Copper futures in London are approaching the record again. Three-month metal has printed a tenth consecutive weekly advance. August spent the month above $14,000 a tonne and tagged about $14,450 late in the month. The all-time high remains $14,527.50. The chart that has desks talking is not a narrative. It is a warehouse map.
United States seaborne imports reached 200,000 tonnes in July. Refined metal keeps flowing into U.S. sheds ahead of a possible import tariff. Every tonne that docks in America is a tonne that is not in Shanghai or on the LME. Shanghai Futures Exchange stocks were already reported down 13% in a week to 63,000 tonnes, the lowest since January 2024. That is how a policy rumor becomes a London price.
Canadian mining readers do not need a lecture on why $14,000 copper matters. They need the part the binary traders keep getting wrong.
The Tariff Is Not a Coin Flip
Rafael Barcellos, who runs Latin American metals research at Bradesco BBI, put the policy error in a note last week. Market consensus treats Section 232-style copper policy as binary: a tariff is bullish, no tariff is bearish. Both endpoints, he argues, are ultimately bearish for the tightness that is lifting London. The bullish object is the uncertainty that sits between them.
If tariffs are not implemented, the massive inventories already accumulated in the United States get redirected into the rest of the world. Global availability rises. The squeeze eases. If the United States instead slaps on a duty, U.S. buyers stop pulling new seaborne tonnes because they have already built the stockpile. Regional demand falls. Tightness elsewhere eases too.
The path that keeps the squeeze alive is the one Barcellos calls most likely: a phased tariff — an incremental rate over years — that leaves U.S. buyers in the seaborne market longer. Demand adjusts slowly. London’s deficit of nearby metal does not vanish in a week. A September 30 deadline on some tariff calendars is therefore not a climax. It is a date that can extend the distortion.
One mid-month tally had already put on the order of 700,000 tonnes into U.S. warehouses on the same logic. July’s 200,000-tonne seaborne print is the monthly proof. BMI, a Fitch unit, has said tariff uncertainty should keep supporting prices and that already-elevated levels are likely to cap near-term gains. Support and cap can be the same market.
The Mine Is the Other Half of the Alarm
Policy can rearrange metal. It cannot print concentrate.
Barcellos’s supply sentence is the one Canadian geologists will recognize. Global mine supply is deteriorating. Severe weather in Chile forced Antofagasta and Lundin to cut 2026 production guidance. Antofagasta’s range moved to 625–655 thousand tonnes from 650–700. Lundin’s moved to 300–325 from 310–335. The named assets were Los Pelambres and Caserones. That is not a modeler’s rounding error. That is a winter on the Andes and a mill that did not get the feed.
International Copper Study Group data cited in Friday coverage showed global mine production down 1.1% in the first half of 2026. Codelco and Freeport-McMoRan recorded double-digit declines. Producers representing about two-thirds of global supply, on Jefferies figures relayed by SP Angel, were down 3.5% in the first half and 4.1% in the second quarter. Chile, still the world’s largest copper country, had its weakest second quarter in nearly two decades and now expects annual output to fall 2.6%.
Morgan Stanley began the year forecasting supply growth. It now expects mine production to finish roughly unchanged or slightly lower — potentially the first annual decline since 2017. The bar chart that travels with that sentence is orange through 2025 and pink in 2026. Accidents, ore-grade decline, project slippage, and weather are the four horsemen every Canadian technical report already lists. This year they showed up in the same half.
On the concentrate side the tightness is feeding the next distortion. Chinese smelters have leaned harder on secondary feedstock. Copper scrap imports rose 15% year-on-year in July and 9% year-to-date. When the mine will not ship clean concentrate, the scrap yard becomes a mine. That is a price signal, not a substitute for a new pit in the Quesnel Trough.
Demand Did Not Take the Week Off
Electric vehicles, grid buildout, and AI data halls are the demand clause every copper note now carries. A planning range used for liquid-cooled data centers still runs 20–40 tonnes of copper per megawatt of applied power. Wood Mackenzie has data centers driving the bulk of a higher U.S. power-demand path. None of that paused because August payrolls printed 162,000. Gold sold off on that number. Copper held. That divergence is the week’s tell: industrial tightness versus a rates punch.
A structurally bullish period, in the language of the notes circulating this morning, is demand accelerating into a mine market that cannot grow. Building a mine still takes a decade. Buying a project can take a quarter. That arithmetic is why Hudbay’s Arizona Sonoran close and Faraday Copper’s agreed purchase of BHP’s San Manuel ground — shares equal to a 30% stake, still subject to remaining conditions — belong in the same file as Los Pelambres weather. At $14,000-plus, metal in the ground with a permit path is a different object than metal in a PowerPoint.
What This Means on a Canadian Desk
TSX and TSXV copper is not LME three-month. It is jurisdiction, capex, and a mill that may not exist. A tenth weekly gain in London does not finance a PEA. It does change the conversation in a boardroom that was waiting for $11,000 to feel brave.
Producers with Chilean or Andean exposure inherit both the price and the weather. Developers in the U.S. Southwest inherit the tariff rumor as a double edge: a duty that protects domestic cathode can also scramble the concentrate book their offtake assumes. Canadian juniors in Ontario and B.C. inherit a multiple that will expand on a squeeze and collapse if U.S. warehouses reverse into the seaborne market the day the policy uncertainty dies.
That last sentence is Barcellos again. Do not underwrite a perpetual tariff fog. Underwrite a mine that works if the fog lifts.
Equity beta will stay loud. Freeport can print a quiet Friday and still be a 2026 winner. A $20 million explorer will gap 8% on a Reuters copper lead and give it back on a guidance cut in Santiago. Treat those as different products. This publication will not publish a target list because a warehouse in New Orleans is full.
The Dates That Can Break the Story
A tariff decision that is clean — yes or no — rather than phased. A Chilean spring that restores Los Pelambres and Caserones faster than the cuts imply. A scrap flood that lets Chinese smelters ignore concentrate. A recession that hits housing and auto harder than data-center copper can offset. Any one of those turns a tenth weekly gain into a tenth week that was the top.
Until then the physical market is doing what tight markets do. It pays the person who has metal in the wrong place a premium to keep it there, and it pays London for the residual tonne.
Conclusion
Copper is approaching records because the United States is stocking against a duty and the Andean mine is having a bad year. The alarm on the chart is not a cartoon. Global mine output is on track for a possible first annual drop since 2017. Codelco and Freeport already printed the declines. Antofagasta and Lundin already cut the ranges.
A Canadian reader should take two sentences home. Uncertainty about a U.S. tariff can support London more than the tariff itself. A mine that does not ship cannot be replaced by a headline. Size the equity as a business. Size the metal as a warehouse. And do not confuse a tenth green week with a permit.
Important information
This article is for informational and educational purposes only. It is based on contemporaneous market reporting, including research notes and exchange data cited in public coverage dated early September 2026. It is not investment advice or a recommendation to buy, sell, or hold copper, copper futures, or any mining security. Company names appear as industry context. Production guidance, inventory figures, and tariff scenarios can change. Forward-looking statements are uncertain. Mining investments can result in loss of principal. Consult a licensed adviser. The author and publisher accept no liability for actions taken on the basis of this article. Past performance is not indicative of future results.