Fervo Energy and its plan to industrialize enhanced geothermal, driven by AI demand

The MIT Technology Review article, part of its "10 Climate Tech Companies to Watch 2026" list, profiles Fervo Energy, a geothermal energy company that aims to extend this power source far beyond the handful of places where it has been practical so far. The download contains the full text of the piece, not a paywall teaser.

Traditional geothermal depends on areas where hot water flows easily through underground rock and carries heat to the surface. Fervo, founded in 2017 and based in Houston, Texas, uses horizontal drilling and hydraulic fracturing to open its own pathways through the rock. This could make geothermal development viable in many more locations.

The company demonstrated this basic approach in 2023 at its Project Red facility in Nevada. It is now developing Cape Station, a much larger plant in southwestern Utah that is expected to reach full commercial operation in 2028. According to the article, the first phase began selling electricity in late September.

The big change this year is the money available and the growth in electricity demand. Fervo went public in May, raising about $2.2 billion in one of the largest clean energy IPOs in history. It plans to invest that capital in accelerating its pipeline of larger projects.

One especially relevant group of potential customers is companies riding the AI boom, which operate huge fleets of energy-hungry data centers. Fervo expanded its relationship with Google this year. In March, the two signed an agreement to drive more geothermal plants: Fervo will propose at least 1 gigawatt of projects within two years, and Google will decide which ones it wants to buy power from. In September, Google agreed to buy at least 396 megawatts from Cape Station for a possible data center in Utah. The agreement covers up to 3 gigawatts of enhanced geothermal projects through 2033, equivalent to roughly three-quarters of the total geothermal capacity currently operating in the US.

Fervo believes this expansion will allow it to cut costs. Instead of designing a different facility each time, it wants to start from a common design that is refined from one project to the next. Each unit, called a GeoBlock, is a plant of about 50 megawatts, and combining several enables projects of hundreds of megawatts.

The company measures its progress partly by installed cost per kilowatt. It is targeting about $7,000 per kilowatt for the first 100 megawatts of Cape Station and $5,500 for the next 400 megawatts. That is a step toward its long-term goal of $3,000 per kilowatt, a level at which, according to the company, it would match and eventually undercut the cost of a new natural gas plant, without the ongoing expense of buying fuel.

As a caveat, the next 400 megawatts of Cape Station are being built alongside the first 100 in Utah. Success there would provide evidence that Fervo can scale production economically, but it does not mean the technology will transfer easily to other parts of the country, according to Shuvajit Bhattacharya, a research associate professor at the Bureau of Economic Geology at the University of Texas at Austin. The subsurface is not a uniform assembly line: geological conditions determine where it makes economic sense to drill and how deep. The company will need to show that its method works in more places than Utah and Nevada.

The coming years will test the strategy: cutting costs as Cape Station is completed, turning conversations with data center companies into operating plants, and replicating that success in other geological settings. The near-term goal is to have one gigawatt operating by the end of 2030, and five gigawatts in total five years later, more than the entire current US geothermal supply.