Fervo Energy teams with PNNL on AI-driven digital twin for next-generation geothermal

Fervo Energy, a U.S. geothermal developer, announced it is partnering with the Pacific Northwest National Laboratory (PNNL) to build a “digital twin” platform intended to advance geothermal development. The effort is designed to bring together continuous field observations, physics-based models, and artificial intelligence, with the goal of improving how geothermal reservoirs and supporting infrastructure are understood and managed.

In a June 22 update distributed through Globe Newswire and republished by Yahoo Finance, Fervo described the platform as a way to integrate real-time data from the field with simulation-grade modeling. Digital twins are computer models that attempt to mirror real-world systems so teams can test scenarios, refine operational decisions, and reduce uncertainty before and during development.

The company said the digital twin approach is aimed at “critical geothermal infrastructure” and at maximizing power generation. While Fervo did not provide project timelines or performance targets in the material available for this story, it positioned the system as a practical tool for decision-making in geothermal projects where reservoir behavior, drilling outcomes, and operating conditions can be difficult to predict.

NVIDIA is also named as a technology contributor. Fervo said it is leveraging NVIDIA accelerated computing as part of the computing environment for the digital twin work. Accelerated computing generally refers to using specialized parallel processors, such as GPUs, to run large simulations and AI workloads more efficiently than general-purpose CPUs alone.

A key element of Fervo’s stated approach is combining physics-based modeling with AI. Physics-based modeling uses established physical laws to represent processes such as heat transfer and fluid movement in subsurface systems. AI, in this context, is described as an added layer that can help interpret noisy measurements, accelerate analysis, or improve model outputs using data from real operations.

For PNNL, the collaboration fits within the lab’s broader role in energy research, where advanced computing and modeling are frequently used to support infrastructure development. However, beyond the partnership framing and the joint focus on digital twins, the announcement did not outline specific PNNL contributions, such as particular modeling methods or datasets, in the information provided here.

What remains unclear from the published announcement is the scope of the platform. Fervo did not specify which geothermal site(s) or subsurface conditions the digital twin is initially targeting, nor did it disclose whether the platform will be used for drilling optimization, reservoir management, maintenance planning, or all of the above. It also did not disclose which exact NVIDIA hardware or software stack is being used, or how the platform will be measured once deployed.

Going forward, the next material developments to watch are details on deployment, performance, and integration. If Fervo later provides updates on pilot results or operational impact, those disclosures would help clarify whether the system reduces uncertainty and improves outcomes in ways that matter to geothermal project economics, such as uptime, reservoir performance, or faster iteration during development.