Chevron Corporation and Microsoft have entered into a landmark 20-year agreement to supply dedicated electricity to a new artificial intelligence data center in West Texas. Under the deal, Chevron’s unit Energy Forge One LLC will construct a massive natural gas power plant with a planned capacity of 2.67 gigawatts, located on more than 2,000 acres in Reeves County within the Permian Basin.
Dubbed Project Kilby, the facility will be developed in partnership with energy company Joulent—backed by investment firm Engine No. 1—and will supply electricity directly to a Microsoft-operated data center campus in Pecos. The project is designed as a behind-the-meter solution, meaning all power is generated on-site and consumed without relying on the strained regional grid, allowing Microsoft to bypass interconnection bottlenecks that have slowed data center expansions nationwide.
Most of the generation capacity will come from large gas turbines supplied by GE Vernova, with additional units from Caterpillar’s Solar Turbines subsidiary. First power deliveries are expected in 2028, with a final investment decision anticipated by the end of 2026. The project is estimated to create around 2,000 construction jobs and generate over $10 billion in state and local tax revenues, and it will use non-potable brackish groundwater and emissions control technology to mitigate environmental impact.
“Our agreement with Microsoft through Project Kilby represents Chevron’s unique ability to deliver power to AI customers with certainty, speed and at a competitive cost,” said Jeff Gustavson, president of Chevron’s New Energies, in a statement. Microsoft President of Cloud Operations and Innovation Noelle Walsh emphasized that “AI and cloud are advancing at a pace that requires a new level of coordination between energy and infrastructure.”
The deal reflects a growing industry trend where data center operators pursue their own power generation, as AI workloads demand enormous, uninterrupted electricity. Project Kilby also positions Chevron beyond traditional oil and gas, tapping into the AI-driven energy market with expected mid-teen returns and cash flows independent of commodity price cycles.
