KoBold’s AI mineral hunt has a $3 billion promise, and a proof problem

KoBold Metals uses AI and geological data to search for critical minerals, but the company’s high-profile claims about what that approach has found remain unverified. A Berkeleyside investigation describes a company valued at nearly $3 billion and operating across 14 countries, while asking how much of its promise the technology can actually deliver.

Watch Desk analysis

What happened

Berkeleyside reports that KoBold is using computing methods to sift through geological records and other data in search of deposits of copper, lithium, nickel and cobalt. The company has around 60 projects across four continents. Its investors include Jeff Bezos, Bill Gates and Sam Altman, among others.

The article describes KoBold’s work on a large copper deposit in Zambia and its exploration plans in the Democratic Republic of Congo. But the Zambia discovery claim has not been independently assessed in a report released by the private company, the article says. Some mining executives quoted by Berkeleyside question how much the AI-assisted approach contributed, and point to the difficulty and expense of extracting deep deposits.

Berkeleyside notes that the article was first published on 23 June 2026. Its account also sets KoBold’s work against the strategic importance of mineral supplies for batteries, energy technologies and AI data centres.

Why it matters

AI here is not a chatbot bolted onto a familiar task. It is being put forward as a way to search a vast, fragmented geological record for resources that are expensive and difficult to find. If that helps identify viable deposits, the payoff could reach well beyond mining. If the discoveries are hard to extract or the technology’s contribution is overstated, the valuation and the grand claims deserve a closer look.

Our read

KoBold’s approach is a genuinely interesting use of computing, but “AI found it” is not the same as proving the deposit can be mined profitably. The distinction matters when the evidence is a company’s claim and the independent assessment has not been published. Berkeleyside’s reporting gives readers the useful tension: a potentially valuable exploration method, a strategically important resource, and major unanswered questions about what comes out of the ground.

What to watch

• Whether KoBold publishes an independent assessment of its Zambia discovery claims.
• Evidence of the technology’s contribution to identifying deposits, rather than simply analysing known geological data.
• The cost, engineering and political challenges of bringing deep deposits into production.
• Whether the company’s exploration results translate into operating mines.

Discussion spark: What should count as convincing proof that AI helped discover a mineral deposit: a company’s exploration results, independent geological evidence, or a mine that can actually be built?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.