the company is seeing “tremendous and accelerating demand for compute infrastructure.” The round’s lead investors reflect a broader shift in private capital toward physical AI infrastructure, with institutional funds seeking long-duration assets similar to utilities or renewables. Energy-first strategy Crusoe’s model began by converting stranded or flared natural gas into electricity used for compute workloads. By 2023 the company had pivoted away from cryptocurrency mining toward AI and high-performance computing, rebranding its platform as Crusoe Cloud. Crusoe’s energy-first vertical integration means it manages the entire stack: generation, construction, cooling, GPUs and cloud orchestration. The company sources power through an “all-of-the-above” mix: solar, wind, hydro, geothermal and gas, often pairing on-site turbines with battery storage. In Iceland, Crusoe operates on 100 percent geothermal power via partner atNorth, expanding capacity there in 2025. Crusoe reports a 45 GW development pipeline, equivalent to “eight to ten New York Cities of power.” A flagship example is the 1.2 GW campus in Abilene, Texas, built for Oracle and OpenAI as part of the Stargate project, due mid-2026. Other builds include a 1.8 GW Wyoming campus (expandable to 10 GW) and renewable-powered sites in Iceland . Infrastructure build-out and financial structure The equity raise follows a $10 billion debt package arranged with JPMorgan and Blue Owl Capital, secured by Oracle’s 15-year Abilene lease. Crusoe acquired Atero in August 2025 to enhance GPU optimization and open a Tel Aviv office. In June 2025 Crusoe partnered with AMD to integrate Instinct MI355X GPUs, followed by Crusoe Spark, a modular edge-data-centre platform. In July 2025 it agreed with GE Vernova to supply 29 aeroderivative gas turbines for flexible on-site generation. Crusoe’s new modular architecture enables rack densities exceeding 150 kW and shortens build times by about 40 percent. Power scarcity and market context AI electricity demand is rising faster than U.S. grid capacity, with regional operators warning of shortages by 2026. Lochmiller told the Financial Times that “power is very scarce right now,” noting that some announced AI facilities may never secure supply. In Armstrong County, Texas, commissioners approved a scaled-back tax-abatement plan, reducing one site from seven to four buildings. Environmental groups in Colorado have voiced concerns about data-centre energy and water use, though no violations involve Crusoe. The company is extending similar hybrid-energy campuses into Alberta, Canada, and studying sites in Norway and Finland. Competitors and relative positioning Crusoe is one of several “neocloud” providers developing AI-specific infrastructure beyond the hyperscalers. CoreWeave, valued around $65 billion after its 2025 IPO, raised more than $7 billion this year to expand GPU capacity through colocation and power contracts. Voltage Park, backed by Jed McCaleb, secured $500 million to rent GPUs from converted crypto facilities. Crusoe differentiates itself by controlling its own energy supply, a slower, capital-heavy path that could gain advantage as grid scarcity grows. Analysts at Morgan Stanley and UBS estimate global AI-data-centre financing needs could exceed $1.5 trillion over the next five years. Crusoe now operates more than 1 GW of capacity and employs over 1,000 people, with new sites planned through 2026 across North America and Europe. Its next stage depends on converting energy control into measurable cost and performance gains as competition in AI infrastructure shifts from GPUs to gigawatts.

Crusoe Raises $1.4 Billion to Build the Power Grid for AI

Energy-first strategy

Keep reading. It's free.

Related stories