The Copper Squeeze Everybody Is Talking About But Not Pricing

How Beijing, Bad Geology, and Paper Markets Are Setting Up A Big Trade

Copper trades at roughly $5.90 per pound. The COMEX-LME spread has blown past 28%. Forty percent of COMEX inventory is unwarranted—locked in arbitrage, unavailable to be settled by the financial markets.. The world’s second-largest copper mine is offline until at least 2027. And China’s smelters just agreed to $0 per metric ton in treatment charges for 2026—the lowest in the history of the industry. They are paying miners for the privilege of refining their copper ore. The market has priced a copper shortage in but not the size. And not the speed.

The Geology Problem Nobody Can Engineer Around

Start with the rock. Average copper ore grades have declined roughly 40% since 1991, according to the IEA’s Global Critical Minerals Outlook. A 2016 study in MDPI Resources found Chilean mines experienced a 25-29% grade decline in just ten years. BHP’s Escondida—the world’s largest copper mine—saw grades fall from 2.5-3% in the 1990s to around 1% by 2025. That means miners must move roughly 1.7x more material to produce the same unit of copper. In Chile—home to a quarter of global output—energy consumption rose 46% against just a 30% production increase over a single decade, according to a study published in MDPI Resources. This is not a cyclical problem. It is thermodynamic.

The pipeline is not coming to the rescue. S&P Global’s mine development study found average lead times have stretched to 17.9 years for projects starting production in 2020-2023, up from 12.7 years for those starting in 2005-2009—a 41% increase. In the United States, the average is 29 years—second only to Zambia at 34 years. Even if copper hit $15,000 per metric ton tomorrow, no new greenfield mine could respond within a decade.

Supply Disruptions Are Structural, Not Episodic

The sell-side treats supply disruptions as one-off events. They are. But when they happen, they have larger impacts then ever before. In September 2025, a catastrophic mudslide at Freeport-McMoRan’s Grasberg mine in Indonesia—the world’s second-largest copper mine—forced a complete shutdown and force majeure declaration. Approximately 800,000 metric tons of mud inundated underground tunnels, killing two workers with five still missing. Freeport now projects 2026 output roughly 35% below previous guidance, with full recovery not expected until 2027. ING estimates the mine accounted for around 3% of global copper supply. The lost production—estimated at 500,000 tonnes over 12-15 months—exceeds the annual output of most mid-tier mines.

Grasberg is not alone. Cobre Panama, which represented roughly 1% of global copper supply, remains closed after Panama’s Supreme Court ruled the mining concession unconstitutional in November 2023. Peru’s Las Bambas and Constancia operations face recurring disruptions from political protests. Benchmark Mineral Intelligence reports Kamoa-Kakula in the DRC lost an estimated 300,000 tonnes across 2025-2026, while Codelco’s El Teniente posted its lowest monthly production in 20 years. These are not tail risks. They are the operating environment.

Beijing’s Smelter Strategy: Capacity to Suffer

China now controls over 50% of global copper refining capacity. CRU Group estimates CSPT smelting capacity alone at 9.1 million tonnes in 2025, rising to 9.6 million tonnes in 2026. In December 2025, Antofagasta agreed to $0/mt treatment and refining charges for 2026—the lowest ever recorded. Spot TC/RCs hit negative $66.60/t by October 2025, meaning Chinese smelters were effectively paying miners to process their concentrate.

The CSPT announced 10%+ utilization cuts for 2026. China’s Nonferrous Metals Industry Association publicly stated it “firmly opposes” free and negative processing charges. But the strategic logic is clear: Beijing is willing to operate smelters at a loss to absorb global concentrate supply and squeeze out non-Chinese refiners. Japan, South Korea, and Spain issued a joint statement criticizing the TC/RC collapse. Japan’s JX Advanced Metals cut output. Glencore required a government bailout to keep Mount Isa running. China produced 9.7% more refined copper year-over-year through October 2025 despite the collapse. This is the airline playbook applied to critical minerals: sustain losses to capture market share, force competitors out, then control pricing. Western financial models that assume Chinese smelters behave like Western companies are pricing in a fiction.

The Chinese know that the United States needs copper to update its electrical grid. To feed its AI Data Centers. To create robots. And to feed the military-industrial complex.

All of these customers are price inelastic. Think $META cares if they pay $9k a ton or $15k a ton when they need to get Hyperion online? They don’t. (More on that shortly.)

The Paper-Physical Disconnect

Financial markets are structurally disconnected from physical copper reality. Sprott’s analysis documents the COMEX-LME spread surging past 28% in mid-2025 after tariff speculation drove massive U.S. inventory accumulation. COMEX stocks rose 388% from year-end 2024 to over 454,000 tonnes by December 2025, while LME inventories dropped 38.6% to roughly 167,000 tonnes over the same period. Benchmark estimates nearly 900,000 tonnes of copper became “economically locked” in U.S. warehouses by the arbitrage—roughly 2% of global demand effectively removed from the tradeable market.

The composition of remaining LME stock makes this worse. A large portion is Russian or Chinese-origin metal that Western consumers are reluctant or unable to purchase due to sanctions and quality concerns. LME warehouses hold less than 200,000 tonnes—just a few days of global consumption. As of January 2026, 38-40% of COMEX copper was unwarranted, meaning it was committed to delivery or locked in arbitrage positions. The headline number—“record COMEX inventories”—masks a market that is physically tighter than at any point in recent memory.

And COMEX stores cathode. The stockpiles are still useless. It’s akin to having hundreds of heads of cattle and telling starving people, that only have a few grills (wire fabricators), that everything is fine and they’ll be able to eat. Maybe. But not as soon as they’d like.

Inelastic Demand from Buyers Who Cannot Substitute

The consensus substitution fallacy runs as follows: at a high enough price, end users switch from copper to aluminum. This is historically accurate for some applications. It is physically impossible for the applications driving marginal demand growth. Electric vehicles require 2.9x more copper than internal combustion engines. AI data centers need copper for thermal management, power distribution, busbars, and circuit boards—applications where aluminum’s inferior conductivity cannot perform in compact form factors. Defense platforms like the F-35 require copper for high-conductivity applications where substitution would require fundamental redesign.

In January 2026, Amazon Web Services signed a two-year agreement with Rio Tinto to purchase copper from a newly reopened Arizona mine for U.S. data center components—the first domestic copper supply deal for AI infrastructure in over a decade. The mine produces 25 million pounds of cathode annually using Rio Tinto’s Nuton bioleaching technology—but even that is not enough for a single hyperscale data center. When Amazon starts buying copper at the mine gate, the substitution thesis is dead.

S&P Global’s January 2026 study projects global copper demand rising roughly 50% to 42 million metric tons by 2040. Without substantial new investment, primary supply could actually decline to 22 million metric tons—less than current production. The gap is not 5% or 10%. It is potentially 10 million metric tons, representing the most significant supply shortfall in any industrial metal this century.

Copper as Gateway Drug: The Critical Minerals Chain

An underappreciated dimension of the copper thesis: copper mining and concentrate processing are the primary source for a suite of critical byproduct minerals—and disrupting one disrupts all of them simultaneously.

Start with rhenium. Roughly 80% of the world’s rhenium supply comes from porphyry copper deposits—it is literally a byproduct of a byproduct, extracted from molybdenite concentrates that are themselves recovered during copper processing. The F-22 and F-35 use third-generation single-crystal superalloys containing 6% rhenium in their turbine blades. There is no substitute at operating temperatures above 1,100°C. Chile’s copper mines produce 55% of the world’s rhenium. When Codelco’s El Teniente posts its lowest output in twenty years, it is not just a copper story. It is a fighter jet engine story.

Tellurium follows the same pattern. Over 90% of global tellurium is recovered from anode slimes during electrolytic copper refining. First Solar’s cadmium telluride panels—the dominant thin-film photovoltaic technology—consume 40% of global tellurium production. Molybdenum is marginally better diversified, but roughly 60% of global supply still comes as a copper byproduct, and it is irreplaceable in the high-strength steel alloys used in tank armor, submarine hulls, and missile components. Selenium, another copper refining byproduct, is a critical input for semiconductor manufacturing.

Here is the strategic problem nobody is modeling: China controls over 50% of global copper refining. That means China also controls the majority of the world’s byproduct recovery for rhenium, tellurium, molybdenum, and selenium. A copper supply disruption does not produce a single-commodity shortage. It cascades through fighter jet engines, solar panels, tank armor, and semiconductor fabs simultaneously. The Pentagon has identified all four minerals as strategic vulnerabilities. The market prices none of them as connected.

What the Market Is Missing

Paper markets dominated by corporate hedging (roughly 70% of commodity market volume) and London-based bank desks running models calibrated to USGS projections with 5-10% deviation bands that are not designed to price a structural physical shortage driven by geology, geopolitics, and industrial policy. The banks mostly exited physical commodities post-2008. There are only a few market makers (shout out J Aaron/GS) compared to FX. Regulatory capital requirements prohibit most financial institutions from holding physical metal. The only players who can express this thesis through physical acquisition—family offices, sovereign wealth funds, and flexible-mandate allocators—are precisely the players least represented in current copper price discovery.

The base case for copper at $17,000/mt or higher within two to three years is not a speculative fantasy. It is the arithmetic of 40% ore grade decline, 17.9-year mine lead times, 500,000+ tonnes of Grasberg production offline, negative TC/RCs, a 28% COMEX-LME spread, inelastic demand from EVs, AI, and defense, and a market where the dominant refiner operates strategically rather than economically. The question for allocators is not whether copper reprices. It is whether you have any exposure when it does. And how to properly express that investment idea.

— — —

Sources and methodology notes available at brubarian.com. Charts built with Datawrapper. All interpolated data points disclosed in chart source lines.