A federal laboratory, a geothermal developer and an AI computing company announced on June 22, 2026, that they are jointly building an AI-powered virtual replica of an enhanced geothermal reservoir, a tool its creators say could eliminate weeks-long analysis waits that today leave underground power plants operating below their potential.
What EGS-Twin is and how it works
Fervo Energy, NVIDIA and Pacific Northwest National Laboratory (PNNL) announced an agreement to develop a next-generation digital twin platform for Enhanced Geothermal Systems (EGS) 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.
Fervo Energy deploys fiber-optic cables and uses acoustic technology to map and gather intelligence from the subsurface, but processing and analyzing all that data takes too much time for operators to act swiftly. Fluctuations in the operating data can indicate issues in wells, reservoirs or pipelines that may require attention. Current models that help represent the dynamics of a geothermal system can take weeks to run. EGS-Twin would run in real time and allow operators to respond quickly to any underground problems that may arise.
To build EGS-Twin, PNNL researchers will use Fervo’s industry expertise and field data to train scalable AI models on NVIDIA AI infrastructure. The trained AI models will then be integrated into the NVIDIA Omniverse libraries, helping geothermal operators more quickly identify and respond to subsurface changes, optimize power generation, and strengthen the scalability of enhanced geothermal systems.
Data sources and the road to commercial deployment
Using currently available proprietary field data from Fervo’s Nevada and Utah sites, the PNNL 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.
The final EGS-Twin will contain anonymized data so that other geothermal plant operators can adapt it to their own operations. That open-access design matters in an emerging industry where a broader, standardized modeling tool could compress the learning curve for every developer that follows.
Fervo’s CTO and co-founder Jack Norbeck said: “We believe that digital twins will expedite the learning curve for geothermal development as we build and operate our GeoBlock assets.” He added that “integrating high-fidelity physics-based models with AI-driven forecasting has the potential to reshape reservoir management, improve heat recovery, and enhance system reliability.”
Fervo’s Cape Station context
Fervo established Project Red in Nevada in 2023, which now generates 3 megawatts for the grid. Cape Station Phase I will deliver 100 megawatts (MW) of baseload clean power to the grid beginning in 2026, and Cape Station Phase II will bring an additional 400 MW online by 2028.
Fervo is 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. Cape Station is expected to be the largest next-generation geothermal development in the world.
Geothermal made up just 0.4% of all electricity generated in the United States in 2023. But according to the U.S. Department of Energy, geothermal could provide potentially 90 gigawatts of firm and flexible power to America’s grid by 2050, assuming that enhanced systems like Fervo’s catch on as a widespread renewable energy option.
Getting there will depend partly on whether operators can extract more from each plant they drill, which is precisely the problem EGS-Twin targets.
Why real-time reservoir intelligence matters now
The EGS-Twin partnership reflects a wider convergence of AI infrastructure and baseload power ambitions. As GCN has reported, only 13 percent of US power projects in the grid interconnection queue reach commercial operation, a failure rate that presses every geothermal developer to demonstrate that operating plants can perform reliably once they are running.
Poor real-time visibility into reservoir conditions is one of the operational risks that can knock capacity factors well below the round-the-clock figures written into power contracts.
Maruti Mudunuru, the PNNL Earth scientist leading the project as principal investigator, said that “current modeling capabilities for geothermal systems are too slow to fully incorporate and analyze production data, which can lead to an underutilized resource,” and that a working digital twin would allow EGS operators to “understand, in real time, the dynamics of their reservoir and act quickly to maximize the power generation potential.”
The team will use currently available results to begin training the digital twin and will continue developing it as additional production data comes online. The platform is expected to be up and running by 2029.
The development also connects directly to the broader push to make geothermal a viable baseload option for data-center power demand. Federal support for EGS has expanded on multiple fronts this year, from Pennsylvania shale conversions to multi-state consortia, and EGS-Twin represents one of the first efforts to apply large-scale AI infrastructure to the operational side of that buildout.
