--- title: "Ormat spent 60 years drilling for steam. Now the buyers are hyperscalers" description: "Ormat runs 1.85 gigawatts of geothermal capacity and wants 2.8 gigawatts by 2028. First-half revenue was $662.7 million, up 43%. The reason is that a data center needs power that never stops, geothermal is one of the few zero-carbon sources that delivers it, and the scarcity of firm supply is now setting the price." category: "Economy" category_url: https://boursel.com/category/economy author: "Daniel Okonkwo" published: 2026-08-22T10:52:00.000Z updated: 2026-08-22T10:52:00.000Z canonical: https://boursel.com/article/ormat-spent-60-years-drilling-for-steam-now-the-buyers-are-hyperscalers tags: ["geothermal", "data-centers", "energy", "ormat", "artificial-intelligence"] --- # Ormat spent 60 years drilling for steam. Now the buyers are hyperscalers Ormat runs 1.85 gigawatts of geothermal capacity and wants 2.8 gigawatts by 2028. First-half revenue was $662.7 million, up 43%. The reason is that a data center needs power that never stops, geothermal is one of the few zero-carbon sources that delivers it, and the scarcity of firm supply is now setting the price. Ormat Technologies has been drilling for underground heat since the 1960s, mostly in places where the geology made it obvious: Nevada, Indonesia, Turkey, New Zealand. It is now selling that output to data centers, and the market has repriced it accordingly. The company operates [1.85 gigawatts of geothermal capacity and is targeting 2.8 gigawatts by 2028](https://fortune.com/2026/08/22/after-60-years-underground-power-cloud-ormat-ai-pivot/), across conventional plants and battery storage. First-half revenue was $662.7 million, up 43% on a year earlier, against a market value of roughly $6.75 billion. Fortune reports contracts with Google and the data-center developer Switch. ## What "firm" power means and why it commands a premium Solar and wind are now the cheapest generation to build almost anywhere. They are also intermittent: the output depends on the weather and the hour. A data center cannot work that way. It draws close to its full load continuously, and the economics of the building are set by utilisation, so idle hours are simply lost money. What its operator needs is firm power, meaning supply that is there at three in the morning in January regardless of conditions. Historically firm and zero-carbon did not go together. Gas is firm but emits. Nuclear is firm and clean but takes a decade to permit and build. Batteries shift renewable output by hours, not seasons, and adding enough of them to firm a wind farm year-round is expensive. Geothermal is the exception. A well tapping a hot reservoir produces at a steady rate indefinitely, with no fuel and essentially no emissions in operation. That combination is rare enough that buyers with net-zero commitments and 24-hour loads will pay above the market rate for it, which is why a company with six decades of unglamorous drilling experience is suddenly interesting. Chief executive Doron Blachar's framing, quoted by Fortune, is that the company has ["the hyperscalers and the AI demand"](https://fortune.com/2026/08/22/after-60-years-underground-power-cloud-ormat-ai-pivot/) and sees endless demand for it. ## The constraint Conventional geothermal works only where nature has already done most of the work: a heat source close enough to the surface, permeable rock, and water in it. That is a short list of places, which is why the industry has stayed small while solar and wind grew enormous. Enhanced geothermal systems are the attempt to remove that restriction. The method is to fracture hot dry rock and circulate water through the fractures to create a reservoir where none existed, which in principle makes geothermal available almost anywhere deep drilling is possible. Ormat is running pilots at Blue Mountain and Desert Peak in Nevada with Sage Geosystems and SLB, with projects expected online in late 2027, and Fortune reports that a single successful installation could supply 500 megawatts to one hyperscaler, against the roughly 100 megawatts a year Ormat typically adds from conventional plants. That is a fivefold change in the unit of production, and it is also the part that is not proven. Deep drilling is slow and capital-intensive, and injecting fluid into rock at pressure has caused induced seismicity that has stopped projects elsewhere. No one has yet run an enhanced system profitably and continuously at the scale a hyperscaler would require. ## How to read it The distinction worth holding is between the revenue Ormat has and the revenue the valuation implies. The existing business is real: 1.85 gigawatts, contracted, generating cash, growing at 43%. That is a good utility-scale energy company benefiting from a genuine shortage of the product it happens to make. The larger story, in which geothermal becomes a general-purpose answer to the power constraint on AI, depends on a technology in pilot. If it works, the addressable market changes shape entirely. If it does not, Ormat is still a well-run geothermal operator in a tight market, which is a smaller and more ordinary thing. Boursel has reported this week that power and grid capacity, not chips, are what now limit where data centers can be built, and that Nvidia has begun investing directly in the companies that secure electricity for them. Ormat is the supply side of that same shortage. The premium being paid for firm, clean power is the clearest price signal in the AI build-out, and it is showing up in the valuations of companies that have nothing to do with computing. ## Sources - [After 60 years underground, powering the cloud: Ormat's AI pivot](https://fortune.com/2026/08/22/after-60-years-underground-power-cloud-ormat-ai-pivot/)