In January, copper briefly traded above $14,500 a tonne. The metal had crossed $12,000 for the first time in its history only the month before. By the standards of a commodity that spent most of the past decade between $6,000 and $9,000, this was a violent repricing of something the world had long treated as boring.

In the same month, a negotiation concluded in a conference room between the Chilean miner Antofagasta and a group of Chinese smelters. They were settling the annual benchmark treatment charge, the fee a smelter earns for turning copper concentrate into refined metal. In 2025, the benchmark had been around $21 a tonne. For 2026 they settled on zero. It was the lowest figure ever agreed in annual negotiations.

Record price. Zero processing fee. Those two numbers look like they belong to different industries, and understanding why they belong to the same one is the whole story.

What a treatment charge is, and why zero is alarming

Copper does not come out of the ground as copper. It comes out as concentrate, a powder that is perhaps a quarter to a third metal by weight. Somebody has to smelt it, and smelters are paid for the service through what the industry calls treatment and refining charges. The miner sells the concentrate at a discount. That discount is the smelter’s income.

When smelters are plentiful and ore is scarce, they compete for concentrate and the charge falls. When ore is plentiful, the charge rises. It is one of the cleanest supply signals in industrial metals, and it has now gone to zero on the annual benchmark, while the spot version of the same charge has been negative since 2024. Negative means the smelter pays the miner for the privilege of doing the work.

A smelter working on the 2026 benchmark earns nothing for processing concentrate. Many stay profitable only by selling the gold, silver and sulphuric acid that comes out alongside the copper, all of which have been fetching record prices.

China built this squeeze itself. Since 2005 the country has accounted for over 90 percent of the growth in global copper smelter output, lifting its share of world capacity from around 15 percent to roughly half by 2025. That capacity was constructed on the assumption that the mines would keep up. They have not. There are now far more furnaces than there is rock to feed them, and the furnaces are bidding against each other for what exists.

The shortage is not in the metal. It is in the rock.

The number that does not respond to price

Here is the fact that sits underneath everything else, and it is not a market fact at all. According to the International Energy Agency, the average grade of the world’s copper mines has fallen roughly 40 percent since 1991.

Grade is the proportion of a rock that is actually copper. A 40 percent decline means that to produce the same tonne of metal, a miner must now dig, haul, crush and process substantially more material than a generation ago, using more energy, more water and more equipment to end up in the same place. High prices cannot create better geology. They can only make it economic to process worse geology, which is a slower and more expensive proposition than a price chart implies.

The consequences show up in the cost of building anything new. The IEA reports that the capital intensity of expanding an existing mine has risen 65 percent since 2020, and now approaches the levels once associated with building a mine from scratch. Wood Mackenzie estimates the industry needs more than $210 billion of capital investment by 2035 simply to meet expected demand.

S&P Global’s assessment is starker. Its January study, co-chaired by the energy historian Daniel Yergin, found that copper supply is on track to fall roughly 10 million tonnes short of demand by 2040, around 24 percent below a projected 42 million tonnes, even assuming recycling doubles. Yergin summarised the bind in a sentence worth keeping: copper is the great enabler of electrification, and the accelerating pace of electrification is an increasing challenge for copper.

Then everything broke at once

Against that structural backdrop, 2025 and 2026 delivered an unusually bad run of operational luck.

Sources: Cochilco, Freeport-McMoRan, Ivanhoe Mines, J.P. Morgan Global Research. Chile alone accounts for around a quarter of world supply.

Chile is the pivot. It produces roughly a quarter of the world’s copper, and its output has been sliding for reasons that are only partly accidental. Falling ore grades and ageing deposits are doing steady structural damage underneath the headline disruptions. When the largest producer on earth is delivering nine percent less than a year earlier, no amount of new capacity elsewhere closes the gap quickly.

And yet the market cannot agree on any of it

A magazine that stopped there would be selling you a shortage. The honest position is that the professionals are looking at the same data and reaching opposite conclusions.

Morgan Stanley sees the largest refined copper deficit in more than twenty years. Nornickel sees a surplus. They are forecasting the same twelve months.

Morgan Stanley projects a refined deficit of around 600,000 tonnes in 2026, which would be the largest in over two decades. ING lands in the same place. UBS estimates roughly 520,000 tonnes, with demand growing 2.8 percent against refined supply growth of 1.7 percent, and has projected $14,000 a tonne by September. J.P. Morgan sees 330,000 tonnes. The International Copper Study Group revised its own 2026 balance from a 209,000-tonne surplus in October to a 150,000-tonne deficit in May. Nornickel expects a small surplus. Goldman Sachs has argued for $10,000 to $11,000 a tonne and sees no genuine shortage before 2029.

Natalie Scott-Gray of StoneX has made the sceptic’s case most precisely. Western speculators are long copper, she notes, while traders in China hold their widest net short position on the Shanghai Futures Exchange since 2021. Two markets, opposite views, which in her reading usually means the Western specs have run too far. She also points out that a refined deficit of a few hundred thousand tonnes sits well below 2 percent of global demand, which is not obviously a crisis.

There is a sharper version of the same doubt. Mining companies have promoted the long-term shortage story so effectively that the market has priced in a scarcity that has not physically arrived. A great deal of the metal that moved into American warehouses ahead of tariff decisions is not being consumed at all. It is sitting in storage, financed against the forward curve. Inventory is not the same thing as demand, and belief is not the same thing as fundamentals.

Why the disagreement does not matter as much as it looks

The forecasters are arguing about a single year. Whether 2026 closes with a 600,000-tonne deficit or a small surplus depends on Chinese stimulus, tariff policy, the speed of the Grasberg restart and whether the global economy slows. Those are real questions and nobody knows the answers.

They are also, for the purposes of this argument, the wrong questions. A 40 percent decline in ore grade over thirty-five years is not a cycle. It does not reverse when China stimulates, and it is not undone by a mine reopening. Every year the industry digs more rock to produce the same metal, and every year the marginal deposit is deeper, poorer and further from a port. Meanwhile the demand side keeps adding claims: grids, electric vehicles, air conditioning, defence, and now the data centres.

The IEA’s projection of a 30 percent supply deficit by 2035 assumes the current project pipeline. Pipelines can be lengthened, and $210 billion can be found. But a new copper mine takes the better part of two decades to move from discovery to production, which means the metal that will be short in 2035 needed its capital in about 2017. That money was not spent. The window in which the shortage could have been prevented by investment has, for a good part of the coming decade, already closed.

Copper is priced today on a fight about next year’s inventories. It will be settled by a fact from 1991.