Nvidia, PNNL and Fervo Energy build AI digital twin to unlock geothermal power 10,000 feet deep

Building a 'digital twin' 10,000 feet underground: PNNL, Nvidia and Fervo team up on geothermal AI

The idea is so simple: generating power from the heat trapped beneath the Earth's crust. It's clean, renewable and potentially abundant. The challenge is how to map what lies beneath -- and efficiently tap it. Pacific Northwest National Laboratory is partnering with chipmaker Nvidia and Fervo Energy, a leading geothermal company, to build a publicly available, digital twin that will create physical models of geothermal reservoirs to optimize power generation. Geothermal energy is produced by drilling wells that push cold water to depths of up to 10,000 feet below the surface -- for comparison, Seattle's Space Needle is 605 feet tall. The water flows through a network of underground fractures, which can be widened and connected through high-pressure injections. Expanding those fractures allows water to reach higher temperatures before returning to the surface, where it produces steam that spins power turbines. Underground rocks at those depths can reach 555 degrees Fahrenheit. "Plant operators need to answer questions like 'How many monitoring wells does the system need? How do we design those wells? How much water should we inject?'" said Maruti Mudunuru, an Earth scientist at PNNL and principal investigator of the project, in a statement. Current models are too slow to provide meaningful insights and guide operators addressing problems in wells, reservoirs or pipelines in real time. Those delays, Mudunuru added, "can lead to an underutilized resource." PNNL researchers will train the AI models; Nvidia will contribute technical expertise and data center infrastructure for the virtual twin; and Fervo is providing proprietary data from its geothermal sites in Nevada and Utah. Geothermal is viewed as an increasingly promising source of clean energy and attracting interest from investors and tech companies hungry for electricity. Earlier this month, Endurance Energy, a Seattle-based startup looking to extract energy from beneath the ocean floor, announced $54 million in funding. Fervo launched its Nevada commercial pilot, Project Red, in 2023, supplying 3 megawatts to a grid serving some of Google's data centers. The company is now building its Cape Station plant in Beaver County, Utah, which is expected to begin delivering electricity to the grid later this year and will ultimately generate 500 megawatts -- enough to power a small city. Fervo's design captures steam in a closed-loop system that returns it below the surface. The company raised $2.17 billion in its initial public offering last month, according to PitchBook. The AI models generated by the project will be incorporated into Nvidia's Omniverse libraries. The final product -- named the Enhanced Geothermal System Twin, or EGS Twin -- should be completed by 2029. It is funded by the Department of Energy's Hydrocarbons and Geothermal Energy Office.

Fervo Energy, NVIDIA and Pacific Northwest National Laboratory Announce Agreement to Develop Next-Generation Digital Twin Platform for Enhanced Geothermal Systems Technology, Known as EGS-Twin

Fervo Energy announced that it has entered into an agreement with NVIDIA Corporation and Pacific Northwest National Laboratory to develop a next-generation digital twin platform for Enhanced Geothermal Systems technology, known as EGS-Twin. EGS-Twin is designed to deliver real-time insight into subsurface behavior and operational performance through the integration of high-resolution field data with physics-based modeling and AI-driven forecasting. To build EGS-Twin, Pacific Northwest National Laboratory researchers will use Fervo Energy?s industry expertise and field data to train scalable AI models on NVIDIA Corporation AI infrastructure. The trained AI models will then be integrated into the NVIDIA Corporation Omniverse libraries, helping geothermal operators more quickly identify and respond to subsurface changes, optimize power generation, and strengthen the scalability of enhanced geothermal systems. Pacific Northwest National Laboratory will develop the workflows and data pipelines, leveraging high-performance computing, including U.S. Department of Energy supercomputing resources, to run large-scale simulations. Using currently available proprietary field data from Fervo Energy?s Nevada and Utah sites, the Pacific Northwest National Laboratory team will begin training the digital twin immediately and continue refining the platform as additional production data comes online, with the platform scheduled for implementation by 2029. Supporting the deployment of 24/7 carbon-free power to meet growing global energy demand, this collaboration represents a major step toward integrating AI and advanced computing into clean energy.