DOE Advances UC Irvine’s AI Geothermal Project to a Three-Year Second Phase

An AI research effort to predict how underground rock will fracture is moving into a three-year second phase. UC Irvine announced on October 8, 2026, that the Department of Energy Office of Science selected its MAESTRO project to advance, with goals of improving geothermal assessments and understanding the earthquakes that geothermal operations can trigger.

MAESTRO—short for Multi-Agent AI Expert for Subsurface Reasoning and Optimization—brings together five University of California campuses, four national laboratories and three private-industry partners. The team combines expertise in geophysics, rock mechanics, chemistry, AI and engineering. UC Irvine did not disclose the amount of DOE funding.

From underground measurements to fracture predictions

Enhanced geothermal systems generate power by circulating fluid through rock with limited natural pathways for water. The fluid extracts underground heat that is used to drive turbines. Making that work requires finding locations where engineers can create and control fractures at suitable depths.

Site assessment involves more than locating heat. Engineers need to understand rock properties, underground stresses, existing cracks, water availability and drilling access. Many subsurface properties cannot be observed directly. Instead, researchers infer them from remote measurements and comparisons with locations already shown to be suitable for enhanced geothermal systems.

MAESTRO researchers plan to combine real-time geophysical observations and prior learning in AI-assisted simulations. Their target is models that predict how cracks form during fracturing operations, helping operators assess the likelihood that a geothermal project will succeed.

UC Irvine says the design includes a safety mechanism intended to keep AI recommendations physically consistent and trustworthy. A related research goal is understanding and controlling induced seismicity: earthquakes that can result from fracturing activities.

The university says tools developed for that work may also provide a basis for early warning of naturally occurring earthquakes. That possible application remains an ambition, rather than an announced warning service.

Turning the research into public guidance

Co-principal investigator Russ Detwiler said the project will be open source, sharing digital resources with researchers, policymakers and the public. He expects eventual real-time guidance to help energy developers reduce the financial risks and time involved in bringing geothermal projects into operation.

Detwiler also described a free browser-based tool that anyone could use to explore geothermal potential, assess risk and review data across the country. The tool is a planned project output, not a released product.