- AI data-center buildouts are driving a sharp rise in electricity use, boosting demand for copper in power delivery, cooling, and grid expansion.
- Forecasts project global copper demand rising about 50% by 2040, with data-center infrastructure requiring over one million tons per year in key markets.
- Companies are moving to secure lower-carbon, domestic copper supplies (e.g., AWS with Rio Tinto's Nuton project), but near-term output is small versus expected demand.
- Palihapitiya's view of copper as the hidden AI investment is supported by demand trends, though permitting, supply constraints, and grid bottlenecks could delay gains.
Read More
The current AI infrastructure boom has sharply increased focus on copper as an essential material, beyond just its traditional uses in wiring or electronics: it is vital for power delivery, cooling, chip interconnects, and grid expansion. Expert forecasts and recent corporate moves jointly validate the idea that copper’s demand profile is entering a secular upswing.
S&P Global’s “Copper in the Age of AI” report projects global copper use will rise roughly 50 % by 2040—from ~28 to ~42 million metric tons—driven by four vectors: core economic demand, the energy transition, AI/data centers, and defense modernization. Data centers alone could take up a rising share of national electricity demand in the U.S., increasing from about 5 % today toward as much as 14 % by 2030, which in turn increases pressure on copper supply for both infrastructure and compute hardware.
On the energy demand side, U.S. data centers consumed approximately 183 TWh in 2024—about 4 % of national electricity consumption—and that number is forecast to rise 133 % to ~426 TWh by 2030 under IEA’s “base case” scenario, with AI-heightened workloads making the demand more intense and continuous[0search0]. Parallel forecasts suggest U.S. data center grid power demand will hit ~134.4 GW by 2030, nearly triple present levels, driven by hyperscalers and co-location operators.
Corporate supply strategies are already reacting. Amazon Web Services recently committed to buying low-carbon copper from Rio Tinto’s Nuton bioleaching project at the Johnson Camp mine in Arizona, aiming for reduced environmental footprint and closer supply proximity. This “first Nuton copper” will be used in U.S. data centers and AWS is assisting with analytics and process optimization. However, the four-year expected production (~30,000 tons) is modest relative to forecasts of millions of tons in demand.
Nevertheless, several strategic risk factors could disrupt copper’s path to “parabolic” gains. These include: permit and environmental delays for mining and processing; the capital-intensive nature and long lead times of mining investments; potential bottlenecks in grid capacity and electric power interconnection; inflationary pressures from energy and materials; and competitive pressures from alternative materials or cooling technologies. Moreover, electricity demand forecasts may be overstated in some jurisdictions due to speculative data center projects[new19].
For investors, this signals a few strategic implications: exposure to copper miners, especially those emphasizing low-carbon production or domestic supply, could benefit. Similarly, copper supply chain innovations (bioleaching, modular extraction) may become differentiators. ETFs or direct copper price play should account for tight global supply projections, but also factor in carbon regulation, trade policy, and technological substitution risks. As Palihapitiya points out, going beyond chip-makers and hyperscalers to the material inputs could capture outsized returns—but it requires patience and careful risk weighting.
Open questions that remain include: how quickly new copper mining projects can be permitted and brought online; how much of the projected copper demand is substitutable with other materials; how effectively energy grids can scale in capacity and emissions performance; and how carbon and environmental regulation (especially in the U.S. and Europe) will shape supply costs.
Supporting Notes
- U.S. data centers consumed ~183 TWh in 2024, or over 4 % of U.S. electricity; projected to rise 133 % to ~426 TWh by 2030 under conservative IEA estimates[0search0].
- Grid power demand from U.S. data centers is expected to reach 134.4 GW by 2030—nearly three times current levels; states like Virginia and Texas are among the highest-growth areas.
- Global electricity demand from data centers is expected to double to ~945 TWh by 2030; U.S. data centers will drive nearly half of that growth.
- S&P Global forecasts global copper demand will grow from ~28 million metric tons today to ~42 million by 2040, driven by AI, energy transition, core economic demand, and defense; yet a supply shortfall of ~10 million tons per year could emerge without new capacity.
- North American copper demand related solely to data center build-out could increase by 1.1-2.4 million tons by 2030, tied to rising power capacity and infrastructure supporting hyperscale facilities.
- Amazon’s AWS has committed to purchase copper from Rio Tinto’s Nuton project, which produces 99.99 % pure copper via bioleaching; the partnership expects about 30,000 tons of refined output over four years, with substantially lower carbon and water intensity than traditional mining.
- Electricity demand growth from data centers and EVs could add ~290 TWh by 2030 in the U.S.—a more than doubling of data center-related electricity load since 2023.
- Electricity rates for consumers are at risk of rising significantly—utilities may carry the cost of new infrastructure tied to speculative data center demand; in some states, regulatory reforms are being introduced to shift risk away from ratepayers[new19].