The problem
According to widely cited projections, humanity needs to mine as much copper in the next two decades as it extracted across all of recorded history. Recycling covers roughly 17% of refined copper supply today, and that fraction cannot scale fast enough to close the gap.
The usual response is to point at mining pipelines and assume supply will catch up. It probably will not, at least not before the demand surge arrives. The reasons matter enormously for investors trying to identify the right exposure.
Four demand forces, arriving at once
Copper was a cyclical metal for most of the 20th century, rising and falling with construction and manufacturing. That framing no longer fits cleanly. Copper's electrical and thermal conductivity make it hard to replace in motors, cables, circuits, cooling systems, and data transfer. No affordable substitute exists at scale. Demand is not arriving sequentially. It is arriving from four directions at once.
Electric vehicles
EVs are the most covered story. They require several times the copper of combustion vehicles, roughly 85kg versus 22kg, concentrated in motors, inverters, and charging infrastructure. More than half of cars sold in China are now electric. Scaled across tens of millions of vehicles annually, the transport shift is a meaningful demand shock.
But EVs are not the dominant new force. The grid is.
Grid reinforcement
Every EV, heat pump, data centre, and industrial motor connects to transmission and distribution infrastructure that was not designed for them. Grid reinforcement, including transmission lines, substation upgrades, transformer replacements, cable, and switchgear, is one of the largest copper demand stories of the next decade. It barely appears in headlines focused on battery chemistry or chip architecture.
The copper sits in the unglamorous layer. The scale of infrastructure required to move power from source to end user dwarfs the copper inside any individual device.
Data centres
Data centres add a different kind of pressure. A large hyperscale facility consumes 500 megawatts or more. Data centres currently account for roughly 710,000 metric tons of copper demand annually. By 2031, that figure is forecast to exceed one million metric tons.
Over the same period, total global refined copper supply is expected to grow from roughly 26.5 million tons to around 29 million tons. If the grid infrastructure connected to data centres is included, the sector could approach 6% to 7% of global supply within five years.
This demand is unusually price insensitive. Copper is a small share of a data centre's construction budget, roughly 3%. If Microsoft, Google, Meta, or Amazon need copper to avoid falling behind in the AI buildout, they pay. Tech infrastructure can outbid traditional housing, plumbing, and appliance markets for tight supply.
Rearmament
Rearmament is the least modelled driver. Precision munitions, drones, radar systems, ships, directed-energy platforms, and electronic warfare systems all require copper alongside rare earths, gallium, antimony, and other critical inputs. Modern military platforms are mobile data centres, with miles of shielded copper wiring inside ships, drones, vehicles, and guided systems.
Defence procurement is estimated to account for 3% to 6.5% of annual global copper supply. Unlike civilian copper, which is often eventually scrapped and recycled, military copper is destroyed, vaporised, or lost. Every ton consumed must be replaced by newly mined ore. Defence buyers are also more price insensitive than ordinary civilian demand and, in some jurisdictions, can receive priority treatment under national security law.
A supply system built for another world
The copper supply system was not designed for four simultaneous demand surges. Its constraints are geological, chemical, regulatory, and political, and they compound each other.
Geology
Average ore grades have declined for decades. The global average now sits around 0.6% copper content, meaning a mine must process roughly 200 tonnes of rock to yield one tonne of copper. Lower grades require more energy, more water, more processing, and more capital per pound of output.
The economics become brittle under input-cost pressure. At elevated oil prices, diesel costs at a 0.6% grade mine can outpace copper price gains and force marginal operations to shut. Record copper prices do not automatically save marginal mines if operating inputs inflate faster than the price of the metal.
Chemistry
Roughly 20% of global copper production relies on leaching low-grade oxide ore with sulfuric acid. That same acid is also used in fertiliser production and semiconductor manufacturing. Middle East shipping disruptions have cut an estimated 50% of seaborne sulphur flows, while China's 2026 restrictions on by-product sulfuric acid exports compound the pressure.
Realised acid prices have exceeded $500 per tonne in some markets. When acid becomes scarce, leaching projects become financially fragile. The constraint applies specifically to SX-EW oxide operations, not to sulfide-concentrate smelting, which means investors can identify which mines face existential input risk and which do not.
Regulation
Permitting a greenfield copper mine in a high-rule-of-law jurisdiction often takes a decade or more, before legal challenges. Some large US deposits have been in permitting and litigation for longer than the modern AI era has existed. The constraint is not always geological uncertainty. It is the institutional process of converting a discovered ore body into a producing mine.
Geopolitics
A substantial share of copper refining capacity sits in China. The risk is not that copper ore disappears. It is that processing know-how, capital equipment, and logistics infrastructure become subject to trade restriction.
The less obvious choke points
A data centre is not a copper machine. It is a basket of materials: copper, gold, silver, palladium, rare earth elements, gallium, indium, tantalum, tungsten, barium, titanium, and more. Several, including gallium, antimony, and graphite, already face Chinese export controls.
Large power transformers show how deep the bottlenecks go. Lead times at major manufacturers have reached 150 weeks. The bottleneck is not just copper. It is grain-oriented electrical steel, hand-wound manufacturing processes that resist automation, and a skilled labour shortage that capital alone cannot fix. The US needs more than 300,000 additional electricians over the next decade. Germany, the UK, and Australia face similar gaps. You can fund a substation expansion. You cannot train an electrician in six months.
The silver feedback loop is the kind of dependency standard commodity analysis misses. Roughly 70% of global silver is produced as a by-product of copper mining. A copper supply constraint also constrains silver supply, raising the cost of photovoltaic cells via silver paste. That slows grid decarbonisation, prolongs reliance on fossil-fuel generation, and maintains pressure on the energy system copper mining itself depends on. Each link is mundane. The combined effect is self-reinforcing.
Concentrate quality is another signal. As ore grades fall, mines produce more complex concentrates with higher arsenic and other penalty elements. Smelters charge penalty fees, and many older facilities cannot handle those concentrates at all. Treatment and refining charges sitting at historic lows indicate that smelters are competing aggressively for feedstock. The bottleneck has shifted from refining capacity to mine-level concentrate supply.
Water closes the loop geographically. The Atacama and similar arid zones host some of the largest undeveloped deposits. Desalination can help, but it requires energy, which is itself under pressure from the same demand drivers that need the copper.
What investors should actually ask
The 18-year copper problem is not primarily a trading question. It is a structural question about which physical constraints become binding, in which sequence, and which actors have the assets, jurisdictions, and capabilities to relieve them.
The useful questions are: which undeveloped projects can realistically produce before 2035; which jurisdictions can permit mines on a relevant timeline; how much incremental copper demand comes specifically from AI versus EVs, grid upgrades, and defence; and which public companies give clean exposure to the bottleneck without simply overpaying for commodity beta?
The constraints are not uniformly distributed. They punish some operators severely and leave others largely insulated. One operator sits at an unusual intersection: some of the highest-grade deposits on earth, self-sufficient power, and self-sufficient acid supply. That combination sidesteps most of what makes marginal copper expensive. That is the Ivanhoe Mines thesis.
Kamoa-Kakula: grade as a structural moat
The Kamoa-Kakula deposits in the DRC grade between 3.5% and 5.5% copper, roughly six to nine times the global average. At a 5% grade, a mill processes about 20 tonnes of rock for one tonne of copper. At 0.5%, it processes 200. Crushing and grinding is the most energy-intensive part of most mining operations. The difference is not incremental.
The grade advantage compounds through the value chain. High-grade ore produces richer concentrate. At Kamoa-Kakula, that concentrate does not leave the site as feedstock for a third-party smelter. The on-site direct-to-blister smelter processes it directly, removing freight, penalty fees, smelter queues, and counterparty risk.
To produce equivalent annual output, a low-grade mine needs far larger processing infrastructure, trucking fleets, tailings facilities, and energy input. The capital deployed at Kamoa-Kakula buys far more productive capacity per dollar than the same capital deployed near the global average grade.
The input-cost advantage
Ivanhoe has also addressed the two input problems that threaten marginal producers.
Energy: instead of relying on diesel generation, Ivanhoe partnered with the DRC's SNEL to rehabilitate the Inga II hydroelectric plant, securing renewable power at costs insulated from oil market volatility.
Sulfuric acid: the direct-to-blister smelter now ramping at Kamoa-Kakula captures sulfur gas and produces roughly 1,350 tonnes of high-purity sulfuric acid per day as a by-product. Ivanhoe can use the acid in oxide leaching and sell surplus into the DRC copperbelt. A line item that destroys marginal producers can become a revenue stream.
The financial results reflect the structural position. In Q1 2026, the Kamoa-Kakula Complex contributed approximately $158 million to adjusted group EBITDA, anchoring total group EBITDA of $191 million. The 2026 production target is 290,000 to 330,000 tonnes of copper in anode or blister, following geotechnical design adjustments in the March 2026 technical report. Underground dewatering continues, with the western zone at roughly 70% completion and the eastern zone at 60%.
Portfolio optionality beyond Kamoa-Kakula
Ivanhoe's other assets add optionality without diluting the core thesis.
- Platreef, South Africa: Shaft #3 construction completed in late March 2026, expanding hoisting capacity five-fold to 5.0 million tonnes per annum. Phase 2 earthworks are underway, backed by a $700 million project finance facility closed in April 2026, targeting over 450,000 ounces of platinum group metals plus nickel and copper by late 2027.
- Kipushi, DRC: the zinc mine produced a record 65,044 tonnes of zinc in concentrate in Q1 2026 at a cash cost of $0.86 per pound, generating $162 million in revenue at a 36% EBITDA margin.
- Kazakhstan exploration: drilling doubles to over 40,000 metres in 2026 following encouraging early assays.
There are still risks. The Lobito rail corridor, Ivanhoe's primary logistics route for copper shipments, is temporarily paused until June for flood damage repair in Angola. Even well-structured operations carry logistics and jurisdictional exposure.
Bottom line
Copper's structural deficit is real, but not all copper exposure is created equal. The constraints on marginal supply, from declining grades to acid scarcity, input-cost inflation, water, skilled labour, and decade-long permitting, create a widening performance gap between operators with structural cost advantages and those without.
Kamoa-Kakula's grade, hydro power, onsite smelting, and acid self-sufficiency place Ivanhoe at one end of that gap. The thesis is not that copper prices only go up. It is that the highest-grade, lowest-input-cost operations can produce economics that marginal mines cannot approach across a wide range of price environments.
The supply problem does not get solved before the demand arrives. The question is who captures the margin when it does not.
This note is for informational purposes only and does not constitute investment advice. Figures cited reflect public company disclosures, market estimates, and third-party projections; do your own research before making investment decisions.