Introduction
The race to power artificial intelligence is pushing the tech industry toward an unlikely energy source: rock heated to extreme temperatures deep beneath the Earth's surface. While conventional geothermal plants have operated for decades, a new wave of startups is pursuing super-hot rock systems that could deliver vastly more electricity from a single well. With Google already committing hundreds of millions to one such project, the technology is moving from experimental to commercial faster than many expected.
What Happened
Google struck a deal worth more than $500 million to buy 50 megawatts of power from Quaise Energy, a startup developing super-hot rock geothermal in central Oregon. The project will drill roughly three miles underground to access heat intense enough to turn water into a supercritical fluid, a state that carries far more energy than ordinary steam.
Around the same time, Mazama Energy, backed by Khosla Ventures, announced a new cooling method called Carbon Chill that uses carbon dioxide instead of water-based mud to keep drill bits from overheating at extreme depths. Meanwhile, XGS Energy received a Department of Energy grant to improve well designs and materials for super-hot conditions.
These developments follow Fervo Energy's public market debut earlier this year, which gave investors their first clear look at enhanced geothermal as an investable sector.
Why This Matters
Data centers, especially those training and running large AI models, are consuming electricity at unprecedented rates. The United States alone has an estimated 57 terawatts of geothermal potential, more than enough to support the next generation of digital infrastructure. Super-hot rock systems could change the economics of geothermal power by extracting more energy from fewer wells. Since drilling can account for up to a third of total project costs, reducing the number of boreholes needed could make clean baseload power competitive with fossil fuels.
Key Takeaways
- Super-hot rock geothermal uses extreme heat and pressure to create supercritical fluids that carry large amounts of energy.
- A single well could potentially produce 15 megawatts or more of electricity.
- Google is paying a premium to secure early supply from Quaise's Oregon project.
- Quaise plans to use millimeter-wave radiation to vaporize rock, while Mazama relies on CO2-cooled conventional bits.
- Drilling represents roughly 25 to 33 percent of geothermal plant costs, so deeper, hotter wells could improve project economics.
- The technology remains unproven at commercial scale, and technical risks remain.
Conclusion
Super-hot rock geothermal is no longer just a research concept. With major corporate buyers, federal support, and hundreds of millions in venture capital flowing into the sector, the industry is testing whether deep-earth heat can become a reliable power source for the digital age. If Quaise, Mazama, and others succeed, the result could be a scalable, carbon-free baseload option for data centers and the broader grid. If they fail, the search for clean, firm power will continue—but the stakes for AI-driven electricity demand have never been higher.

Discussion
Join the conversation
Thoughtful reactions, questions, and follow-up ideas help shape the next story.