AI Capex Hitting Energy and Copper Constraints

The AI supercycle infrastructure narrative is colliding with hard supply constraints in two critical areas: copper and electrical capacity. Every megawatt of hyperscale data center infrastructure requires approximately 27 tonnes of copper for transformers, substations, power distribution, cooling systems, grid expansion, and high-capacity cabling. Copper has rallied to $13,619 per tonne on the London Metal Exchange, a fresh three-month high and only 6% below the January 2025 all-time peak near $14,500. The Copper Market is now described as "starting to look tight again," with timing that coincides precisely with the second quarter AI infrastructure deployment cycle.

China's Zhaojin Mining, one of the world's largest gold and copper miners, is actively scouting for acquisitions in Africa and Central Asia to scale copper and gold assets, particularly those divested by Western firms. This signals that major miners see multi-year supply growth as essential and are willing to move capital internationally to access it. On the energy side, SoftBank invested billions in AI data center battery backup and storage; Voltagrid, an energy infrastructure startup, raised $1 billion from Blackstone and Halliburton and is valued above $10 billion. These are signs that energy margins are being squeezed and that grid upgrades are now existential requirements for AI capex planning.

The cross-asset implications are significant. Copper miners and battery/energy storage companies are likely to see sustained capex tailwinds, but rising input costs could compress semiconductor manufacturer margins if copper input prices remain elevated. Energy companies and utilities are gaining embedded option value as data center operators bid for reliable power. Materials inflationThe rate at which prices rise across an economy. pressures that are independent of Fed policy could keep long-term yields sticky. Gold, silver, and hard commodity miners are all rallying on both inflation hedging and AI copper demand.