OpenAI Assembles AI Supply Chain

This partner map shows that OpenAI is no longer buying compute like a normal cloud customer, it is assembling its own AI supply chain across capital, power, sites, chips, and cloud operations. SoftBank is the financial lead for Stargate, OpenAI runs operations, Oracle is building and hosting capacity, Microsoft continues to provide cloud services, NVIDIA supplies the core GPU systems, and Crusoe is executing the first flagship campus in Abilene.1 2 3 4 5 6

• The split between Oracle and Microsoft matters. Oracle is the main infrastructure operator inside Stargate, while Microsoft remains a critical cloud partner outside and alongside it. That means OpenAI is widening its compute base rather than replacing one provider with another.1 3 4
• Crusoe is not just a contractor pouring concrete. It is the hands on builder behind the first Abilene campus, which went live in phases on Oracle Cloud Infrastructure, with Oracle delivering NVIDIA GB200 racks and early training and inference workloads already running there in 2025.2 5 6
• The scale is closer to utility infrastructure than normal tech capex. OpenAI said Stargate would target $500B and 10 GW over four years, then said the Oracle expansion plus Abilene brought capacity under development to more than 5 GW, with over 2 million chips planned.1 3

Going forward, frontier AI competition will hinge less on model demos and more on who can lock in land, power, financing, and chip flow years ahead. Stargate pushes OpenAI toward that model, where the winning lab increasingly looks like a coordinated industrial network, not just a software company.1 2 3 4