Chevron and Microsoft Partner to Build Natural Gas Power Plant for Texas Data Center
July 2, 2026
Chevron and Microsoft Partner to Build Natural Gas Power Plant for Texas Data Center
Chevron Corporation has entered into a strategic partnership with Microsoft to develop a dedicated natural gas-fired power facility for the tech giant’s expanding data center campus in Texas. The co-located facility, named Project Kilby, is designed to provide a dedicated electricity supply to Microsoft’s data center operations for a 20-year period, signaling a deepening collaboration between the energy and technology sectors to meet the surging power demands of cloud computing and artificial intelligence.
Announced on June 22, the project aims to support the expansion of Microsoft's data center capacity by 2 gigawatts. The power supply facility is expected to begin operations by 2028 and will gradually ramp up to a total capacity of 2.67 gigawatts over time. The initiative is projected to generate more than 6,000 construction jobs and hundreds of permanent operational roles, underscoring the significant economic impact of large-scale digital infrastructure development in the region.
Chevron is expected to announce its final investment decision on Project Kilby before the end of this year. Last year, the company confirmed it is collaborating with investment firm Engine No. 1 and electric services company GE Vernova on the project. This partnership highlights a growing trend among energy companies to secure long-term, large-scale power purchase agreements with hyperscale cloud providers, as data center electricity consumption continues to rise sharply across the United States.
By co-locating power generation directly with data center facilities, Chevron and Microsoft are aiming to bypass constraints on the traditional grid and ensure reliable, dedicated energy supply for high-density computing workloads. The move reflects the broader industry shift toward integrating energy infrastructure with digital infrastructure, particularly as AI model training and inference require increasingly stable and substantial power sources.
Source: finance.yahoo
