AWS signs a two-year supply deal with Rio Tinto to access its Arizona copper mine, the US' first new source in 10+ years, as AI data centers create huge demand
The copper, which is being produced in Arizona by Rio Tinto with bacteria and acid, will be used for data-center construction
Context & Ripple Effects
AWS's agreement turns a construction input into a directly managed part of its AI buildout. It follows reporting that AI-focused facilities require substantially more copper per megawatt than conventional data centers, sharpening the relevance of copper-intensive AI capacity.
The deal also connects cloud expansion to an Arizona industrial base already drawing semiconductor investment, while Rio Tinto's output is described as the first new U.S. copper source in more than a decade.
First-order effects
- AWS gains contracted access to Rio Tinto's Arizona copper for data-center construction over two years, reducing its exposure to spot procurement for that portion of its buildout.
- Rio Tinto adds a named hyperscale buyer for output from its new Arizona source, tying the mine more directly to AI infrastructure demand.
Second-order effects
- Other data-center builders may face greater pressure to secure copper earlier in the construction cycle as large buyers lock in supply; the concern echoes warnings that data-center growth could deepen copper constraints.
- The agreement strengthens the commercial link between U.S. mining capacity and regional data-center development, rather than treating copper as a generic, globally sourced input.
Third-order effects
- If similar contracts proliferate, AI infrastructure procurement could extend beyond chips and power into long-lead industrial materials, making physical supply chains a more visible constraint on deployment schedules.
- The pattern would favor developers and suppliers able to coordinate multi-year capacity planning; whether it materially changes market pricing depends on how quickly new copper output and data-center construction expand.
The trend: AI data-center expansion is broadening the infrastructure bottleneck from compute hardware and electricity to the raw materials required to build facilities.