-1-1.png?height=120&name=Curation%20Logo%20Transparent%20(1)-1-1.png)
The Pentagon has transferred its AI-powered critical minerals pricing initiative to a private nonprofit, the Critical Minerals Forum (CMF), to improve supply chain transparency and secure Western deals. Backed by over 30 members including Volkswagen, MP Materials and RTX, the CMF uses government-sourced data to reduce reliance on China-dominated markets. The programme aims to forecast long-term pricing by factoring out distortions like overproduction and tariffs. Nevada is among the first to collaborate, eyeing copper smelting projects. (Reuters)
Why does this matter? The trading of critical metals is notoriously complex with global supply chains for these materials often opaque, volatile, and subject to various risks, including geopolitical events and market disruptions. The recent trade war has further complicated the market. The AI model seeks to bring clarity, reflecting "one of the boldest efforts to date to transform the ways certain metals are bought and sold", as reported in Reuters.
The CMF's pricing model arrives at a pivotal moment, as the US rushes to reduce its reliance on China, which dominates the rare earths market – producing 61% and processing 92% of global supply. Between 2020 and 2023, the US imported 70% of its rare earth compounds and metals from China when, in April, China restricted US exports of 7 "heavy" rare earths crucial for defence technologies, squeezing US supply. Meanwhile, prices for nickel, cobalt, lithium, and other battery metals have been distorted by Chinese overproduction, often operating at a loss in Indonesia and the Congo, to secure market dominance. In response, Washington has invoked emergency powers and forged new mineral deals, including one with Ukraine.
The Pentagon’s AI model draws on over 70 mining-related datasets to forecast the supply and pricing of critical metals over the next 15 years. Using public and proprietary information from data providers like FactSet, Benchmark Mineral Intelligence, and the US Commerce Department, the model calculates what a metal will cost when labour, processing and other costs are factored in. Further, the model will anticipate how disruptive market events - such as export restrictions or tariffs - could influence price and supply. The model aims to assure the buyer and seller of the economics of a metal deal, now and in the future.
The model will be managed by a team of fewer than ten people and is supported by the Defense Advanced Research Projects Agency (DARPA), which has committed funding through at least 2029 to support its development and application in strategic investment planning.
While many believe the model will bring some clarity to an opaque market, some disagree – "I've tried to politely say I think this is worthless," said Ian Lange, a mining economics researcher at the Colorado School of Mines, "can we predict the price of oil better now than five years ago? The answer is no. Machine learning doesn't help".
AI is set to have an ever-increasing role in the mining sector. For example, AI has been utilised to analyse geological data, identifying high-potential mining zones. Last year, KoBold, backed by Bill Gates and Jeff Bezos, used this technology to discover a major copper deposit in Zambia. Predictive maintenance is another area AI is optimising; here, mining equipment failures are forecast before they happen, reducing costly downtime. Through adopting this technology, Anglo reported a 75% drop in unplanned outages.