Artificial Intelligence Expands Demand for Critical Minerals and Energy

The expansion of artificial intelligence does not depend solely on algorithms and computing capacity. Behind the models, chips, and data centers lies a physical infrastructure that requires critical minerals, energy, water, and large land areas.

Growth in training compute drives demand for copper, lithium, gallium, and rare earths. According to analysis presented by engineer Eduardo Barrera, data centers may require between 27 and 33 tons of copper per megawatt, a figure that can reach 47 tons in facilities dedicated to intensive training tasks. If transmission lines and substations are incorporated, total demand could reach between 100 and 150 tons per megawatt.

Gallium is used in high-performance chips and its demand linked to data centers could exceed 10% of global supply by 2030. Meanwhile, cooling systems employ neodymium-iron-boron magnets, associated with rare earths, whose demand would grow about 7.5% annually through 2040. Energy storage related to artificial intelligence could also reach some 300 gigawatt-hours by 2030.

The impact extends to water consumption and land use. AI data centers used approximately one trillion liters of water during 2025, while larger-scale campuses can occupy from tens to thousands of hectares. Examples mentioned include Google facilities in Maryland, Meta in Louisiana, and the Stargate project in Texas.

The relationship between both activities is not solely one of competition for capital and resources. Artificial intelligence can also contribute to mining through tools to accelerate exploration, process large volumes of information, and improve operational efficiency. Thus, a technology that increases pressure on mining supply chains could also help enable more precise resource utilization.

Information from Panorama Minero.