The AI boom beneath our feet: How data centres are rewriting mineral demand

Artificial intelligence (AI) may run in the cloud, but its foundations lie deep underground. Behind every autonomous vehicle algorithm, every training run, every AI-generated video, sits a physical network of hyperscale data centres, which are rapidly becoming one of the strongest new demand engines for critical minerals.

For Canada’s mining sector, the rise of AI represents a rare convergence: a rapidly expanding global market, strong alignment with Canada’s mineral strengths, and growing recognition that secure supply chains matter just as much as performance and price.

AI systems require enormous computational horsepower. As data centres scale to meet that demand, they are consuming not only vast amounts of electricity but also unprecedented volumes of minerals. According to the International Energy Agency (IEA), data centres already account for roughly 2% of global electricity use, a figure expected to more than double by 2030. In the U.S. alone, electricity demand could climb from under 200 TWh in 2023 to between 400 and 600 TWh by 2030, driven largely by AI workloads.

But beyond power, these facilities rely on a substantial — and diverse — suite of minerals to operate.

A recent U.S. Geological Survey breakdown highlights just how broad the material footprint is

• AI chips: silicon, germanium, gallium, indium, arsenic

• Server boards and circuitry: copper, silver, gold, tin, palladium, platinum, tantalum

• Cooling and heat sinks: copper and aluminum

• Storage and magnets: barite and rare earth elements (REEs)

Many of these elements are sourced from concentrated global supply chains. China dominates the production and processing of REEs and several byproduct metals essential to semiconductors. Some — like gallium, indium, and tantalum — are 100% import-dependent in the U.S. With AI accelerating, Western economies are reassessing where their minerals come from and how secure those supplies are.

No metal stands to benefit more directly from AI-linked infrastructure than copper. AI data centres are essentially copper frameworks. The metal carries high-voltage power from the grid into server halls; it enables heat exchange; it is woven into transformers, switchgear, interconnects and vast webs of cabling. According to BloombergNEF, the industry average shows about 27 tonnes of copper used per megawatt of AI data center capacity. The metal can represent up to 6% of total data-centre capex.

Data centre demand for copper will average 400,000 t/y over the next decade, peaking at 572,000 tonnes in 2028. By 2035, global data centres could cumulatively lock up more than 4.3 million tonnes — equivalent to the annual production of the world’s top five mines combined.

The challenge, however, is that global supply is tightening. Years of under-investment and lengthy permitting timelines suggest copper output will reach only 29 million tonnes by 2035, far short of the roughly 35 million tonnes needed under current demand forecasts.

Battery chemistries are also evolving quickly. While lithium iron phosphate (LFP) dominates data-centre UPS installations today, alternatives like sodium-ion, zinc-based, and redox-flow batteries are advancing rapidly. This diversification creates a wider playing field for Canadian mineral suppliers in this new mineral economy as follows:

Abundant deposits across copper, nickel, cobalt, lithium, REEs, and aluminum.

A global reputation for responsible mining practices and transparent governance.

“Preferential access” to the largest and fastest-growing AI data-centre market in the world: the U.S.

A federal “Critical Minerals Strategy” designed to build supply chains for batteries, semiconductors, and clean energy technologies.

The surge in AI-driven mineral demand offers a powerful tailwind for Canadian miners, but it also brings new layers of complexity.

Commodity markets linked to AI — especially copper — are likely to experience sharper cycles, as forecasts tighten and new supply struggles to keep pace. These dynamics underscore the need for more agile planning frameworks; ones that blend rigorous long-range forecasting with the ability to adjust quickly when markets shift. Companies that can model price volatility, adjust capital allocation and maintain optionality in project sequencing will be better positioned to capture value across the cycle.

Supply-chain resilience will also become a core differentiator. As governments and technology companies scrutinize upstream risks, the reliability of sourcing — not just the volume — will matter. Canada’s reputation for high ESG standards and predictable regulation is a strength here, but miners will need to reinforce that advantage through transparent reporting, strong community partnerships, and consistent operational performance. In a climate where buyers increasingly care about how minerals are produced, not just where, Canada has the potential to lead.

Project execution risk is another area demanding attention. If global forecasts materialize, the world will require an unprecedented wave of new mines, expansions, and processing facilities. Delivering these on time and on budget will be critical, especially as cost inflation, labour shortages, and supply-chain challenges become more frequent. Advanced risk modelling, stronger contractor oversight, and proactive insurance strategies — particularly around DSU and construction risk — will play an essential role in protecting returns and supporting investor confidence.

Finally, the rapid evolution of battery technologies and semiconductor materials means miners must remain closely attuned to technological pathways. Today’s dominant chemistries may not define tomorrow’s demand profile. Maintaining a diversified commodity portfolio and monitoring shifts in AI hardware design will help ensure capital is aligned with enduring demand, not short-lived cycles.

In the AI era, risk management is not a defensive exercise. It is a competitive advantage — and one that Canadian miners are well-positioned to leverage.

AI is not just transforming digital workflows; it is reshaping the physical economy that supports them. The build-out of hyperscale data centres is poised to drive some of the strongest mineral demand growth of this decade, led overwhelmingly by copper but extending across a wide suite of critical minerals.

For Canada, the opportunity is clear. Our geology, our governance, and our proximity to the world’s largest AI market position us uniquely to supply the minerals that will power the next wave of global innovation. The companies that rise to the challenge will be those that combine development ambition with strategic, forward-looking risk management — the firms ready to fuel the world’s digital future, one tonne at a time. 

Raul Munoz is managing director and North American industry leader for mining & natural resources at Marsh. Based in Vancouver, he leads Marsh’s mining advisory, risk management, and insurance strategy across Canada and the U.S., working closely with global mining companies on complex operational, construction, and energy-transition risks.