Back
Microsoft, Chevron sign 20-year power deal for West Texas data centers

Microsoft, Chevron sign 20-year power deal for West Texas data centers

Chevron will build the behind-the-meter 2.67 GW Kilby gas project near Microsoft's Pecos campus, with first generation targeted in 2028 and a final investment decision expected by end-2026.

Fact Check
Reuters confirms every specific element of the claim: a 20-year natural gas-fired power agreement between Chevron and Microsoft, Project Kilby as a 2.67-GW co-located (behind-the-meter) facility at Microsoft's Pecos, Texas campus, first power/generation in 2028, and a final investment decision expected by year-end 2026. WSJ and Hart Energy independently corroborate the 2.7 GW Kilby project in West Texas. The numeric figures, locations, timelines, and parties all match precisely.
    Reference: 123
Summary

Microsoft and Chevron have signed a 20-year agreement to support the tech company's West Texas data centers through Chevron's planned Kilby natural gas project, a behind-the-meter facility next to Microsoft's Pecos campus that is expected to scale to 2.67 GW. Microsoft expects the campus to add 2 GW of data center capacity over the next five to seven years, reflecting the heavy power needs tied to AI workloads. Chevron is developing the project with Engine No. 1 and GE Vernova, while Caterpillar's Solar Turbines unit will provide additional generation capacity. The companies are targeting initial power generation in 2028 and a final investment decision before the end of 2026. Microsoft has described the Pecos investment as a multi-billion-dollar commitment expected to create more than 6,000 construction jobs and hundreds of permanent operational roles. The structure keeps generation on-site rather than routed through the regional grid, helping avoid interconnection bottlenecks that have slowed other data center and power projects.

Terms & Concepts
  • behind-the-meter: A power setup where electricity is generated and used on-site, rather than delivered through the wider grid.
  • AI workloads: Computing tasks used to train or run artificial intelligence systems, which can require large and variable amounts of power.