This Startup Finds Critical Minerals in the Unlikeliest of Places

A century ago, prospectors tromped through remote areas, hoping to spot valuable, mineralized rocks simply poking out of the ground. Eventually, after they found all of the obvious stuff, they progressed to doing airborne geophysical surveys that used tools such as electromagnetic sensing to identify minerals that were just below the surface or highly concentrated. But there’s always been a lot more out there than we had the mechanisms to find. So now, companies are training artificial intelligence models on heaps of historical data to help locate untouched reserves of minerals that are key to clean energy technologies such as electric vehicle batteries and wind turbines.

One of the biggest players in this space is Earth AI, a Sydney-based startup that today announced a $20 million Series B round, bringing the company’s total investment to over $38 million since its founding in 2017. While the company had initially sought to raise $15 million in this round, investor interest was so strong that it exceeded its target by $5 million. Lead investors were Tamarack Global and Cantos Ventures.

Earth AI has a two-stage business model. First, it uses its proprietary software to locate likely deposits and purchases the mineral rights to the land. Then, it sends in its drilling rig and in-house team of geologists to produce mineral samples. The team then shows these samples to mining companies to prove that the area warrants further development and — assuming the miners are interested — sells them the mineral rights. “Because of the scarcity of this, because of the deficit of where we are and where we’re going, these price tags are $500 million to $2.5 billion,” Monte Hackett, Earth AI’s chief financial officer, told me. Huge as that may sound, it’s much less than what a mining company would typically spend doing traditional exploration themselves. This latest funding will allow the startup to purchase additional drill rigs and increase its project pipeline to over 50 sites.

“The problem is obvious,” Hackett told me. “We need $10 trillion of critical metal production by 2050. We’re producing $320 billion, a $9.7 trillion shortfall.” To better locate the trillions of dollars worth of minerals necessary for the energy transition, the startup’s CEO, geoscientist Roman Teslyuk, digitized decades of old Australian geophysical survey data, then overlaid that with remote sensing data such as satellite imagery to train a model to recognize the Earth systems and geological processes that created minerals deposits millennia ago. “Another way I like to think about it is that our algorithm looks for the geological shadow that is cast by a dense mineral body,” Hackett explained in a follow-up email.

Hackett told me that Earth AI focuses specifically on “greenfield” applications — that is, areas where no mining activity or substantial minerals exploration has previously occurred. So far, the company’s discoveries include a significant deposit of palladium, which also contains platinum and nickel, as well as a gold deposit and a molybdenum deposit. “We’re finding things that go against what has been the common sense of the industry so far,” Hackett told me, referencing the company’s palladium discovery. “There was geological consensus that there was no platinum palladium on the East Coast of Australia, and our algorithm learned what it looked like on the West Coast and then identified it on the East Coast.”

While Hackett said Earth AI is open for business anywhere, right now all of its projects are in Australia, where the company has “600 ‘x’s on our treasure map” — that is, likely areas for deposits, Hackett wrote in a follow-up email. Outlining the advantages of doing business there, he explained, “We don’t have to go somewhere where there’s unsavory working conditions or there are issues where we have to put our principles into balance. Here, everything is very regulated and above board.” Plus, mining has long been an important component of the Australian economy. “They’re incredibly efficient at doing this well. So the timeline for development is the shortest compared to other places,” Hackett told me.

This tech could have important domestic implications too, though. As the newly-inaugurated President Trump prioritizes ramping up U.S. production of critical minerals, Earth AI could one day help locate deposits here, as well. And since Australia is a close American ally, the nation could play a key role in helping wean the U.S. off of Chinese imports, providing the U.S. with critical minerals that it can’t now, or perhaps ever, produce in sufficient quantities itself.