CHEYENNE, Wyo. — The generating station American Electric Power plans to build outside this city will sell its electricity to one buyer for twenty years. That contract, and the load behind it, is doing most of the work that a regulated rate case would normally do.
AEP said Wednesday it will invest about $4 billion in a 702-megawatt fuel cell project near Cheyenne, under a long-term power purchase agreement with Google. The project company, Cheyenne Fuel Cell LLC, is roughly 92 percent owned and controlled by AEP, with nearly 8 percent held by funds advised by Basalt Infrastructure Partners.
The plant will use fuel cell technology developed by Bloom Energy, which generates electricity without combusting natural gas, according to the company. AEP said the project includes plans to integrate carbon capture to deliver near-zero-emissions power.
Construction is scheduled to begin this year, with full operation expected by the end of 2028. The electricity will serve Google’s growing demand in the greater Cheyenne area, where the search company is building data center capacity.
“As power demand continues to grow, AEP is bringing together new technologies and strong partnerships to deliver reliable, innovative energy solutions quickly and at scale,” said Bill Fehrman, AEP’s chairman, president and chief executive, in a statement.
The announcement follows agreements Google reached with Black Hills Corp., effective Sept. 30, to serve its planned Cheyenne data center. Those arrangements run through 2048 and commit the utility to building 564 megawatts of new gas-fired generation, part of a package valued at about $1.8 billion.
Utility executives have spent two years describing the same problem. Data center operators want power quickly, and the equipment that supplies it is booked years out. Gas turbine order books at the major manufacturers stretch toward the end of the decade, which leaves projects that need electricity in 2028 looking for something that can be assembled faster than a combined-cycle plant.
Fuel cells fit that gap. They are modular, shipped by truck and commissioned in stages, which appeals to a developer trying to match supply to a tenant’s phased build-out. The trade is cost and a fuel supply that still runs through the same pipeline network.
Bloom Energy and AEP are not new to each other. The two companies announced a gigawatt-scale fuel cell procurement agreement in late 2024, framed at the time as a way to bring power to AI data centers that could not wait for grid upgrades. The Cheyenne project turns that framework into a named site, a named customer and a signed twenty-year contract.
The Wyoming project also sits outside AEP’s traditional service territory, which covers parts of eleven states from Ohio to Texas. That matters for how the cost is borne. Power sold under a contract to a single large customer rather than through regulated retail rates means the accounting depends on the credit of that customer, not on a rate base approved by a state commission.
AEP is in the middle of a $78 billion capital plan for 2026 through 2030, aimed at serving load growth across its footprint. Projects of this kind let the company point to investment that does not wait on a regulatory calendar.
Utilities have argued that large-load customers should pay for the infrastructure built on their behalf, through special contracts and tariffs, so that residential ratepayers do not absorb the risk of demand that may not materialize. Wyoming regulators and AEP’s own state commissions will test that principle as similar deals accumulate.
For Google, the arrangement is one of a set. The company has been signing supply contracts with utilities, developers and nuclear operators as it tries to secure firm power for data centers whose electricity needs are measured in gigawatts. The Black Hills agreements and the AEP project together give it a foothold in a state with cheap fuel and available land.
The risk sits in the forecast. A fuel cell fleet built for one tenant is worth what that tenant’s demand is worth, and twenty years is a long time in a business where computing architectures and model designs change within eighteen months. Google has not published a load forecast for Cheyenne, and AEP has not said what happens to the plant if the tenant’s needs fall short of plan.
Carbon capture is the other assumption. Projects that promise near-zero emissions by adding capture equipment to a generating asset have a mixed record on cost and schedule, and the technology here is described as a plan rather than a signed contract.
What the deal shows is how the electricity system is being reorganized around a handful of very large buyers. A utility that once built generation to serve a whole territory now builds it to serve a named customer on a negotiated price, with the state’s role reduced to approving the terms after they are set.
Whether that produces cheaper power for everyone else is the question regulators have not yet answered at scale. Cheyenne will be one of the places where they start.
