By Jinjoo Lee

Energy startups are in demand on the promise that they will deliver power to artificial-intelligence hyperscalers. The mere prospect of doing so at some time in the future is enough to command multibillion-dollar market capitalizations.

Fervo Energy, a geothermal startup that made its public debut this month, is closer than many to fulfilling that promise.

The company applies hydraulic-fracturing techniques to tap underground heat. Its initial public offering was a success: Fervo's shares are up 42% from its IPO price, with a market cap of $12.4 billion.

That is hefty for a company that doesn't yet generate meaningful revenue, and it is comparable with two nuclear-energy players. Oklo, backed by OpenAI's Sam Altman, and Amazon.com-backed X-Energy are each valued at around $11.5 billion.

Fervo has a much more visible path to revenue than those other companies. The first unit of its commercial project -- Cape Station in Utah -- is expected to start generating power for its customers by Oct. 1. The next two units are slated to deliver power by Jan. 1. These are contracts with consequences: If the company doesn't meet those deadlines, it must pay its customers.

By contrast, most nuclear startups expect commercial delivery by 2030. And even that could be optimistic. Many small modular-reactor, or SMR, designs have yet to receive full design certification from the Nuclear Regulatory Commission.

In total, Fervo has 658 megawatts -- or $7.2 billion -- of binding power-purchase agreements with companies including Southern California Edison, Alphabet's Google, NV Energy and Shell.

Geothermal energy does have one real constraint compared with nuclear SMR. It only makes economic sense in certain places, primarily the western part of the U.S., where hot rock can be found in shallower surfaces.

Fervo estimates it can develop 42 gigawatts across Utah, Nevada and Idaho, comparable to the amount the U.S. solar industry installed in 2025. Fervo finance chief David Ulrey says geothermal projects could become economic in such places as the East Coast over the next decade as drilling costs decline.

One selling point for geothermal energy is that it is an always-available "baseload" power, unlike more-intermittent sources such as wind and solar. Moreover, other sources of always-available power face high costs and bottlenecks in the supply chain.

Wood Mackenzie estimates that by the end of 2027, gas-turbine prices will be nearly triple what they were in 2019. Ordering a turbine today means getting it delivered five or more years later.

Fervo says a permitted geothermal project takes about three years to build; the goal is to get that down to 18 months. Its plants require a different type of turbine that faces much less of a supply-chain constraint.

Power produced by Fervo's first project won't be cheap at roughly $7,000 a kilowatt, comparable to some nuclear SMR designs. Eventually, it expects costs to fall to $3,000 a kW, which would make it more comparable to a new natural-gas power plant.

Meanwhile, data indicates cost reductions. BloombergNEF says Fervo's drilling costs per well have declined by about 29% with each doubling of cumulative wells drilled.

Earlier-stage geothermal projects that use drilling techniques could yield annualized returns of 11% to 14%, rising to 15% to 21% in later phases as costs decline, according to Project InnerSpace, a nonprofit geothermal energy research organization. That is much better than the single-digit returns traditionally seen in solar and wind, and in line with some Permian shale projects.

In part, this is because tech companies are willing to pay a higher price for clean, always-available power. It also helps that geothermal energy qualifies for federal tax credits through 2033.

The big question for Fervo is whether its geothermal wells will generate consistent heat over the life of its 15-year contracts. Fervo's S-1 filing says its pilot project generated "consistent, stable temperature, " but that project has only been operational since 2023.

To keep the heat output steady, the company might have to refracture existing wells or drill additional ones, says Dani Merino-Garcia, vice president of research at Project InnerSpace. Additional wells will be a small part of the total project's capital expenditures, according to Fervo's Ulrey.

There will be other adjustments, too, such as figuring out the optimal water flow to make sure rocks don't cool too quickly and create fractures in such a way that the water flow through them is homogenous, according to Merino-Garcia.

Another geothermal company worth watching is Ormat Technologies. It is a longstanding giant of the industry that is working on pilot projects that use oil-and-gas drilling techniques.

The company this year signed a 150MW power-purchase agreement with Google and has been able to extend expiring contracts on older plants at higher prices. Ormat's market cap is modest compared with shinier energy upstarts, at around $8 billion.

Always-on power is a scarce resource for hyperscalers, and there is a limited universe of companies that can provide it. Those that actually deliver the power deserve a premium.

Write to Jinjoo Lee at jinjoo.lee@wsj.com

(END) Dow Jones Newswires

May 23, 2026 09:45 ET (13:45 GMT)

Copyright (c) 2026 Dow Jones & Company, Inc.