- Copper prices have risen more than 60% since April 2025, and the International Copper Study Group forecasts a global refined copper deficit of roughly 150,000 metric tons in 2026, driven by AI data center construction requiring power densities far above traditional facilities.
- J.P. Morgan Global Research estimates lithium demand will rise 16% year-over-year in 2026, with EVs accounting for 58% of that incremental demand, while UNCTAD projects lithium demand will increase by more than 350% by 2040.
- China’s nonferrous metals sector reported 117.1% year-on-year profit growth in the first five months of 2026, reflecting elevated copper and aluminum prices tied to AI infrastructure and new-energy materials demand.
AI data centers are now one of the most copper-intensive construction projects on the planet, and mines cannot be opened fast enough to keep up. Copper prices have climbed more than 60% since April 2025, driven by rack densities that can reach 130 to 300 kilowatts per rack, far above what traditional facilities require. The supply gap is structural: new mines typically need development timelines exceeding a decade, which means the deficit the International Copper Study Group forecasts for 2026 is not a short-term pricing anomaly.
AI Infrastructure Ignites Copper Demand
The numbers behind that deficit are specific. J.P. Morgan Global Research estimates copper demand from data centers will reach 475,000 metric tons in 2026, up 110,000 tons year-over-year. A June 2026 Crux Investor report noted that new mine development requires multi-million-dollar capital commitments alongside those decade-plus timelines, making supply highly inelastic relative to demand. The International Energy Agency projects data center power demand will rise 165% by 2030, which compounds the pressure on every metal in the supply chain, not just copper.
The policy and procurement implications are real. Enterprises building or expanding data center capacity are now competing for the same copper supply as grid operators and EV manufacturers. Procurement teams that treated copper as a commodity input are increasingly facing allocation constraints more typical of semiconductor supply chains.
New Energy Transition Accelerates Broader Metal Needs
AI infrastructure is one pressure on the metals market. The energy transition is another, and the two are compounding. The global non-ferrous metals market is projected to grow from $1.23 trillion in 2025 to $1.77 trillion by 2034, a compound annual growth rate of 4.03% from 2026 to 2034. Copper, aluminum, nickel and lithium are the primary drivers, each tied to a different layer of the transition: grid electrification, EV manufacturing, battery storage and renewable generation.
Lithium illustrates the scale of the long-term shift. J.P. Morgan Global Research forecasts 16% year-over-year demand growth in 2026, with EVs accounting for 58% of that incremental demand. UNCTAD‘s July 2, 2026 report, “The Shifting Dynamics of Critical Minerals Trade,” projects lithium demand rising by more than 350% by 2040, with magnet rare earths up 65% over the same period. Aluminum’s role is quieter but durable: its combination of light weight, corrosion resistance and recyclability makes it a structural material in both EV platforms and renewable energy infrastructure, demand that does not fluctuate with any single technology cycle.
Critical Minerals Face Supply Chain Tensions
Concentration is the central risk. The same UNCTAD report flags that the value chain for critical energy transition materials, lithium, nickel, cobalt and rare earth elements, remains geographically concentrated from mining through processing. China accounts for an estimated 69% of rare earth element production, a figure that shapes procurement decisions for manufacturers across the EV and defence supply chains. That concentration is the reason governments from the US to the EU have moved to diversify sourcing; the dependency is not theoretical.
Cobalt entered 2026 under specific structural pressure: the Democratic Republic of Congo’s export quota system is capping shipments and tightening the market at a point when battery demand is still rising. Recycled cobalt is expected to contribute a meaningful but still limited share of global supply by the late 2020s, not enough on its own to offset primary supply constraints. Nickel faces a different problem. The transition toward NMC battery chemistries keeps it essential, but the market is also navigating oversupply in certain grades while higher-purity battery-grade nickel remains tight. These are not uniform commodity markets; each metal has its own supply architecture and failure mode, and enterprise buyers building exposure across several of them face genuinely different risk profiles in each.
For broader context on how concentrated supply chains are shaping enterprise AI infrastructure decisions, see our coverage of the Nvidia chip smuggling investigations and what they reveal about hardware supply vulnerabilities.
Industrial Modernization and Profit Growth
The demand surge is already showing up in industrial earnings. China’s nonferrous metals sector reported 117.1% year-on-year profit growth in the first five months of 2026, reflecting elevated copper and aluminum prices driven by AI-related electronics and new-energy materials. That figure is notable both for its size and its breadth: it covers the sector overall, not a single firm or commodity cycle.
Semiconductor materials are following a related trajectory. The shift toward Wide Bandgap materials, specifically Gallium Nitride and Silicon Carbide, is accelerating in AI data centers, 5G infrastructure and high-efficiency power conversion. GaN-on-Silicon 12-inch wafer production is scaling as manufacturers work to reduce unit costs. High Bandwidth Memory production for AI accelerators is simultaneously driving demand for metal plating and high-purity non-ferrous inputs at the chip packaging layer. These are not niche applications: they are the materials layer underneath the AI hardware buildout, and supply constraints here carry through to inference capacity just as copper constraints carry through to data center construction timelines. Enterprise leaders planning AI infrastructure investments over a three-to-five year horizon are effectively taking a position on metals markets whether they recognise it or not. For more analysis on enterprise AI strategy, visit our Enterprise AI section.



