Ask any data‑center operator and they’ll tell you: the internet eats copper. Now imagine that appetite multiplied by generative AI, a global rearmament and, just for fun, a few hundred million humanoid robots.
A new study from S&P Global has thrown cold water — and a hot spotlight — on an uncomfortable possibility: demand for copper could rise roughly 50% by 2040 while mine production struggles to keep up, creating what the firm calls a potential systemic risk to global economies.
The math: kilotons, data halls and unlikely robots
S&P’s base case projects copper demand climbing to about 42 million tonnes by 2040, up from roughly 28 million tonnes today — a near‑term doubling in some uses and a steep cumulative rise overall. Traditional drivers such as construction and electrical infrastructure remain the backbone of that growth, but the marginal drivers are newer and faster: electric vehicles and batteries, expanding power networks, and the rapid build‑out of AI infrastructure.
The report flags data centers and AI as especially fast‑growing consumers of copper. S&P estimates global installed data‑center capacity could almost quadruple by 2040; the metal used in racks, cabling, cooling and power distribution scales with that footprint. You can see why projects like Google’s Project Suncatcher — aimed at radically rethinking where and how we host compute — matter beyond tech headlines: they’re responses to the same appetite S&P is quantifying.
S&P also calls out defense spending and even speculative scenarios. In a vivid thought experiment the researchers note that if 1 billion humanoid robots reach the market by 2040 (a far‑out case), that could demand about 1.6 million tonnes of copper annually — roughly 6% of today’s consumption. Combine AI, data centers and defense and you add several million tonnes of extra demand by mid‑century, S&P says.
Where new demand collides with shrinking supply
On the supply side, S&P’s outlook is less cheerful. Global copper production, the report says, may peak near 33 million tonnes around 2030 and then drift lower as ore grades fall and existing mines become less productive. Building new large‑scale mines is capital‑intensive, slow and politically fraught: permitting delays, financing hurdles, local opposition and environmental reviews all stretch timelines. That mismatch — accelerating demand and sluggish supply growth — is the heart of the squeeze.
Prices are already sending signals. Copper rallied to record levels in recent months, in part because of mine outages and in part because of stockpiling in anticipation of trade frictions and policy shifts. Bloomberg and market data show benchmark copper in London trading above previous peaks as buyers scramble for physical metal; such price moves can feed through to everything from construction costs to the economics of electrification.
Not just an industry problem: systemwide risk
S&P’s framing is notable: they warn the shortfall could become a systemic risk — a phrase commonly reserved for financial contagion and infrastructure breakdowns. If access to copper tightens, the knock‑on effects could touch manufacturing, grid upgrades and the very deployments meant to cut emissions (electric vehicles and renewable grids rely on significant copper inputs). In other words, shortages would not be a niche commodity story but a potential drag on broader economic and technological transitions.
Policymakers have started to notice. Some governments and large buyers are stockpiling or prioritizing domestic sourcing, which can amplify market distortions. Trade policy risks — including tariff talk in the U.S. — add another layer of uncertainty for supply chains that operate on global just‑in‑time logic.
Mines, permits and money — the practical bottlenecks
The technical hurdles to closing the gap are prosaic and stubborn. Discovering ore is one thing; developing a large, modern mine is another. Projects face long lead times (often a decade or more), capital scarcity for risky long‑dated investments, and increasingly strict environmental and social standards that can — rightly — slow or block developments.
At the same time, recycling and efficiency gains will help, but they’re unlikely to replace the vast volumes required for a truly electrified future. Companies will need to innovate on both supply and substitution fronts: better exploration tech, more efficient electrification materials, and circular‑economy approaches to reclaim copper from old infrastructure.
How companies and countries are reacting
Mining firms are pitching new projects and seeking partners and capital, while utilities and tech giants re‑engineer procurement and design to manage exposure. On the tech side, product and platform choices matter: decisions about where to place compute, how to cool it and what hardware to standardize can all affect copper demand. That’s why conversations about AI deployment tie into infrastructure planning in unexpected ways — from office racks to orbital data hubs — and why product and platform shifts (like agentic booking features in cloud services) ripple into infrastructure needs and purchase plans. See how developments in AI deployment are reshaping infrastructure planning with news about Google’s AI Mode and enterprise search and indexing innovations like Gemini Deep Research that expand practical compute demand.
For investors, manufacturers and policymakers, the simple reality is this: copper isn’t optional. It’s a core input in the green transition and in the data economy. If S&P’s scenarios are even directionally right, markets and planners will need to treat copper the way they treat oil or semiconductors — as a strategic commodity whose availability shapes the pace of technological and economic change.
Markets will adapt, as they always do. But adaptation takes time. And in a race where the finish line is an electrified grid, fleets of EVs and sprawling AI compute farms, time is exactly what we don’t have a lot of.