Eric Schmidt, former Google CEO, speaks during the Collision 2022 conference at Enercare Centre in Toronto. Lukas Schulze/Sportsfile for Collision—Getty Images

Former Google CEO Eric Schmidt is getting into the AI and data center race with his new startup, and he’s betting on rural West Texas and a failed railroad turned oil giant to help him build enough power to light up 7 million homes.

Schmidt’s new Bolt Data & Energy is taking the one-stop shop approach for hyperscalers’ land, power, and water needs for their data center campuses. Bolt has teamed up with Texas Pacific Land, a little-known oil and gas player with a long history and a $20 billion market cap that happens to offer 882,000 acres of West Texas land—more acreage than Rhode Island—with easy access to natural gas and renewable energy resources. Oh, and the company just so happens to have its own water services business for oil and gas that can translate to help for thirsty data centers as well.

“Energy is the main constraint in scaling AI. If we want to keep America competitive, we have to solve this problem. Bolt was created to address this challenge,” Schmidt said in an emailed interview with Fortune. “We realized that combining my technical expertise with TPL’s unrivaled land, abundant water, and access to low-cost energy could create the infrastructure needed to meet the virtually infinite demand for compute.”

Having literally co-authored the book on AI—The Age of AI: And Our Human Future, in 2021, a year before the launch of ChatGPT—Schmidt sees the age of AI and advanced robotics as the “Fourth Industrial Revolution.” He believes data center campus developers such as Bolt are necessary to compete with China in the global AI race.

“Our platform begins with West Texas’ abundant natural gas but is designed to transition to renewable and clean energy, with nuclear power also included in future plans,” Schmidt said. “By integrating land, power generation, and data centers, we can create a scalable, resilient infrastructure capable of meeting the growing global demand for compute. Our goal is to ensure AI develops responsibly, supports American competitiveness, and delivers technology that benefits humanity while minimizing climate impact.”

Schmidt, 70, served as Google’s CEO for a decade, from 2001 to 2011, and then continued as executive chairman of Google and then Alphabet through 2017 and as technical advisor until 2020. He’s stayed plenty busy since, though. He’s also now the CEO of aerospace manufacturer Relatively Space, and cofounder of the non-profit that organizes the AI+ Expo for National Competitiveness.

Schmidt is the chairman of Bolt, and he cofounded it with Investors Todd Meister and Allan Tessler, who is a major investor in Texas Pacific Land. To date, Bolt has raised $150 million in initial capital, with TPL contributing a $50 million investment, including right of first refusal to supply critical water resources to the new data center projects.

“We felt like we wanted to capture more of the value chain than just a land lease or a water contract, so that’s why we actually invested in Bolt,” Texas Pacific Land CEO Ty Glover told Fortune. “When you’re looking at who you might want to partner with in a space that you’re not an expert in, then who better than a titan of that industry like Eric Schmidt.”

West Texas as an AI epicenter

To understand how Texas Pacific Land came by such a massive acreage holding, it helps to look back at its history of more than 150 years.

The legacy dates to 1871, when a federal charter was granted to build a national railroad from Texas to California. At the time, railroad companies received federal land grants in exchange for laying tracks.

The railroad failed for a variety of financial reasons, but it resulted in the formation of the Texas Pacific Land Trust to manage the railroad’s acreage. That acreage became quite valuable when the Texas oil boom took hold in the Permian Basin more than a century ago.

Texas Pacific has been publicly traded for almost 100 years, but it existed as a sleepy trust collecting oil and gas royalties until 2021, when an investor feud resulted in the trust converting into a much more proactive corporation.

“Coming from a failed railroad to a gorilla in the oil and gas space and now entering the AI space is exciting. It’s a new frontier for us and for West Texas,” Glover said.

As legacy data center regions like Virginia get saturated with facilities, the frontier regions such as West Texas are going be more attractive, Glover said, with easier regulatory environments and more sparse populations.

“Our hope is we’re moving dirt on projects within the next couple of years,” he said. “What’s attractive about TPL is we can really scale this. You can build multiple, multi-gig data center campuses with one owner. Just like in other industries, scale really matters here.”

Schmidt said Bolt plans to start with one anchor customer and grow from there. He name-dropped many potential anchors: Google, Microsoft, Meta, Amazon, Oracle, OpenAI, Anthropic, xAI, Palantir, and even the White House’s new Genesis Mission for AI.

Bolt is taking a bespoke approach similar to that of Texas-based AI power startup Fermi, backed by former U.S. energy secretary and Texas governor Rick Perry. Fermi launched an IPO in October before it had even started collecting revenue and quickly surged to a $16 billion market cap, though its value has since plunged to $5 billion at the end of 2025. However, Bolt is staying private and not banking on public investor interest in the AI boom.

The plan is to start with natural gas-fired power and grow to 1 gigawatt capacity, Schmidt said, then build more campuses as the power generation sources expand to include wind, solar, and battery power and, eventually, nuclear power over time. The goal is to grow to 10 gigawatts of power—enough to electrify about 7 million homes—on Texas Pacific Land acreage.

“We’re taking a different approach from traditional data center models that lease space and buy power from the grid. By vertically integrating energy ownership with advanced data infrastructure, we can design a platform that is both efficient and resilient,” Schmidt said.