Summary
- AI model aims to promote deals by offering more certainty on future pricing
- Pentagon will continue to fund the AI program through at least 2029
- More than 30 companies have joined non-profit Critical Minerals Forum
(Reuters) – A U.S. government-created artificial intelligence program that aims to predict the supply and price of critical minerals has been transferred to the control of a non-profit organization that is helping miners and manufacturers strike supply deals.
Launched in late 2023 by the U.S. Department of Defense, the Open Price Exploration for National Security AI metals program is an attempt to counter China’s sweeping control of the critical minerals sector, as Reuters reported last year
Now, more than 30 mining companies, manufacturers and investors – including auto giant Volkswagen – have joined the Critical Minerals Forum non-profit and will be its first users, according to Rob Strayer, a former U.S. diplomat and the organization’s president.
“Everyone in the critical minerals sector is looking for more price transparency,” said Seth Goldstein, a lithium industry analyst with Morningstar. “Any tool like the CMF that could help would be welcome.”
Other members include copper miner South32 rare earths producer MP Materials and defense contractor RTX. The CMF held its first meeting with members in November. The privatization and CMF’s membership have not previously been reported.
Armed with the AI model, the CMF aims to help manufacturers curb their reliance on China by signing more metal supply deals with Western mines, according to more than two dozen industry consultants, purchasing agents, analysts, regulators and investors who told Reuters the program reflects one of the boldest efforts to date to transform the ways certain metals are bought and sold.
The goal is for the AI model to calculate what a metal should cost when labor, processing and other costs are factored in – and Chinese market manipulation is factored out – and thus give buyer and seller confidence in a deal’s economics.
Some deals with the CMF are beginning to take shape. Nevada officials this week said they would work with the CMF and its AI model to help attract copper smelting to the state. The U.S. has only two copper smelters and as such imports nearly half of its demand for the red metal.
The program has already faced skepticism over whether it can achieve the goal of transforming the long-established ways metals are bought and sold.
Yet it is aimed less at heavily traded metals – such as aluminum – and toward lightly traded metals or metals that see heavy overproduction from some in an attempt to sway market pricing.
For example, the CMF model could help manufacturers forecast available nickel supplies in 2028 if the U.S. were to impose a 100% tariff on that metal from Indonesia, the top global producer.
That data that could help a manufacturer determine whether to invest in a U.S. nickel mine or agree to buy its future production, a step that would help obtain financing for a mine’s construction.
In such a scenario, the nickel buyer would use the AI model’s data to negotiate a long-term deal for guaranteed supply, regardless of whether Chinese miners boost production and drive down market prices, as they have done in recent years.
The CMF’s aim with the AI model does assume that a buyer would be comfortable paying more than the market price for a metal if supply were guaranteed.
CHINA SQUEEZE
The CMF’s entrance into the complex metals markets comes as Beijing restricts critical minerals exports, the very kind of market interference that the CMF officials said underscores the need to build more U.S. mines and processing facilities to power the energy transition.
Prices on the London Metal Exchange and other futures exchanges for nickel, cobalt and some other battery metals have been dominated in recent years by overproduction from Chinese miners operating at a loss in Indonesia and Congo to boost market share.
Many niche-but-essential battery minerals on which Beijing has imposed export controls are not traded or lightly traded, including rare earths – a group of 17 metals used to make magnets that turn power into motion – as well as germanium and gallium.
In response to a request for comment about the CMF, the Chinese embassy in Washington, D.C., said that China manages its exports of rare earths in accordance with rules from the World Trade Organization.
“China will continue to work with other countries to jointly undertake the responsibility of global rare earths supply,” said embassy spokesperson Liu Pengyu.
Volkswagen and some other CMF members said they see the CMF as helping boost visibility into what can be an opaque critical minerals supply chain. MP Materials and RTX did not respond to requests for comment.
U.S. President Donald Trump has already ordered his administration to work with private developers to boost U.S. crucial minerals production, a step that could be aided by the data CMF aims to provide markets, program officials said. The president has also launched a study into potential tariffs on all U.S. minerals imports.
Drawing on its government connections, the CMF aims to connect mining projects with potential investors and manufacturers needing more-secure metals supply, said Strayer.
Massachusetts-based rare earths processing startup Phoenix Tailings hopes the CMF can help create U.S.-based prices for minerals tied to actual production costs, said CEO Nick Myers.
Phoenix aims to use data from the CMF as negotiating leverage with potential customers, including manufacturers that are themselves CMF members, Myers said. “In a sector that is opaque, it is one of the tools to get more information,” Myers said.
Not all market observers are convinced that the CMF’s AI model is revolutionary.
“I’ve tried to politely say I think this is worthless,” said Ian Lange, who teaches mining economics at the Colorado School of Mines. Lange contrasted the goals of the Pentagon’s AI model with the much-larger and more-complex global oil market.
“Can we predict the price of oil better now than five years ago? The answer is no. Machine learning doesn’t help,” Lange said.
‘ENCOURAGE MORE VISIBILITY’
The Pentagon’s AI model is being trained using more than 70 mining-related data sets and aims to guide investment decisions out for at least 15 years based on how unexpected market shocks – export restrictions, for example – could affect the production or price of a metal.
FactSet Benchmark Mineral Intelligence and other pricing providers are supplying data, as is the U.S. Commerce Department, officials said.
It is access to analysis of that data – some of which is not public – that the CMF says it believes sets the Pentagon’s AI program apart from ChatGPT or other AI programs.
And that data is the CMF’s biggest cost, part of the reason why the Pentagon’s Defense Advanced Research Projects Agency (DARPA) will fund it for the next few years while the CMF determines whether to charge all members or create a tiered structure with basic members getting free access and others paying for more granular data, officials said.
S&P Global AI developer Charles River Analytics, and software firm Exiger with price reporting agency partner Metal Miner have developed the model, according to the Pentagon.
S&P Global declined to comment. Charles River Analytics did not respond to a request for comment. Exiger said it believes its data can help forecast a material’s cost and availability and boost supply chain visibility.
The CMF has been organized as a nonprofit trade association with a board of directors comprised of its members. Its staffing is small – fewer than 10 employees – and its annual budget is not disclosed.
DARPA does not have a representative on the CMF board, but is funding the program through at least 2029 and plans to transfer the AI model’s intellectual property to the CMF by the beginning of 2027, officials said.
There are no plans to make the CMF a for-profit entity, although there may be charges in the future for access to more detailed data sets, officials said.
The CMF is launching a campaign to attract more members – especially from the semiconductor, aviation and defense industries – and offering free membership for the next 14 months while the Pentagon funds data collection, Strayer said.
Foreign governments are also studying whether to join the CMF and use its data, including copper-rich Zambia and cobalt-rich Democratic Republic of Congo, CMF officials said, adding they aim to make the program international in scope to boost metals market transparency.
The Zambian and DRC embassies in Washington, D.C., did not respond to requests for comment.
As Western miners begin to demand green premiums for their metals, those new agreements increasingly require the very market intelligence the CMF model aims to provide.
“Any mechanism that can give you better modeling of markets is obviously enormously valuable,” said Brian Menell, CEO of TechMet, a mining investor and CMF member.
The AI model introduces another variable for the LME to contend with, especially as the exchange is struggling as rivals in Chicago and Shanghai try to take market share for some niche battery metals.
The LME declined to comment.
Reporting by Ernest Scheyder; Editing by Veronica Brown and Claudia Parsons
Source: worldautoforum.com
