As OpenAI prepares to set up shop on Richmond’s waterfront, the company’s co-founder Sam Altman and other billionaires funding the artificial intelligence revolution are pouring money into a Berkeley-based firm that uses AI to hunt for the raw materials it runs on.
Ever since KoBold Metals went looking for capital, billionaires have flooded it with cash. Jeff Bezos, Bill Gates, Mark Zuckerberg, Altman along with Mark Andreessen, Michael Bloomberg and Jack Ma, are among KoBold’s many mega-rich backers. President Donald Trump’s administration has tried to bolster KoBold’s bottom line.
The rush of investment has driven KoBold’s valuation to nearly $3 billion, making it the second-most valuable tech company in Berkeley, after Rigetti, a quantum computing firm.
It’s spurred by the company’s use of artificial intelligence to search for critical minerals like copper, lithium, nickel and cobalt, essential elements in the building of electric vehicles, wind turbines, robots, fighter jets, batteries, and — of special interest to many of those billionaires — AI data centers.
Sam altman’s company coming to Richmond
In April, KoBold broke ground in Zambia on what may be one of the world’s largest copper deposits, a mineral that is needed in enormous quantities for hyperscale AI data centers that companies like OpenAI are racing to build. It is one of the company’s 60 projects in 14 countries across four continents.
“We invested in KoBold because their full-stack AI combines decades-old geologic information with new data sources … to find new deposits of critical battery materials,” said Altman, whose company OpenAI is part of a consortium investing $500 billion to build new U.S. data centers, said in a testimonial on KoBold’s website. “No one else in the world is doing anything like it.”
KoBold’s billionaire backers are gung-ho about the company’s prospects. But can the company deliver on its promises?
Not everyone is convinced. Some longtime mining executives question whether KoBold’s much-hyped AI-assisted technology was truly responsible for discovering the new copper vein in Zambia, which sits next to another copper lode. They point out that there are enormous technical challenges to retrieving ore that lies a mile underground in one of the country’s wettest regions. And since KoBold, a private company, has not yet released an independent report assessing its claims, they cannot be verified.
“I don’t have reason to question whether the copper is there,” said Mothusi Pahl, a consultant who spent decades in Africa building and operating power projects for mining operations in Zambia, Mozambique and South Africa. “But the complex cost and energy requirements to get that copper out of the ground, and the political and economic risk associated with that work, are where the complexity is, and that’s the part being glossed over.”
KoBold’s growth benefits from U.S. competition with China over critical minerals
While Altman’s assertion that KoBold’s approach is unparalleled is a bit hyperbolic — there are other companies leveraging new technologies to find mineral deposits — it reflects the urgent sense that the U.S. needs to better control its critical mineral supply and KoBold may be one way to do it.
The United States and China are locked in a global battle to dominate the extraction and production of critical materials that will fuel the next energy revolution. And so far, China is winning. The U.S. produces only about 14% of the world’s critical materials, and Australia, 6%. China, in contrast, controls 60% to 90% of global mining and refining of critical minerals. It has used its dominance for international leverage.
“Deng Xiaoping [the leader of China from 1978 to 1989] is famous for having said OPEC has oil, but China has rare earths,” said John Pomfret, who spent years in China as a Washington Post correspondent and now works for Garnault Global, a strategic risk advisory firm. “What China did, starting in the 1980s but then accelerating in the 1990s and 2000s, was to move to dominate completely, not simply the mining of rare earths, but also the whole supply chain.”
The U.S. and China battle reached a new intensity in April 2025 and has accelerated since then. In response to Trump’s decision to increase tariffs, China restricted exports of seven rare-earth minerals as well as magnets made from them. That chokehold forced auto makers Ford and Suzuki to temporarily shut down some auto plants because of a lack of materials. On Monday, China barred 10 American companies from purchasing other rare earths. “Not since the Arab oil embargo in late 1973 and early 1974 had the United States experienced such a drastic loss of access to critical minerals,” a reporter for the New York Times wrote in December.
The trade war has prompted the Trump administration to prioritize increasing U.S. control over the production of rare-earth minerals. It has tied foreign aid and military support to access to minerals, promised hundreds of millions of dollars to build infrastructure to bring those minerals to market, and encouraged Western companies to invest more in mining exploration. In December, the State Department launched Pax Silica, an international effort to compete with China by establishing more reliable supply chains for AI-enabling technologies.
Trump administration’s pursuit of minerals linked to KoBold in Greenland, Zambia, Democratic Republic of Congo
When Bezos, Zuckerberg, Elon Musk and Google’s Sundar Pichai were seated in the second row at Trump’s inauguration in January 2025, right behind his family members and in front of his Cabinet picks, it signaled the importance of the tech industry to the incoming president. Trump later stood at a podium with Altman to announce the launch of the multibillion-dollar push to build more AI data centers, and appointed two of KoBold’s investors, Zuckerberg and Andreessen, to serve on the President’s Council on Science and Technology.
Days later, Trump renewed his call for the U.S. to take over Greenland, an action necessary, he said, to thwart Russian and Chinese influence in the Arctic. But Trump also acknowledged the U.S. wanted access to Greenland’s rich deposits of rare-earth materials, and later made that access part of the framework of a deal he worked out with NATO over the future of the island. Observers noted that many of the men attending Trump’s inauguration stood to benefit from increased U.S. control of Greenland.
“Our new president’s obsession with buying Greenland … looks less like a joke and more like a big handout for the tech companies that backed him during the election,” wrote a reporter for the tech news site Gizmodo.
KoBold had deployed its AI technology to look for minerals in Greenland, but sold its stake in the Disko-Nuussuaq nickel-copper-cobalt project in 2022 and no longer had an investment there. However, the company could still collect future royalties from the country.
The White House also began pushing policies that would help KoBold’s operations in Africa. During a January 2026 White House meeting, administration officials urged an Australian company to sell its interests in a major lithium mine in the Democratic Republic of Congo (DRC) to KoBold, Bloomberg News reported. Lithium is critical for the manufacturing of batteries for electric vehicles, drones and military communications equipment.
In March, the State Department mulled denying drugs to people with H.I.V. in Zambia, where KoBold has a huge operation, to force the African country to agree to give the U.S. more access to its copper and other critical minerals. It also announced that the U.S. would spend $250 million in foreign assistance to increase investments in critical mineral extractions.
One of those strategic alliances was with the Democratic Republic of Congo. The U.S. said it would offer investment and military help to DRC President Felix Tshisekedi, who is combating Rwanda-backed rebel groups and seeking an unconstitutional third term, in exchange for access to DRC minerals.
In April, just as Ebola began another deadly march through the country, KoBold announced what it called the largest lithium exploration in history. It will use its proprietary technology to explore 13 licensed sites spread over 3,000 square kilometers.
“This happened because of the courage and vision of the DRC government and the tireless work of the U.S. State Department to create the conditions for American companies to succeed here,” Kurt House, the company’s CEO, said in a press release published in the trade publication Mining Engineering.
KoBold declined to make anyone available to speak with Berkeleyside, Richmondside’s sister publication, for this story. “We’re heads down and focused on the work at hand, so we won’t be able to join you for an interview,” Justin Hamilton, KoBold’s PR consultant, wrote in an email. He later answered some questions submitted in writing.
KoBold’s founders devised new AI tools to locate mineral deposits and Silicon Valley money followed
Kurt House and Josh Goldman met as doctoral students at Harvard. The men were both “energy nerds,” Goldman said on the podcast Pioneers of AI, and used to take field trips to power plants and coal mines for fun. Goldman went on to work as a consultant to oil and gas companies and write software that scoured large datasets to find untapped deposits. House ran a company focused on carbon sequestration and taught about entrepreneurship at Stanford University.
In 2014, they teamed up with Jeff Jurinak, a veteran of petroleum exploration and production, to form Phase Change Resources, a private equity company that invested in oil and gas. That lasted just a few years.
“After doing that for a short while, we decided we wanted to … start working on the materials of the future,” said Goldman, now KoBold’s president.
The men realized that the market for critical earth materials was enormous, yet current mining methods could not keep up with demand. House predicted on the Volts podcast that there would be two billion electric vehicles on the planet by mid-century, meaning the need for lithium could increase by 4,000%. The world will need $10 trillion to $15 trillion in rare earth minerals to meet future needs, he said.
“Any way you slice it, those are just gigantic numbers, and they require gigantic amounts of lithium, cobalt, copper and nickel,” said House. “[If] that’s how much we need … to be mostly off fossil fuels by 2050, how much exists in the reserves of current mines?”
House, Goldman and Jurniak (who has since left the company) determined that the world will need to find about 1,000 new mines to meet future demand, and currently only a handful are developed a year. Most exploration projects fail to become mines and the industry is spending three times more to make 60% fewer discoveries than 30 years ago, according to a paper by an Australian mining consulting group.
The trio decided to apply modern computing methods to mineral exploration, in part through the use of artificial intelligence. They named the company for a creature in Germanic folklore that cursed silver miners and poisoned their ore. The company’s goal is “turning exploration from a slow, manual process into a fast, data-driven repeatable science,” according to its website.
For centuries, people discovered ore deposits by looking for outcroppings protruding from the earth’s surface and then mining there. But most of the earth’s ore deposits are below ground, making them difficult and expensive to find.
There is already a tremendous amount of information about the Earth’s crust, but it is scattered around the globe, documented in old maps, handwritten surveys and century-old government reports, and difficult to analyze. One of KoBold’s first initiatives was to quantify that material. The founders frontloaded the company not with mining experts but with data scientists and software engineers from places like Google, Apple and Meta. They worked with geoscientists to develop algorithms and technologies to assemble and analyze large datasets.
To further narrow down the location of mineral deposits, KoBold invented the Hyperpod, which provides three-dimensional information about the Earth’s magnetic field using its high-resolution camera and other sensors. KoBold mounts it to the side of a helicopter or small airplane and flies over areas it believes might have ore deposits. The terabytes of new information are used to develop algorithms that detect geological patterns indicative of potential mineral deposits.
“We don’t drill for metals, we drill for information,” House told the New York Times. “It puts the science into eureka.”
KoBold’s flagship mine in Zambia is a big bet that some experts say could be too expensive to operate, though copper prices are rising
Not everyone is bullish on using AI to revolutionize mining. Roland Gotthard, an exploration geologist in Perth, Australia, questioned KoBold’s Series C funding round, which closed in 2025 with $537 million in funding. He expressed concern that AI might misinterpret geological data, leading to misguided predictions.
“In any event, AI won’t help get mines drilled faster once the discovery is made, and it won’t get you through ESG [Environmental, Social, and Governance] and permitting faster,” Gotthard wrote on his LinkedIn page. “We should expect the one or two discoveries will stagnate as usual. Or maybe AI chatbots will convince farmers, indigenous communities and NIMBYs to set aside their objections in some Silicon Valley technofantasy.”
Some critiques have focused on the company’s flagship mine in Zambia.
In 2024, KoBold announced that it had uncovered what may be the largest copper deposit discovered in the country in the last decade — a finding that seemingly confirmed the effectiveness of its AI mining technology.
The Mingomba Mine in the Chililabombwe region is expected to produce 300,000 tons of very high-grade copper a year, according to House, who called it one of “the world’s greatest undeveloped copper deposits” during a late April ceremony to mark the sinking of a shaft. Zambian President Hakainde Hichilema, the U.S. ambassador to Zambia and other notables attended the groundbreaking.
The discovery has raised international expectations about the mine’s impact. KoBold will invest $600 million in Zambia by the end of 2026, “the largest expenditure of private American capital into Zambia since independence,” House said at the dedication ceremony. It has already built a team of about 300 people, most of them Zambian, including its in-country CEO, Mfikeyi Makayi. That investment is expected to grow to more than $2 billion when the mine is built out. Zambia will also profit, as the government-controlled ZCCM-IH, which stands for Zambia Consolidated Copper Mines Investment Holdings, holds a 20% interest in the project. KoBold has hired local suppliers and is funding international studies for some young Zambians.
But the Mingomba Mine faces many challenges. Some mining experts even doubt that KoBold made a new copper discovery since the area has been known as the Copperbelt since the 1920s. KoBold was exploring in a “brownfield,” an established area, not a “greenfield” where ore has never been found. The find is an extension of an existing series of ore bodies that had been previously mapped, said Mike Anglin, an executive with over 30 years of experience in international mine operations, including the discovery of copper deposits in Chile, Peru and Argentina.
“They took the data and then made a big deal about using AI to find this deposit,” said Anglin. “Major mining companies, if they believe they could develop an economically viable deposit, would have dedicated enough exploration geologist time and resources to tell you there was something there to continue exploring.”
Hamilton, KoBold’s spokesperson, responded that other copper mines in the region are concentrated in a small area and were developed after geologists noticed ore protruding from the ground. It took KoBold’s proprietary mining to find the ore hidden deep below the surface, he wrote in an email. “We drilled the two deepest holes in Zambia’s mining history to help define the size and grade of the resource.”
Another issue is that KoBold, a private company, does not have to release an independent report assessing its claims, leaving its investors to rely on what the company tells them. That’s in contrast to publicly traded mining companies, which are required in the U.S., Canada and elsewhere to release third-party reports to validate claims, said Anglin.
KoBold has said it expects to release its engineering study on the copper mine in 2027. “KoBold has drilled more than 150,000 meters at Mingomba,” Hamilton wrote in an email. “Every meter produces core. Every core gets assayed. We have a clear understanding of the mineral resource.”
Mothusi Pahl, the veteran consultant on African mining operations, who previously worked for a start-up backed by one of Bill Gates’ funds, said he wrote a letter of concern to the billionaire in April. Pahl, who knows Gates and says he has no commercial interest in KoBold or in any competing projects, pointed out the challenges the copper mine faces — challenges KoBold has not publicly acknowledged, he said. Since the mine, which will be at least 1,700 meters deep, is situated in one of the wettest regions in Africa, KoBold will have to spend millions on energy costs to pump out water, ventilate the mine and remove the ore.
An adjacent, well-documented mine, Konkola, sank a shaft to 1,505 meters (the deepest on the African continent). The British company operating the mine, Anglo American, had to contend with the Kafue River, which ran through the Konkola mine. The dewatering operation alone consumed an estimated 660 to 900 megawatt-hours of electricity per day, said Pahl. The British company encountered so many obstacles at a time when copper prices were low that it abandoned its stake in 2002.
KoBold’s deeper workings will sit below the existing mine’s water table and will be expensive to mitigate, said Pahl. He believes KoBold has lowballed the true cost of the mine, he told Gates.
“Mining operations get harder the deeper you go, and power is expensive,” Pahl wrote in an email. Pahl estimated that creating a power supply could cost $500 million, raising questions about KoBold’s current $2 billion budget for the mine.
Copper prices, however, have surged almost 50% in the last two years, which could help justify the company’s investment.
Hamilton dismissed Pahl’s concerns. “Underground mining with water management requirements is among the most well-established disciplines in the industry,” he wrote. “The engineering toolkit for this has been refined over decades in some of the most challenging geological environments on Earth.”
If KoBold is overselling its technology and can’t deliver large amounts of copper economically, Zambians will suffer the most, both Pahl and Anglin said. While copper generates 80% of Zambia’s earnings from export, large swaths of the Copperbelt are marred by abandoned open pit mines. (In 2025, a Chinese-operated copper mine released 13 million gallons of toxic waste into the Kafue River, just one incident of tailings polluting water supplies. Farmers in Zambia have filed an $80 billion lawsuit against the Chinese companies involved.)
“President Hichilema has personally positioned Mingomba as a centerpiece of Zambia’s economic recovery and KoBold’s stated production targets have been incorporated into Zambia’s national 2031 copper output goals,” wrote Pahl. “A high-profile failure here, backed by globally recognized technology investors, risks damaging the mining community AND the political economy of southern Africa at a global scale.”
Anglin, who worked with Pahl on the letter to Gates, agrees.
“One thing that drives me nuts is promoters and the damage they do to the industry,” he said. When promoters “make a lot of noise and then can’t deliver [that] damages the industry because investors then back off.”
Companies like KoBold, he said, take risks that scare off traditional mining investors. So they have “to go to non-traditional mining investors, you know, the Bezoses, the Gates[es], the whoevers. And they have to use the ‘We’re smarter,’ the ‘Silicon Valley’ label.”
