Copper

Copper has been rerated from a construction-cycle metal into the bottleneck commodity of electrification. The world consumes roughly 27 million tonnes of refined copper a year, with China taking well over half, and the new demand layers stack on top of the old ones: a battery-electric vehicle contains roughly 50 to 80 kilograms of copper against roughly 20 in a combustion car, grid operators in the US, Europe, and China are in the largest transmission build-out in decades, and hyperscale data centers draw copper by the tonne for busbars, cabling, and cooling. That demand story carried LME copper to a record $11,104.50 a tonne in May 2024, and after the September 2025 Grasberg mine disaster in Indonesia knocked out one of the world's largest sources, the rally went vertical: a new record above $11,200 in late October 2025, through $13,000 in early January 2026, an intraday peak of $14,527.50 a tonne on January 29, 2026, and then a new record of $14,875 on September 10, 2026.

Supply is struggling to answer. First Quantum's Cobre Panama mine, roughly 350,000 tonnes a year and about 1 percent of world supply, was ordered shut in November 2023 after Panama's Supreme Court ruled its contract unconstitutional, and it has sat idle since. Chilean grades keep falling; Codelco, still the largest producer, has been mining its lowest output in roughly a quarter century. The growth is in the Democratic Republic of the Congo, where Ivanhoe's Kamoa-Kakula complex ramped past 400,000 tonnes a year and pushed the DRC past Peru into second place among mining nations. Meanwhile China built so much smelting capacity that spot treatment charges, the fee smelters earn for converting concentrate, went negative in 2025: smelters effectively paying miners for feed.

The falling-grade debate in copper is worth considering. US mines recovered about 2.5 percent copper from their ore in 1906 and about 0.4 percent a century later (see the ore-grade chart further down this page), and Cochilco puts Chile, the largest producer, near 0.6 percent today against roughly 1 percent at the start of the 2000s. Lower grade means more rock blasted, hauled, ground and floated for each tonne of metal, so more energy, water and capital, and costs rise with it. But grade is partly a choice. When the price rises, rock that was waste becomes ore, so the average grade falls because the mine is working more of its deposit. The same chart shows the reverse in 1933: at the bottom of the Depression only the richest US mines kept running, and the average jumped above 2 percent before falling back as prices recovered. MinEx Consulting found that the twelve giant porphyry deposits that held over half the world's known copper in 1929 now account for roughly six times the metal anyone credited them with then, mostly because the cut-off grade fell, so a falling grade also measures how much larger the usable resource has become. The real squeeze is time: a deposit found today can take close to two decades to produce, the IEA expects mined supply from known projects to fall about 25 percent short of demand by 2035, and a 2026 study co-led by Adam Simon of the University of Michigan put the price needed to fund enough new mines above $20,000 a tonne.

Copper trades on three venues at once: the LME 3-month forward is the global reference, COMEX is the US leg, and SHFE sets the tone in Asian hours. The May 2024 COMEX squeeze and the 2025 tariff arbitrage, when COMEX traded as much as roughly 28 percent over the LME before the July 30, 2025 cathode exemption crashed the premium in a day, showed that the three legs can disconnect violently when policy or positioning overwhelms the physical arbitrage.

The US tariff price rally began in 2025. The July 2025 proclamation put a 50 percent tariff on semi-finished copper products from August 1 and exempted refined cathode, but it said a tariff on refined copper, 15 percent from January 2027 and 30 percent from 2028, would be considered after a Commerce Department update due by June 30, 2026. That update was delivered, and as of early October 2026 no decision has followed. A tariff that might arrive is a reason to ship metal into the US before it does, so COMEX warehouse stocks rose from under 100,000 short tons in early 2025 to nearly 770,000 by late September 2026, while LME stocks fell from about 400,000 tonnes in April to about 250,000. The world as a whole is in surplus this year, but all of the surplus is in the US: Wood Mackenzie estimates the rest of the world is short about 460,000 tonnes. LME copper set its record in September and then fell about 6 percent when Reuters reported White House hesitation over the refined tariff. So the price carries two things at once, a real shortage outside the US and a stockpile inside it that exists because of a policy decision still pending, and a clear answer either way would release the second. The stockpile has also put the COMEX curve into a contango of 5 to 6 percent a year, a cost borne by anyone who holds copper through rolling futures.

The bull market that runs through all of this began in 2003 and 2004. Copper had spent the turn of the century cheap, near 60 cents a pound in 2001 and 2002, and miners had cut capacity and exploration to match. Then China, newly inside the WTO and building cities faster than any economy before it, began buying copper faster than the world could mine it. The ICSG counted refined deficits of several hundred thousand tonnes in both 2003 and 2004. The LME price averaged about $1,800 a tonne in 2003, $2,900 in 2004 and $3,700 in 2005, and by mid-2005 visible stocks on the LME, Shanghai and COMEX together came to about 72,000 tonnes, under two days of world use, with the LME at its lowest since 1974. A new mine takes the better part of a decade, so supply could not answer and the price had to ration demand instead. A Chinese state trader then made it worse: Liu Qibing of the State Reserve Bureau built a short position of about 130,000 tonnes on the LME in 2005 and disappeared in October, leaving the bureau to buy back into a rising market. Commodity index funds, which grew quickly from 2004, added a buyer that never used the metal. Copper reached about $8,800 a tonne in May 2006, roughly five times its 2003 average.

Substitution answers part of a high price, slowly. Aluminium carries about 60 percent of copper's conductivity for the same cross-section, but a cable of equal capacity weighs roughly half as much and costs far less, which is why nearly every high-voltage overhead line in the world is already aluminium. The live fronts are air conditioners, where aluminium tube is now in an estimated 40 percent of units, power and building cable, car wiring harnesses, and radiators; plastic has already taken most new water pipe and optical fibre most telecom cable. The trade watches the copper-to-aluminium price ratio, and switching is generally thought to accelerate above about 3.5 to 4. The ratio reached a record of about 4.3 in January 2026. J.P. Morgan estimates substitution now removes about 2 percent of world copper demand and could reach about 6 percent by 2030, and Wood Mackenzie reckons the last supercycle cost copper 400,000 to 500,000 tonnes of demand a year. What does not switch easily is the core of the electrical economy: the windings of efficient motors and generators, printed circuits, and the dense wiring inside data centres and electric vehicles, where space and heat matter more than the cost of the metal. Each switch also means redesigned products, new connectors and, for building wire, changed safety codes, so the realistic range is a few percent of demand, reached over years.

There is no single price at which demand breaks, because what matters is whether fabricators can pass the cost on to their customers. The signs are consistent, though, and they show first in China, which uses more than half the world's refined copper. In May 2024, with COMEX at a record $5.20 a pound, Chinese wire and rod makers delayed orders and cut run rates to about 66 percent of capacity, the lowest for the season since at least 2017. Shanghai Futures Exchange stocks rose to a four-year high of about 291,000 tonnes, the premium for imported cathode in Shanghai fell to a record discount, and China exported a record amount of refined copper. The LME price fell about a fifth by August. Bloomberg reported the same worry about Chinese buyers near the January 2026 peak. The warning lights to watch are the Yangshan import premium, Chinese fabricator operating rates, SHFE stocks and the copper-to-aluminium ratio; when they turn together, the price has reached the level buyers will not pay.

Why do central banks hold gold rather than copper? Value density is most of it. At October 2026 prices a tonne of gold is worth about $133 million and a tonne of copper about $15,000, so $1 billion of gold is about 7.5 tonnes, a cube under a metre on each side, while $1 billion of copper is about 67,000 tonnes, enough to fill roughly three Olympic swimming pools. Copper is also used up: nearly all the gold ever mined still exists, while copper goes into walls and machines and is only partly recovered. And a reserve asset has to hold its value in a crisis, which is when copper falls, because its price tracks industry; it lost about two-thirds of its value in the second half of 2008, while gold ended the year higher. People do hold copper. China's State Reserve Bureau keeps a stockpile as industrial insurance, and the Sprott Physical Copper Trust has held metal for investors since 2024, at a fee of about 1.4 percent a year that reflects the cost of warehousing something this bulky. Copper was money once. Sweden, rich in copper, struck it at full metal value from 1644 to 1776, which meant plates rather than coins; the largest, the 10-daler plate, weighed 19.7 kilograms. People left the plates in storage and traded the receipts, and Stockholms Banco issued Europe's first banknotes in 1661.

A US penny minted before 1982 is 95 percent copper, about 3 grams of it, and at 2026 prices the metal in it is worth more than 4 cents. Melting them has been a federal crime since 2006, a rule written for exactly that arithmetic. The Mint had already switched to copper-plated zinc in 1982 because copper had made the old penny too expensive to produce, and in November 2025 it struck its last penny.