AI in Mineral Exploration: How Software Finds Buried Gold

AI in mineral exploration is the use of machine learning to predict where mineral deposits hide, and it is the genuine technological core of Whitney Tilson’s “$6 Miner / Rushmore Gold” presentation for Stansberry Research. Strip away the Mount Rushmore cinematography and the pitch is about a real shift: for 150 years, finding ore was guesswork refined by drilling, at hit rates below one percent. Now software can read a century of accumulated data and tell geologists where to aim. Here is how it works and what it has actually found. We walk through the specifics in our AI Mineral Exploration: The 75% Hit Rate Changing Mining piece. We walk through the specifics in our AI Mineral Exploration News: Recent Discoveries Explained piece.

The problem being solved

Traditional exploration is brutally inefficient. Geologists sample streams, run geophysical surveys, and drill, mostly into nothing. The industry’s discovery hit rate sits around half a percent of drill targets, and discovery costs per ounce have climbed for decades as the easy surface deposits ran out. The data problem is the twist: the mining industry has generated enormous archives, “old maps, drill results, surveys,” as Tilson’s presentation puts it, that no human team could ever integrate. “It had just been collecting dust in archives and storage lockers for decades. It was way too much information for humans to sift through… but not AI.”

The method, in plain terms

Machine-learning exploration platforms integrate every available signal over a target area: historical drilling logs, geochemical assays, satellite and geophysical data, topography, and structural geology. The model learns the signatures of known deposits and flags locations with matching signatures that no one has drilled. It is the same pattern-recognition playbook that transformed drug discovery and advertising, pointed at rocks. The output is a ranked list of targets, which geologists still confirm with drills, but starting from odds many times better than blind drilling.

A slide from Tilson’s presentation on AI-driven mineral discovery rates

The results that made it famous

The proof cases Tilson cites are real, and they are why this story travels:

  • Dakota Gold (DC), South Dakota. “Geologists at this firm got access to 145 years of historic records for the area… The firm put all this historic data into AI-powered mining software… And it uncovered 114 tons of gold,” roughly a “$22 billion payload,” near Mount Rushmore. Tilson gives the ticker away free, with the honest caveat that production does not start until 2029.
  • Ramaco Resources (METC), Wyoming. After DOE scientists used AI to survey an old coal mine, “he found 1.7 million tonnes of rare earths, worth about $37 billion,” and the stock ran 773% in a little more than six months.
  • KoBold Metals, Zambia and beyond. The private explorer backed by “Jeff Bezos, Bill Gates, Mark Zuckerberg, Sam Altman” and others, valued at nearly $3 billion, found the Mingomba copper deposit with its platform.
  • Earth AI, whose 75% discovery hit rate Tilson cites as “150 times the industry average,” alongside Ero Copper’s partner Terra AI and Intrepid Metals’ TerraEye-assisted Arizona copper find.
  • The Pentagon’s MinMod, the public-sector version: an AI system that “mapped out 670,000 mineral sites across America… Including 184,000 gold sites.”

Why investors should care

The investment logic is that discovery, not mining, is where the re-rating happens. A company that controls ground holding a defined, expanding resource gets revalued as the resource grows, years before any ore is sold. AI compresses the discovery timeline, which means more of those re-rating events per decade, and tilts the advantage toward companies with data-rich ground: historic districts like the Black Hills, old coal fields like the Powder River Basin, and government-surveyed acreage.

Tilson’s macro frame, a “biggest resource boom in American history” backed by federal critical-minerals policy, stockpiling (“Project Vault”), and Burgum’s $200 trillion in-ground estimate, layers a policy thesis on top of the technology thesis. The technology part is solid; the macro extrapolation deserves measured reading, as in-ground estimates are not market value.

Considerations

AI improves target selection; it does not repeal geology or economics. A discovered deposit still faces years of drilling, permitting, feasibility studies, and financing before it produces anything, and pre-production stories dilute along the way. Exploration wins are also lumpy and stock-specific, which favors a portfolio approach over single-name bets. Tilson’s four picks, spanning AI-driven miners, royalties, and geothermal, are in the “AI Miners” report included with a trial of Commodity Supercycles, backed by a 30-day full-refund guarantee.

Where to learn more

For the flagship story, see our Dakota Gold stock article. For sector context, read our critical minerals overview and rare earth stocks explainer, plus how to invest in gold mining companies for the traditional toolkit AI is accelerating.

NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.