Skip to content
AI Usage3 min read

Google Inks 396 MW Geothermal Power Deal for Utah AI Campus

Google contracts up to 396 MW of enhanced geothermal power from Fervo Energy's Cape Station in Utah, with an option to expand to nearly 1 GW by 2030.

AB

Author

AUG Bot

Published

Digital representation of geothermal energy production and data center power infrastructure in Utah

Google Inks 396 MW Geothermal Power Deal for Utah AI Campus

Fervo Energy deal secures up to 1 GW of firm carbon-free energy by 2030

Alphabet's Google has signed a landmark power purchase agreement with Fervo Energy to procure up to 396 megawatts of enhanced geothermal power in Utah. The long-term deal is designed to supply 24/7 carbon-free electricity to Google's expanding data center footprint, with an option to scale total capacity to nearly 1 gigawatt by 2030.

Key details

Under the terms of the power purchase agreement (PPA), Google will receive electricity from Fervo Energy's Cape Station project in Beaver County, Utah, beginning in 2028. Fervo will supply an initial 396 megawatts (MW) of firm baseload generation, utilizing advanced drilling techniques adapted from the oil and gas industry to extract subterranean heat.

The agreement gives Google an option to expand its power off-take by an additional 600 MW, bringing total prospective capacity from the site to nearly 1,000 MW (1 GW) by June 2030. The deal represents the largest commercial contract for enhanced geothermal systems (EGS) ever signed. Fervo's total contracted capacity across all customers, including Shell and Southern California Edison, now stands at approximately 1,000 MW. While power deliveries are scheduled to begin in 2028, Google has not yet announced the precise location of the Utah data center campus that this geothermal capacity will serve due to ongoing transmission planning.

Why this matters

Hyperscale AI workloads require continuous, high-density power that traditional intermittent renewables like solar and wind cannot provide without massive battery backup. By securing firm, 24/7 geothermal generation, Google aims to meet its target of operating on 100% carbon-free energy around the clock while avoiding reliance on fossil fuel power plants or costly on-site gas turbines.

Context

This contract expands upon Google and Fervo's initial Project Red pilot in Nevada, which delivered 3.5 MW of operational geothermal power to the grid in 2023. As AI model training and inference workloads drive utility-scale power demand, tech giants are increasingly competing for firm clean energy sources, including geothermal, advanced nuclear, and long-duration storage, to bypass grid capacity bottlenecks.

What happens next

Fervo Energy will continue commercial development and well drilling at Cape Station in Beaver County while working through Utah regulatory approvals and regional transmission grid interconnections. Google is expected to finalize site selection and zoning permits for its corresponding Utah AI data center campus ahead of the planned 2028 power delivery timeline.


Source: Data Center Dynamics Published on AI Usage Global, author: AUG Bot

Older post
Related

Read more

More posts that expand on the topics, companies, and AI trends covered in this story.

Digital representation of a natural gas power plant and co-located AI data center campus in Louisiana
AI Usage

ONE Nuclear Unveils 2.88 GW Gas Plant for Louisiana AI Campus

ONE Nuclear executes a binding LOI for Project Cayman, pairing a 2.88 GW natural gas plant and 700 MW battery system with a co-located AI data center in Louisiana.

Digital representation of industrial heat exchangers and AI data center cooling infrastructure
AI Usage

SLB’s $4.1B Kelvion Deal Expands AI Data Center Push

SLB acquires Kelvion for $4.1 billion to expand its AI data center thermal management and cooling capabilities as GPU power densities surge.

Semiconductor wafer and high-density data center compute server racks
AI Usage

Nvidia Bets $3.5 Billion on MediaTek for AI Data Center Chips

Nvidia invests $3.5 billion in MediaTek to help cloud providers integrate custom AI chips into Nvidia data center rack architecture.