Summary
Bernstein's analysis suggests that forecasts tying surging AI and data center buildouts directly to a significant rise in copper demand may be overstated. Although global data center capacity could approach 200 gigawatts by 2030, improvements in power-distribution architecture and shifts in how operators source electricity are expected to materially reduce copper intensity per site, muting the net impact on copper consumption.
Data center capacity versus copper intensity
The report projects global data center capacity climbing toward 200 GW by 2030. Crucially, Bernstein expects the industry to adopt higher-voltage direct-current architectures - notably 800-volt direct-current (800VDC) systems - especially in AI-focused facilities. Because higher-voltage systems carry less current for the same power transfer, they can lower copper requirements significantly. Bernstein estimates that the transition to such architectures could cut copper usage per installation by as much as 45% compared with existing infrastructure.
As a result, the firm projects copper demand attributable to data centers will remain in the range of about 400,000 to 500,000 tonnes annually over the next several years and then decline in the 2030s even as aggregate capacity continues to expand.
Power sourcing and implications for grid copper demand
Bernstein also flags a weaker-than-expected link between data center expansion and transmission-and-distribution copper requirements. Long grid interconnection delays in the United States and Europe are prompting many operators to pursue behind-the-meter generation solutions. That approach reduces immediate needs for large-scale transmission and distribution investment tied to new facilities, and thus lowers short-term copper demand from grid upgrades.
In parallel, energy storage systems are emerging as an increasingly important source of future copper demand. Bernstein's forecast sees global energy storage capacity rising from roughly 550 gigawatt-hours in 2025 to about 1,500 gigawatt-hours by 2030, supported by more renewable deployment and stronger demand for grid flexibility.
Supply-side adjustments and market outlook
On the supply front, Bernstein upgraded its view on certain projects - notably BHP's Escondida OGP2 and First Quantum's Cobre Panama - while adopting a more cautious stance toward Barrick Mining's Reko Diq development. The report maintains a forecast of a copper market deficit in 2026, followed by a broadly balanced market through 2029. From 2030 onward, deficits are expected to widen as electrification-related demand outpaces supply growth.
Bernstein also raised price targets for Antofagasta and Freeport-McMoRan while keeping Market Perform ratings on both companies, noting that copper-focused miners are trading at meaningful premiums to historical valuation averages.
Implications for markets and sectors
The analysis has implications across several sectors: data center operators and cloud service providers may benefit from lower capital intensity per gigawatt of capacity; power grid planners could face altered timing and scale of copper demand; energy storage and renewables stand to drive a growing share of copper consumption; and miners face a shifting demand profile with near-term balancing and longer-term deficits that influence pricing and valuation.
Concluding observation
Bernstein's view reframes the narrative that raw copper demand will surge in lockstep with AI-driven data center growth. Instead, the firm emphasizes efficiency gains, alternative power sourcing and the rising role of storage as the key drivers shaping copper demand through the remainder of the decade.