Microsoft to Build 2GW AI Data Center Campus in West Texas With Onsite Gas Power
Microsoft said it will build a new data center campus in Pecos, Texas, adding about 2 gigawatts of capacity in one of the company’s largest single expansions as the race to secure power for AI infrastructure accelerates.
The multibillion-dollar project in Reeves County will be developed over the next five to seven years and is expected to support more than 6,000 construction jobs at peak build-out, along with hundreds of permanent operational roles once the campus is running, Microsoft said Monday.
The announcement underscores how the AI buildout is increasingly being shaped by access to electricity as much as access to land, chips or capital. Microsoft said the Pecos campus will be paired with a dedicated onsite energy supply, initially through a co-located natural gas power plant operating behind the meter, meaning the facility will serve the data center directly rather than drawing from the public grid at launch.
“Critically, the energy infrastructure required to power this datacenter is being funded by Microsoft,” the company said. “We are paying for the new generation and supporting infrastructure needed to serve our own operations.”
It was also reported Monday that Chevron has signed a 20-year agreement to supply power to Microsoft’s Pecos data center campus through a co-located natural gas-fired facility known as Project Kilby. The project, being developed with Engine No. 1 and GE Vernova, is expected to provide first power by 2028 and eventually scale to 2.67 GW.
The deal marks one of the clearest examples yet of a new model emerging around AI infrastructure: hyperscalers and energy companies pairing large computing campuses with dedicated power plants to avoid long grid interconnection delays and limit political backlash over ratepayer costs.
Microsoft said the Pecos campus will support demand for AI and cloud services from companies, startups, governments, healthcare providers and educational institutions. The company framed the project as both a response to immediate customer demand and a long-term bet on advanced compute infrastructure.
The project also arrives as Texas is moving to tighten oversight of large electricity users, especially data centers. Last week, the Public Utility Commission of Texas approved ERCOT’s “Batch Zero” process for large-load interconnection requests, creating a new framework to study projects of 75 MW or more together rather than one at a time. ERCOT is evaluating more than 438 GW of large-load requests, about 89% of which are tied to data centers.
That queue is far larger than the state’s current peak electricity demand, highlighting the scale of speculative and real demand facing the Texas grid. The Microsoft project’s behind-the-meter structure appears designed partly to address those concerns by bringing new generation alongside the data center rather than relying initially on existing grid capacity.
Microsoft said the gas plant and data center may eventually connect to the broader grid in coordination with utilities and local authorities. The company said the long-term plan is for the project to become part of the regional energy system and contribute to grid resilience, though it did not provide a detailed timeline for any grid connection.
The Pecos project also puts Microsoft deeper into the Permian Basin, where abundant natural gas, large tracts of land and growing interest in power-intensive computing have made West Texas a focal point for AI infrastructure proposals. The region has become attractive for developers seeking to pair data centers with gas-fired power generation, particularly as traditional grid connections become more contested and slower to secure.
Microsoft sought to emphasize local economic benefits in its announcement. Reeves County Judge Leo Hung, the county’s top elected official, said the investment would create opportunities for local businesses, support workforce development and reinforce Pecos as a place for “forward-looking companies.”
The company pointed to its existing Texas footprint near San Antonio, where it has operated data centers for nearly a decade. Microsoft said its Datacenter Academy partnerships in that region included a $545,000 investment that has reached more than 450 students, while statewide workforce programs such as TechSpark have helped create more than 1,100 jobs and engaged 20,000 Texans in digital skilling.
Microsoft also said it will extend similar workforce, local hiring and small-business support efforts to West Texas.
The company’s announcement leaned heavily on water and emissions commitments, two issues that have become central to local debates over data center development in Texas. Microsoft said the Pecos campus will use closed-loop cooling systems designed to sharply reduce water demand, requiring an initial charge of water but no additional water consumption during steady-state operation. It also said it will seek to use nonpotable water where possible.
For power generation, Microsoft said the onsite natural gas facility will include emissions controls such as Selective Catalytic Reduction systems to reduce nitrogen oxide emissions. The company also said it has already contracted 4.7 GW of renewable electricity for its Texas power use, while continuing to pursue its broader carbon and water commitments.
Still, the Pecos buildout illustrates the tension at the center of the AI infrastructure boom. Microsoft and other hyperscalers are under pressure to rapidly expand compute capacity for AI products and enterprise cloud demand, while communities and regulators are increasingly scrutinizing how much electricity and water those facilities require and who pays for supporting infrastructure.
In Texas, that scrutiny has intensified as data centers, bitcoin miners and industrial users crowd the large-load queue. The state’s latest regulatory moves show that even in one of the country’s most development-friendly power markets, large AI campuses are being asked to prove they can bring enough generation, transmission funding and operational flexibility to avoid shifting costs onto households and smaller businesses.
