As I write this in July 2026, the role of geothermal electrical power generation at scale is still uncertain. To be sure, there is progress. Big news earlier this year in the US was the initial public offering of Fervo Energy on 13 May. Under the ticker symbol FRVO, the NASDAQ stock closed its first day of trading with a market capitalisation of over $10 billion.
Fervo is the technical and market leader in a new style of geothermal development that utilises methods and equipment from the oil and gas sector to drill horizontal injector–producer doublets. Typically, the injector lateral is completed using multi-stage hydraulic fracturing to create permeability in otherwise tight igneous and metamorphic rock. Brine is pushed through the reservoir, where it absorbs heat, before passing through a surface binary-cycle power plant to generate electricity. Except for the power plant, this all sounds pretty familiar to anyone in the oil and gas industry. The shale revolution has been under way since 2009, transforming the US from a net importer into the world’s largest oil producer.
The timescale is important. We are 150 years into the age of oil, 20 years into the age of shale oil, four years into the age of AI and two years into the age of Fervo-style geothermal power. I say we are in year four of AI because ChatGPT—the first publicly available large language model—was launched on 30 November 2022.
“We are 150 years into the age of oil, 20 years into the age of shale oil, four years into the age of AI and two years into the age of Fervo-style geothermal power.”
I will not go into the nature or technology of AI; our concern here is the enormous demand for energy from the data centres required as AI scales exponentially. Currently, the US alone has 55 GW of installed data-centre capacity, with another 400–800 GW announced or planned. How are all these data centres to be powered? Early facilities simply used grid power, but that will no longer work. Opposition is growing across the US, with some states legislating for a moratorium on new data centres because of anticipated increases in energy costs for everyone.
‘Behind-the-meter’ options are announced every day: dedicated gas turbines from established or start-up companies, solar or wind farms, small nuclear reactors and—this is the point—geothermal. Fervo has a 3 GW agreement with Google to provide power for data centres and, as we have seen, this represents only a small fraction of the growing demand.
The new geothermal power boom is centred on the Basin and Range Province of the western US, an area with high geothermal gradients of 75–100°C/km. In such areas, a substantial geothermal power project can be developed at a drilling depth of less than 3 km, a case in point being the 500 MW Fervo Cape Station currently under construction in Utah. Like any technology, geothermal has advantages and disadvantages: the advantage is its minimal CO₂ footprint; the disadvantage is its tremendous need for water, which is scarce in the Utah desert.
Right now, in the US at least, climate concerns are taking a back seat to the data-centre construction boom. This has focused attention on pure market forces that favour energy sources with the lowest unsubsidised levelised cost of electricity (LCOE). The word ‘unsubsidised’ is important because federal government subsidies currently available for geothermal power can offset up to 40% of plant construction costs, and these are expected to continue for several years.
“In the US petri dish of capitalism, geothermal electrical power generation will surge or sputter based on its LCOE.”
It is too soon to say what the LCOE will be for Fervo Cape Station—a metric eagerly awaited over the next few financial quarters. In the US petri dish of capitalism, geothermal electrical power generation will surge or sputter based on its LCOE—unless, or until, low-carbon energy solutions once again become a priority.
Prof Cristopher L. Liner is Maurice F. Storm Endowed Chair, Department of Geosciences, University of Arkansas.
Views expressed in this column are solely those of the author.
