AI-Driven Multi-Scale Framework for Critical Mineral Discovery and Recovery from Subsurface and Unconventional Sources

Critical minerals such as cobalt, rare-earth elements, and thorium are essential ingredients in electric vehicles, renewable energy systems, and modern electronics. The United States currently relies heavily on foreign sources for these materials, creating vulnerabilities in our energy and technology supply chains. Finding and recovering more of these minerals domestically is a national priority, but doing so is difficult: deposits are often hidden deep underground, spread across large and rugged landscapes, or locked up in places we don't normally think of looking, like old mine waste piles or groundwater.

This project builds GeoFinderAI, a smart, AI-powered tool that acts like a highly trained geological detective. Instead of relying on any single clue, it combines many different types of information at once, satellite images of the ground surface, measurements of underground rock structures, water and soil chemistry, and even the DNA of microbes living in the soil, to build a complete picture of where critical minerals may be hiding and how they might be recovered. The AI learns the patterns that connect surface clues to what lies underground, allowing it to make predictions far faster and across far larger areas than traditional methods allow.

The team will test GeoFinderAI across two well-known mineral-rich areas in Idaho: the Lemhi Pass belt, rich in rare earth elements and thorium, and the Blackbird belt, known for cobalt deposits. Over nine months, the system will be built, trained, and rigorously tested to demonstrate that it can outperform conventional approaches.

All tools and data produced by this project will be made freely available to the broader scientific and mining community, supporting a stronger and more self-sufficient domestic critical minerals industry.