Kitco News: Copper’s AI boom exposes a hidden strain on the power grid
kitco news frames a market story that starts far from the trading floor: in data centers, substations, and the long runs of wire that turn electricity into computing power. Artificial intelligence may be built on code, but it runs on copper, and that reality is becoming harder to ignore as utilities and hyperscale operators rush to expand capacity.
Why is copper suddenly central to the AI buildout?
The answer is physical, not abstract. Data centers require enormous amounts of electricity, and electricity requires copper. As hyperscalers race to secure power capacity for AI workloads, utilities are accelerating grid upgrades, transformer deployments, and transmission expansions. That surge in infrastructure demand is quietly reshaping the copper market.
For years, copper’s investment case was tied mainly to electric vehicles and renewable energy. Now AI infrastructure is adding demand that did not meaningfully exist five years ago. The result is a tighter supply picture, with long permitting timelines and scarce large-scale discoveries slowing the response from miners. In commodity cycles, the lag between rising demand and supply growth can matter as much as the demand itself.
How are hyperscalers changing the copper story?
A January agreement between Amazon Web Services and Rio Tinto’s Nuton venture in Arizona signaled that copper is becoming part of AI procurement strategy, not just a background material. Amazon Chief Sustainability Officer Kara Hurst said, “Securing access to lower-carbon materials produced close to home strengthens our operational resilience. ” Amazon later said it plans to spend $200 billion in capital expenditures in 2026, with most of that directed to AWS data center expansion. CEO Andy Jassy said the company is “monetizing capacity as fast as we can install it. ”
Microsoft is also committing heavily to infrastructure, with $34. 9 billion in capital expenditures in the first quarter of fiscal 2026. Roughly half went to GPUs and CPUs, while the rest funded long-duration infrastructure, including data centers, power systems, and finance leases for land and electricity. CEO Satya Nadella described the focus as “maximizing tokens per dollar per watt, ” a phrase that captures how AI economics now depend on energy and materials as much as software. Microsoft expects to double its data center footprint within two years.
Oracle has taken a similar path through its partnership with OpenAI. The company plans to raise up to $50 billion in 2026 through debt and equity to fund AI infrastructure as part of the $300 billion Stargate project. Chairman Larry Ellison described one U. S. facility as “sized to fit eight Boeing 747s nose-to-tail. ” Oracle’s cloud infrastructure revenue grew 68% year-over-year in its most recent quarter, and its remaining performance obligations rose to $523 billion, a pipeline that will still need copper before it becomes operational capacity.
What does this mean for miners, utilities, and households?
The numbers are large enough to affect the broader system. A conventional data center uses between 5, 000 and 15, 000 tonnes of copper, while a hyperscale AI facility can require up to 50, 000 tonnes per campus. UBS forecasts copper deficits exceeding 400, 000 tonnes by 2026 as mine disruptions intersect with accelerating demand. S&P Global projects that AI, defense, and electrification together could drive even more pressure on supply over time.
That shortage is not just a market statistic. It can influence how quickly grids are upgraded, how fast data centers come online, and how expensive it becomes to build the infrastructure that powers daily digital life. Utilities are already moving to expand transmission and transformer capacity, but those systems take time to build and longer still to scale.
What responses are being tested now?
The response is emerging in both procurement and capital spending. Hyperscalers are trying to lock in power, land, and materials earlier in the build cycle. Miners and suppliers, meanwhile, face the challenge of meeting demand in a market where new capacity takes years to arrive. That mismatch is why copper has moved from a supporting role in the energy transition to a central piece of the AI trade.
For investors, the significance is less about a single company than about the structure of the cycle. AI may be the headline, but the infrastructure behind it is where the constraint lives. In that sense, kitco news is telling the story of a commodity being pulled into the center of a technological race, one transformer, one transmission line, and one data center at a time.
