Copper: supply, demand and price over the long horizon

Copper is trading around US$14,250 a tonne in August 2026 — and, for the first time on record, above the 2011 super-cycle peak even after adjusting for inflation. A nominal record says nothing on its own; a real one says a great deal. It means the market is now paying more purchasing power for copper than at any point in its history, which forces the question that a commodity with a fifteen-year investment cycle has to be asked in decades rather than quarters: is this a cycle, or a change of regime?

This page does not answer that with a number. It answers it with a model: six assumptions you set yourself, and one pricing identity that follows from them. It is not a spot-price forecast — over a twenty-year horizon that would be absurd. It is a way of showing what price is required for supply and demand to meet under a given set of assumptions, and how sensitive that price is to what you believe.

The base case is built from published market data and consensus midpoints; it is not an Atlas Minerals view on what will happen. Every constant is written out in the assumptions table at the foot of the page.

Supply — why new copper arrives slowly

Copper is not short in the ground. It is short of economically available copper in places where it can be permitted, built and connected to a grid within a sane timeframe. Four structural forces hold supply growth below what the base-case demand path would need.

• Falling head grades. The global average head grade has fallen from roughly 0.9% Cu at the turn of the century to about 0.53% today. To hold output flat in metal terms, mines must move almost twice the ore — with the matching increase in energy, water and waste. The model carries this escalation as 0.9% a year of real cost growth.
• Discovery-to-production lead times. The median for a large copper deposit runs 15 to 20 years. The price signal of 2026 therefore cannot become physical metal before the second half of the 2030s — and the projects producing in 2030 were sanctioned when copper was near US$8,000.
• Capital intensity. A greenfield project costs roughly US$25,000–30,000 per tonne of annual capacity. A single new 300 kt/yr line is an eight-billion-dollar investment decision, usually in a jurisdiction with a long permitting path and rising local-content requirements.
• Geographic concentration. Chile and Peru together supply about a third of world mine production, and the Democratic Republic of the Congo is the fastest-growing source. Supply risk is therefore not diversified — one water crisis in the Atacama, or one change of regime in the Central African copperbelt, reprices the whole balance.
• Disruption is the rule, not the exception. The industry routinely books a 4–5% annual allowance for strikes, slope failures, water shortages and litigation. Actual output has almost never matched the plan set at the start of the year.

The mine supply growth slider is the net rate after depletion and disruption. For reference: world mine production grew about 2.3% a year between 2000 and 2024, but the past decade has been materially slower. The base case uses 1.0% — roughly what the sanctioned and under-construction project pipeline delivers without new incentives.

Secondary copper is modelled separately. Scrap arises from the installed stock of metal rather than from mine output, so it is carried as a share of refined consumption — otherwise fast mine growth would manufacture secondary supply out of nothing.

Demand — the grid decides, data centres accelerate

The copper demand story is usually told through electric vehicles. That is the smaller half of it. An EV does carry roughly three times the copper of a combustion car, but even in 2045 transport is only about one sixth of total demand under the base case. The decisive volume sits elsewhere: in the transmission and distribution networks that make electrification possible at all.

Every megawatt of renewable generation needs lines, transformation and a connection. Every fast charger needs a reinforced feed. Every data centre needs distribution, cooling and backup power. Copper is an infrastructure commodity far more than an automotive one — which is why the grid and renewables investment slider is the single most sensitive input in the model.

Data centres are carried as a separate segment because they behave unlike anything else: they start from a very small base — under 1 Mt in 2026 — and grow at a rate no other segment reaches. The model deliberately decays that rate each year: fifteen per cent sustained for two decades would make data centres the largest single consumer of copper on the planet, which is not a scenario but an extrapolation error.

Against demand sits substitution and thrifting. Aluminium displaces copper in transformer windings, busbars and part of the distribution cable market — historically whenever the copper-to-aluminium price ratio passes roughly three. That is a real ceiling on price, not a theoretical one: it removes 5% of demand by 2045 in the base case, and 12% under demand disruption.

Price — how the model gets there

The price in this model is not a drawn line. It comes out of two steps, neither of which contains a free parameter tuned to make the answer come out nicely.

Step one: cost support. The last tonne needed has to be paid for. The guide is the incentive price of a project at the 90th percentile of the cost curve — the price at which a new greenfield mine makes sense. Today that sits near US$10,800/t in 2026 dollars. The model escalates it for two things: the real cost growth that falling grades and rising strip ratios impose (0.9% a year), and the steepening of the cost curve that expansion itself causes — the more that gets built, the dearer the next increment is.

Step two: market clearing. A physical deficit cannot exist — somebody simply does not get metal. Price is what closes it. The model therefore solves the standard long-run identity in which both supply and demand respond to price with their own elasticity:

Finally, the model recognises that spot currently trades above long-run equilibrium. The 50% US import tariff on copper, effective 1 August 2026, has opened the COMEX–LME spread; exchange stocks cover roughly six and a half days of world consumption; and the market is restocking. The model decays this cycle premium over about five years — which is why the modelled path falls at first, before structural tightness pushes it back up. That shape is not an error; it is the difference between a cycle and a regime.

The band around the modelled path is not a statistical confidence interval. It is the same identity computed under a conservative and an aggressive reading of the two least certain constants — cost escalation and elasticity. It expresses the range of defensible opinion, not probability.

What this means for Morocco

If the model's conclusion is even roughly right — that copper's cost support is stepping up and staying there — it changes the economics of exploration in jurisdictions that have so far sat at the margin. The Moroccan copper belt is one of them.

• Proximity to market. Morocco sits within direct reach of the European market, with port and power infrastructure most new copper jurisdictions lack. Where a premium for security of supply is priced in, logistics is part of the deposit.
• High Atlas activity. Morocco Strategic Minerals Corp (TSXV: MCC) published assay results from Ouneine on 29 July 2026 — individual samples up to 14.88% Cu and 696 g/t Ag — and holds an option over a portfolio of five mining licences together with the environmental permit for a planned central flotation plant in the Ouneine valley.
• Other operators. Atlas Mining is advancing its TMN copper project across the Taroudant and Siroua areas of the Anti-Atlas; Aterian is tracking its own Moroccan copper targets. The sector is junior and pre-production, but it is no longer empty.
• Regulatory direction. The mining-law amendment (draft bill 72.24) simplifies procedures and establishes a national commission for strategic minerals; the digital mining cadastre has been live since April 2026. Both shorten the most expensive part of a project's life — permitting.

The assay results quoted are individual samples published by third parties. They do not state or imply resources, reserves or average grades. Atlas Minerals claims no ownership of any permit or block referenced here.

Assumptions and method

The model has six inputs you control, plus the fixed constants below. There is nothing else in it.

The historical price series is the annual average LME cash, grade A; 2026 is the January-to-August average. The real series is deflated to 2026 dollars on CPI-U. Each chart carries a Show data toggle with the underlying table.