How Chevron Became the AI Darling of Big Oil
Key Takeaways
- •Chevron's Project Kilby is a 20-year agreement to provide Microsoft with 2.67 gigawatts of gas-fired electricity for AI data centers in Reeves County, Texas, with operations beginning in 2028.
- •The project uses self-contained, behind-the-meter power generation that bypasses grid interconnection queues, allowing progress despite Texas Governor Greg Abbott's freeze on new data center grid connections.
- •The deal involves multiple partners including Texas Pacific Land, investment firm Engine No. 1, GE Vernova turbines, and Caterpillar Solar Titan turbines, with provisions for solar and battery storage expansion.
- •Chevron is evaluating additional hyperscaler partnerships in other gas-rich regions including the Rockies, Eagle Ford Shale, and the western Midwest, viewing Project Kilby as a platform for growth.
- •Microsoft has relaxed certain climate commitments to accelerate AI infrastructure deployment, while both companies plan to collaborate on carbon capture and cleaner energy integration.

Rural Reeves County, Texas — home to roughly 4,000 households spread across 2,600 square miles of arid West Texas terrain — is about to host a colossal industrial transformation. Chevron is set to bring enough gas-fired electricity to the county to power more than 2 million homes. Rather than serving local residents, however, all of that power will be dedicated to Microsoft data centers.
The landmark agreement, known as Project Kilby, positions Chevron as the emerging AI energy leader among major oil companies. Chevron views Kilby — scheduled to begin operations in 2028 and scale up through 2031 — as the first of what could become several large-scale AI hyperscaler partnerships spanning West Texas and other gas-rich regions, including the Rockies and the Midwest, according to Jeff Gustavson, president of Chevron New Energies.
The project is named in honor of Texas Instruments engineer Jack Kilby, who invented the integrated circuit — the microchip — and the handheld calculator, both foundational technologies on the path to artificial intelligence.
"After making the announcement about Kilby [in June], everyone wants to talk to us," Gustavson told Fortune. "They now see us as a higher credibility player in this space that can actually put together everything you need on these projects, including, importantly, a customer."
By "everyone," Gustavson refers not only to hyperscaler companies but also to gas turbine manufacturers — a critical advantage given lengthy equipment order backlogs — as well as the broader ecosystem of third-party suppliers and contractors. Chevron contributes scale, project management expertise, land, natural gas resources, and more.
The 20-year power purchase agreement with Microsoft covers 2.67 gigawatts of natural gas-fired electricity, with provisions to expand using solar generation and battery storage. The deal encompasses Chevron's land and gas reserves, supplemental land and water services from Texas Pacific Land, financial backing from investment firm Engine No. 1, at least seven large GE Vernova turbines, and several smaller Caterpillar Solar Titan 350 turbines. Chevron brings a market capitalization of approximately $370 billion and recently reported its most profitable quarter on record.
"We think we can do more of these. This is a platform for growth, which we think differentiates us versus our peer set," Gustavson said. "We can be selective here because we bring a set of capabilities that really are unique. There are not many companies that can do all the things a Chevron can, and we have just proven that we can actually make this real."
Big Oil's Strategic Evolution
Although Chevron and Microsoft have maintained a cloud computing relationship for a decade, Chevron initially found itself excluded from actual AI energy deals when the boom began.
"They were very focused on renewable power only. That was all they were really talking about," Gustavson recalled. "For us, renewable power is not a core strength."
The landscape shifted approximately two years ago as speed and reliability became paramount in the global AI arms race. Electricity demand surged as tech companies raced to build out AI training and inference infrastructure, and hyperscalers recognized that the U.S. power grid — already strained by the retirement of coal plants and aging transmission lines — could not support the loads their facilities required.
"You started to have some of the concerns around consumer electricity prices and all of that," Gustavson said. "And, really, the discussion shifted to, 'Hey, we need reliable, affordable energy. Can we talk about natural gas?'"
Chevron is the second-largest oil and gas producer in the Permian Basin of West Texas — the center of U.S. energy extraction and the most productive oil and gas basin in the country — trailing only ExxonMobil. Yet the rural region has minimal power demand. The strategic logic: build data center complexes close to natural gas production sites, where abundant land is also available, eliminating the need for long-distance pipeline construction and avoiding competition with established grid users.
"West Texas we always thought was the place you would want to do this first because you have a very large natural gas resource," Gustavson said. "It's a state where it's a little easier to get things done. You have a deregulated power market. We know the stakeholders. You put all that together and the conversations really took off. We reserved equipment very early — a year and a half ago — and put together the project."
Microsoft ultimately agreed, having relaxed — though not abandoned — its climate commitments to accelerate AI infrastructure deployment.
"The rapid growth we're experiencing in AI and cloud, driven by customer demand, requires energy infrastructure that can scale quickly and reliably," said Noelle Walsh, Microsoft president of the Cloud Operations + Innovation group, in a statement. "Our agreement with Chevron helps ensure we'll have dedicated, large-scale power to support the evolution and reliability of advanced compute."
Even with Texas Governor Greg Abbott imposing a temporary freeze on new data center grid interconnections, Chevron and Microsoft can move forward because they are building behind the meter with self-contained power resources — an arrangement that bypasses grid interconnection queues that, nationwide, can delay projects by several years.
"Texas and the administration in Austin have been very clear, 'If you want to build data centers in this state, we need you to bring your own power. This cannot impact our consumers,'" Gustavson said.
When grid capacity eventually catches up, Project Kilby could connect to the Texas grid, though Gustavson indicated there is no urgency to do so in the near term.
The Competitive Landscape
Numerous large-scale AI campuses are on the drawing board across Texas. OpenAI's pioneering Stargate campus is progressing more slowly than anticipated. Other ambitious proposals — Fermi's Project Matador, Pacifico Energy's GW Ranch, and Energy Abundance's Data City, Texas campus — currently lack committed customers.
Chevron, by contrast, holds a 20-year agreement with a hyperscaler and capacity for expansion. Only ExxonMobil rivals Chevron's potential scale, but Exxon has concentrated on developing carbon capture solutions for data center campuses rather than leading projects directly. Chevron is also planning carbon capture growth.
At Chevron's Houston headquarters — 560 miles from Reeves County — chairman and CEO Mike Wirth emphasizes how the company has transformed the Permian Basin into its primary free cash flow engine, controlling costs while maximizing returns.
Wirth expressed willingness to increase natural gas production if it enables additional projects on the scale of Kilby.
"Could the Permian grow? One hundred percent," Wirth said during his July 31 earnings call. "The Permian is one of many options we have into the next decade for growth."
Beyond West Texas, Chevron's major U.S. gas resources include the Rockies region — spanning Colorado to North Dakota — where land is similarly plentiful. Gustavson indicated the company is also evaluating South Texas' Eagle Ford Shale region near the Gulf Coast, as well as the western Midwest, including Ohio, Illinois, and Indiana. All of these areas could support additional hyperscaler projects, he said, while West Texas remains the top priority.
"We'll be customer-led on that," Gustavson noted.
Chevron is also mindful of the environmental concerns associated with the AI-driven surge in gas-fired power generation. While the natural gas industry routinely emphasizes its advantages over coal, Gustavson stressed the potential for carbon capture at every gas plant, along with expanded renewable energy and, eventually, clean geothermal power.
"This is the big question. The world is faced with this challenge. We're always trying to balance reliability and affordability with ever-cleaner energy," Gustavson said, noting that AI advancements should ultimately contribute to addressing climate change.
"For Microsoft, this is still very important to them, and it's very important to us. We'll continue to work together with them to see how we can best balance that equation."
This story was originally featured on Fortune.com.