Mining data to mine battery metals
A startup is using artificial intelligence to find new sources of metals that power electric vehicles and renewable energy technologies.
Why it matters: The world desperately needs new supplies of lithium, cobalt and other metals to accelerate the shift to EVs and renewable energy, and machine learning can help narrow the search.
Driving the news: This morning, the Silicon Valley-based startup KoBold Metals — which is backed by Bill Gates and Jeff Bezos, among others — announced a partnership with the world's largest mining company, BHP Group, to use its technology to locate battery metal deposits.
- The partnership will focus on a more than 193,000-square-mile area of Western Australia, a region KoBold CEO Kurt House notes is "bigger than California."
- The search will focus on copper and nickel deposits believed buried 650 feet to 5,000 feet below the surface.
The big picture: "We need roughly $10 trillion worth of new discoveries of lithium, nickel, cobalt, copper and rare earth minerals just to be able to fully electrify the world's passenger vehicle fleet," says House.
- "But exploration success rates have been declining in recent years because the easy surface deposits have been found and because there just hasn't been a lot of investment in exploration technology," he adds.
How it works: Conventional mining exploration involves collecting physical deposits from field sites and analyzing them, but that takes time, with plenty of false starts. By one assessment, only 1 in 100 sites will have a deposit that can be profitably mined, and that proportion has been dropping in recent years.
- KoBold argues it can boost discovery rates 20-fold by tapping 160 years' worth of scattered geological data, as well as new data taken from fresh surface readings, and using machine learning approaches to analyze it and indicate where mineable deposits are likely to be found underground.
- "This is fundamentally an information problem with a search problem that's perfectly set up for the kind of advanced scientific computing that has had minimal investment up to this point," says House.
Between the lines: BHP's partnership with KoBold is part of the mining giant's effort to build out its portfolio of what it calls the "future-facing commodities" that will run the renewable-energy economy.
- Anglo-Australian mining company Rio Tinto has increasingly tapped artificial intelligence as part of its exploration process in recent years, including one major copper find in Western Australia.
- Such efforts need to speed up. A May report by the International Energy Agency found that deploying enough clean technologies to prevent 2°C of warming would increase demand for energy-storage minerals more than 30-fold by 2040.
The catch: Mining with AI is still mining, with all the political risks, human rights issues and environmental destruction that digging metals out of the ground can entail.
- Aboriginal groups recently rejected a bill intended to protect cultural heritage in mining areas in Western Australia, arguing they should have the final say over protection of their sacred sites.
- For its part, KoBold — which does exploration work itself, rather than simply selling its software to the highest bidder — has said it will only work in areas where it can get community buy-in.
Be smart: KoBold shows how technology plays an underappreciated role in just how large our supplies of vital natural resources really are.
- As market demand for minerals like lithium or cobalt rises, so does the willingness of mining companies to invest in finding more, as new technology can make previously unprofitable or unknown deposits profitably accessible.
The bottom line: We do need lots of lithium and cobalt to save us, and House is convinced "we'll have enough to electrify the economy."
- "I just don't know if we'll have enough to do it in time, which is why we're running as fast as we can."