Earth AI uncovers six critical mineral prospects with predictive models

Traditional mineral exploration usually looks like this: seasoned geologists hiking across deserts for months with rock hammers, or mining giants drilling endless holes around decades-old deposits. Earth AI is rewriting that playbook by letting predictive algorithms spot critical metals from miles away.

The predictive exploration startup announced that ground verification has officially confirmed six brand-new mineralized prospects across its 100% owned tenements in Australia.

Hitting paydirt in data deserts

The discoveries span two distinct projects covering over 900 square kilometers, pinpointed by the company's proprietary Mineral Targeting Platform.

At the Kooranjie Project, located roughly 500 kilometers northwest of Sydney in New South Wales, Earth AI targeted a 512-square-kilometer area in the Cobar Basin. Surface sampling confirmed tungsten concentrations up to 0.26%, alongside silver, molybdenum, and tin, with elevated lithium and topaz in the broader system.

Further north, at the Elkedra Project in the Northern Territory, Earth AI confirmed three new prospects across a 400-square-kilometer expanse. Laboratory assays revealed cobalt grades up to 0.39%, copper at 1.39%, and gold at 0.685 grams per ton.

Custom hardware meets frontier AI

Most exploration tech focuses on areas near existing mines where historical drill logs are abundant. Earth AI took the hard road, training its models on remote sensing and geophysics specifically to operate in data-poor greenfield environments.

They did not stop at code. The company also builds its own proprietary exploration drill rigs, which streamline site logistics, drill smaller exploratory holes, and sharply reduce environmental impact and site mobilization costs.

According to Earth AI, this full-stack setup delivers an industry-high 75% discovery hit rate while accelerating exploration timelines fourfold and slashing costs by up to 75%. Maiden drilling at Kooranjie is slated for May 2025.

Why it matters

The global energy transition is starved for critical metals like tungsten, cobalt, and copper, yet traditional greenfield discovery rates have collapsed over the past two decades. Prospecting is notoriously slow, capital-intensive, and prone to expensive dry holes.

Earth AI is showing that machine learning can genuinely break new ground rather than just recycling known geology. By venturing into unexplored territory and validating findings with real surface assays, it lowers the cost of entry for critical supply chains that Western governments are desperate to de-risk.

We think their most pragmatic move was avoiding the pure software trap. The mining industry is notoriously skeptical of slide decks; you cannot convince a mining board without core samples in hand. By acquiring its own land packages and deploying proprietary drilling rigs, Earth AI controls the entire discovery feedback loop.

That said, high-grade surface rocks do not guarantee an economically viable mine. Everything now hinges on the May 2025 drill program, which will reveal whether these anomalies hold continuity and scale at depth. Earth AI found the door, but it still has to prove what lies behind it.