Crusoe, the data center developer best known for building the sprawling Stargate campus in Abilene, Texas, for OpenAI, has raised $3.9 billion in a Series F funding round, the company confirmed this week — one of the largest private capital raises of the year and a fresh signal that investors remain willing to underwrite the physical layer of the artificial intelligence boom even as its costs mount.
The raise lands alongside a significant governance upgrade: Crusoe has added three heavy hitters to its board of directors — Cloudflare chief financial officer Thomas Seifert, former Digital Realty chief executive Bill Stein, and Redwood Materials founder and CEO JB Straubel, a Tesla co-founder and one of the most closely watched figures in the domestic energy supply chain. The appointments, first reported by Yahoo Finance, pair capital-markets and real-estate expertise with deep operating knowledge of batteries, materials and grid-scale power.
A war chest for concrete and copper
Crusoe co-founder and CEO Chase Lochmiller discussed the raise and the company's buildout strategy at Bloomberg Invest Dallas, in conversation with Bloomberg's Dina Bass, where the conversation centered less on model architectures than on the unglamorous constraints of the AI era: megawatts, substations, land, water and construction labor.
The company's pitch is that compute demand has outrun the infrastructure designed to serve it. Crusoe says the new capital will fund both very large campuses — the kind that host hyperscale AI training clusters — and a newer product line of "small modular AI factories," containerized or prefabricated data center units that can be manufactured off-site and energized faster than a conventional greenfield build.
Coverage of the round emphasized different things. CNBC's interview with Lochmiller focused on the scale of demand for AI data centers; MSN's headline framed the raise around "massive data centers and small modular AI factories"; TechSpot, meanwhile, zeroed in on the modular thesis with a blunt summation:
"This startup believes smaller, modular data centers can solve AI's infrastructure bottleneck."
The Abilene problem: costs are rising
The round arrives at a delicate moment for Stargate, the multi-gigawatt venture announced by OpenAI with partners including Oracle and SoftBank. The Information reported this week that developers of OpenAI's Stargate data center are facing higher costs — a development that cuts to the heart of the economics for every company in the supply chain, including Crusoe, which has been the most visible builder attached to the Abilene site.
"Developers of OpenAI's Stargate Data Center Face Higher Costs."
Rising costs are not unique to one project. Across the industry, contractors report that switchgear, transformers, backup generation and high-voltage interconnection have all lengthened lead times, while utilities in Texas and Virginia wrestle with queues for grid capacity. In that environment, a builder's ability to procure equipment and secure power at scale becomes as much a competitive moat as its engineering talent — one reason the infusion of $3.9 billion matters beyond the balance sheet.
Why the board matters
The three new directors map neatly onto Crusoe's three biggest risks. Seifert, a veteran of the software and semiconductor industries who serves as Cloudflare's CFO, brings public-company finance discipline at a moment when AI infrastructure companies are being scrutinized for debt-funded expansion and capital intensity. Stein spent years running Digital Realty, one of the largest data center REITs in the world, and understands the leasing, siting and land-banking mechanics that determine whether a campus is financeable. Straubel, through Redwood Materials, sits at the center of domestic battery recycling and critical-materials supply — a domain increasingly relevant as operators pair data centers with on-site generation and storage.
Taken together, the appointments read as a maturity signal: Crusoe is staffing up for a phase in which execution, procurement and grid relationships matter more than storytelling.
The modular argument
Crusoe's smaller-factory concept is a bet that not every AI workload needs a half-gigawatt campus. By standardizing design and manufacturing modules in factories, the company argues it can compress construction timelines, reduce on-site labor risk and place capacity closer to power sources — including stranded or flared gas, a legacy of Crusoe's origins in flare mitigation for oilfields.
Skeptics note that modular designs still depend on the same scarce transformers and turbines, and that hyperscalers have historically favored bespoke buildouts for their largest training clusters. The counterargument, increasingly common among infrastructure investors, is that inference — the cheaper, more distributed half of the AI workload — will drive demand for edge-adjacent, rapidly deployable capacity.
What to watch
- Whether Crusoe discloses a valuation and the debt-versus-equity mix of the $3.9 billion.
- How cost overruns at Stargate affect construction schedules in Abilene and future phases.
- Whether the modular product wins its first anchor customers outside the Stargate ecosystem.
- Grid interconnection timelines in ERCOT, where Texas' deregulated market has been both an accelerant and a source of volatility.
For now, the message from Dallas is unambiguous: the AI trade has moved from software margins to the balance sheet of the electric grid. Crusoe's $3.9 billion is a wager that whoever builds that bridge fastest — and cheapest — controls the next phase of the boom.
