Bernstein debunks 3 copper demand myths tied to data centers
Bernstein argues that copper demand from AI data centers may be overestimated, as efficiency gains and voltage shifts could cap long-term consumption despite capacity growth.
Bernstein has challenged three common misconceptions about copper demand from data centers, suggesting that expectations tied to AI-driven expansion may be overstated. The firm projects global data center capacity could approach 200 gigawatts by 2030, but advances in power distribution technology are expected to reduce copper intensity per facility. Copper is a key industrial metal used extensively in electrical wiring, and data centers require significant amounts for power distribution, grounding, and cooling systems. However, Bernstein argues that efficiency gains will curb the metal's demand growth.
For traders, this reassessment matters because copper prices have been partly supported by narratives of surging demand from the tech sector. If Bernstein's analysis proves correct, the long-term demand outlook for copper could soften, potentially weighing on prices. Copper futures on the London Metal Exchange (LME) have been volatile amid AI hype, and a demand slowdown could shift market sentiment. Current copper pricing can be tracked on NowPrice's commodities page for real-time context.
Looking ahead, the key factor is the industry's expected shift toward 800-volt direct-current (800VDC) architecture, particularly in AI-focused facilities. This transition could significantly lower copper usage per megawatt, meaning that even as capacity grows, total copper demand may plateau or decline after the early 2030s. Traders should monitor adoption rates of 800VDC and broader data center capex trends to gauge future demand. Additionally, copper supply constraints from mines could interact with demand shifts, making the balance critical for price direction.