Crusoe opens 400,000-square-foot Tulsa facility for AI infrastructure manufacturing

The expansion signals a shift toward vertical integration in the AI hardware stack to bypass global supply chain bottlenecks for power distribution equipment. By manufacturing switchgear and enclosures in-house, Crusoe reduces lead times for high-density data centers that power streaming and AI workloads. This move places pressure on traditional infrastructure providers as vertically integrated firms gain speed-to-market advantages in the race for compute capacity. For the broader ecosystem, this localized production model serves as a hedge against volatile international hardware markets. Watch for the completion of the Stargate Project in Texas to validate if this internal supply chain can meet the massive power demands of next-generation AI factories.

The Stargate Project, a $500 billion AI infrastructure initiative backed by OpenAI, SoftBank, Oracle, and MGX, has become the largest single driver of new data center construction in the United States. Crusoe is one of several companies racing to build the physical infrastructure required to meet its compute demands. In January 2025, President Trump announced the Stargate Project at the White House alongside SoftBank's Masayoshi Son and OpenAI's Sam Altman, framing it as a national effort to keep American AI leadership ahead of China. The project's first data center campus broke ground in Abilene, Texas, and Crusoe's Tulsa manufacturing expansion is explicitly tied to supplying that buildout with electrical infrastructure components.

Crusoe's vertical integration strategy mirrors a broader trend among AI infrastructure developers who are bringing critical supply chain elements in-house to avoid bottlenecks. The company has raised more than $1.4 billion in total funding, including a $600 million Series E in late 2024 that valued it at $2.8 billion, according to Crusoe's own announcement of the round led by Founders Fund. Meanwhile, competitors in the AI data center space are pursuing similar strategies. CoreWeave, which provides GPU cloud infrastructure to AI companies, completed its initial public offering in March 2025 at a $23 billion valuation, raising $1.5 billion to fund additional data center capacity. The race for power distribution equipment, cooling systems, and custom electrical components has created acute supply constraints that favor companies with in-house manufacturing.

The technical demands of next-generation AI workloads are driving these infrastructure decisions. High-density GPU clusters for large language model training now require power densities exceeding 100 kilowatts per rack, far beyond the 10 to 20 kilowatts typical of traditional data centers. Crusoe's Tulsa facilities produce medium-voltage switchgear and copper busbars specifically engineered for these loads. The company's first Tulsa facility, which opened in 2024, focused on modular data center enclosures. Crusoe co-founder and CEO Cully Cavness has described the company's approach as building "the full stack" of AI infrastructure, from energy sourcing through electrical distribution to compute deployment. Oklahoma Governor Kevin Stitt and Tulsa Mayor Monroe Nichols both attended the September 2026 opening, underscoring the state's positioning as an AI manufacturing hub. Recent industry shifts have also seen Perplexity taps Crusoe Cloud for NVIDIA GB300 powered AI inference to further scale its compute capabilities.