Copper nears record high as AI infrastructure boom drives demand

Three-month copper futures on the London Metal Exchange (LME), climbed to $13,791 per metric ton, according to Nikkei — just below the all-time record of $14,527.50 set on January 29, 2026. That day saw the sharpest single-day price jump since 2008, nearly hitting the LME's 12% daily trading limit.

Copper demand is directly fueled by the AI infrastructure boom: the metal is the backbone of the power grids that keep AI data centers running. Per the Copper Development Association, a single hyperscale AI data center can consume up to 50,000 tons of copper, compared with 5,000-15,000 tons for a conventional data center. Electric vehicles add a separate source of demand: estimates vary, but EVs typically require 3-4 times more copper than internal combustion engine cars — around 72 kg of copper per EV versus roughly 25 kg for a gasoline equivalent.

• Current price: $13,791 a ton, near the $14,527.50 record set January 29, 2026
• A hyperscale AI data center uses up to 50,000 tons of copper versus 5,000-15,000 tons for a conventional one
• An EV requires 3-4 times more copper on average than a gasoline-powered car

For the crypto market, this copper crunch isn't an abstract commodities story: the same grid and data-center infrastructure is needed by companies increasingly converting mining capacity into AI hosting. We previously covered how Galaxy Digital opened a second Texas data center under its Merlin project, and how Quantum Solutions and Hyperscale Data sold ETH and BTC reserves to fund data center construction — rising copper prices directly raise the capital costs of exactly these projects, where power grids and cooling already eat up the largest share of the budget.

Forecasts see copper demand from AI infrastructure peaking around 572,000 tons in 2028 — without even factoring in additional pressure from electric vehicles and ordinary economic growth. If copper supply can't keep pace, the resulting shortage will hit both conventional data centers and the former mining farms now mass-converting into AI hosts just as hard.

None of this should be read as personalized investment advice.