|
Part I
The Mechanism
|
Copper smelters.
The headline machine is running COMEX copper at $6.49 a pound and calling it a supply story. Bloomberg says record prices. CNBC says electrification. The analysts are pointing at mine output, at Chile, at the demand side of the ledger. They're not wrong about copper being tight. They're just looking at the wrong part of the engine.
The part that's actually breaking is in the middle. Not the mine. Not the end-user. The smelter — the facility that turns rock into metal. The entire global midstream layer of copper processing is operating on economics that no longer make mathematical sense. And almost nobody on a trading desk is talking about it.
Treatment and refining charges — the fees miners pay smelters to convert concentrate into finished metal — are the heartbeat of midstream economics. They've been the smelter's core revenue line for over a century. This year, for the first time ever, the annual benchmark settled at zero. Not low. Not compressed. Zero.
And then spot went negative. On September 18, SMM's spot copper concentrate TC index hit –$221.89 per dry metric tonne. That means smelters are paying miners to take their own raw material. It's like a restaurant paying the farmer for the privilege of cooking the steak. Nobody designed a business model for this. And yet here we are.
|
Part II
The Diagram
|
Story off. Diagram up.
Here's how you build a smelter crisis. Global copper smelting capacity is projected to grow from roughly 26 million tonnes in 2024 to 30 million by 2028, per ICSG data. China alone accounts for about half of global smelting and more than 90% of the capacity growth since 2005. In 2026, CSPT member capacity hit 9.61 million tonnes — 83% of China's total primary smelting.
Meanwhile, global mine production grew roughly 1% last year. China's refined copper production grew 8%. That's the gap. That's the entire story in two numbers.
The TC progression tells the story mechanically: $80/t in 2024. $21.25/t in 2025. Zero in 2026. And spot has blown past zero into territory the industry considered mathematically impossible three years ago. Monthly average spot TCs tracked by Mysteel fell from –$46.78 in January to –$120.03 in June to –$127.20 in early July. Individual spot trades have reportedly hit –$220.
To understand why, you need to see the flow:
Too many smelters chasing too little rock. China built the capacity because smelting is downstream industrial policy — controlling the conversion layer means controlling the metal. Whether it makes money on TC/RCs is secondary to Beijing. Whether it makes money at all is now the question nobody wants to answer out loud.
Japan's Sumitomo Metal Mining already cut 2025/26 fiscal year refined output by 2.6%. Pan Pacific Copper fought for a rollover on benchmark fees rather than accept the zero. The non-Chinese smelters don't have access to cheap state capital, subsidized energy, or advanced byproduct recovery circuits. They're the first dominoes.
|
Part III
The Weak Link
|
The consensus assumption — the one propping up every model that still shows copper smelter margins stabilizing — is that CSPT's announced 10% production cut will hold. It was headlines in December. It was reassuring. It was also, apparently, optional.
When the actual earnings guidance came out, the story inverted. Yunnan Copper raised 2026 output guidance to 1.71 million tonnes, up from 1.64 million produced last year. Jiangxi Copper — China's largest — raised 2026 output guidance to 2.39 million tonnes, up from 2.38 million in 2025. Mining.com's headline put it plainly: no sign yet of China's plan to cut copper output in major smelters' results. I've seen this movie before in OPEC. The cartel announces. The members cheat. The cut never materializes.
They could get away with it because sulfuric acid was doing the heavy lifting. Smelters produce 3 to 4.5 tonnes of sulfuric acid per tonne of refined copper. When acid prices climbed from ¥1,015/tonne in January to ¥1,768 in June, that byproduct stream papered over the TC losses. It was the safety valve. Run the furnaces at a loss on processing, make it back selling acid to fertilizer plants and chemical manufacturers.
That valve just shut. Since July 10, sulfuric acid prices from copper smelters have declined for 11 consecutive weeks. Some Chinese regions are printing below ¥1,000 per tonne — roughly 43% below the June peak. Acid fell 11% in September alone. The overcapacity in smelting created overcapacity in acid output. The cure was always going to poison itself.
So now the smelters are losing money on TC/RCs. They're losing the acid cushion. Gold and silver byproducts help the ones with advanced recovery circuits, but not everyone has that hardware. And CSPT is trying to expand its membership — now courting every smelter in the country — which tells you the current 16-member group couldn't enforce discipline the first time around.
I've been on the wrong side of a smelter squeeze exactly once, in 2022 with zinc. The dynamics are slower than a mine disruption — it takes quarters, not days, for the midstream pain to translate into refined metal scarcity. But when it does, the move is larger, because the market wasn't watching. The smelter is the boring part. Nobody writes headlines about smelter margins until the refined metal doesn't show up.
|
Part IV
The Chain Reaction
|
The sequencing from here is mechanical. You don't need a narrative. You just need to follow the pressure through the system.
Step one is already happening. Sumitomo cut. Pan Pacific is fighting for margin preservation. Seven Chinese smelters confirmed maintenance shutdowns in recent quarters. SMM reports some have already signaled they'll cut production and warned clients that long-term contract deliveries will be reduced. The voluntary cuts CSPT couldn't enforce? Physics is about to impose them involuntarily.
Step two is the one the refined copper market hasn't priced. China's refined output growth is expected to slow sharply — SMM already flagged this — but "slow" is doing a lot of work in that sentence. If the acid cushion stays below ¥1,000 through Q4 and spot TCs stay below –$200, you're not looking at slower growth. You're looking at an outright output decline in H1 2027. Against demand that's still climbing on grid buildout, EV wiring, and data center infrastructure.
Step three is where the copper price goes places the consensus models aren't built for. LME already hit $14,527.50/t in January on the mine-side story alone. The smelter-side story hasn't been priced in yet. When refined copper gets scarce — not concentrate, not ore, but the finished cathode — fabricators don't wait for the chart to tell them what to do. They scramble. We saw the cash-to-three-month premium blow out to $545/t in August. That was a preview, not the main event.
Where does the capital go? Not into the smelters. Their margins are collapsing in real time. The edge, if there is one, sits upstream — with the miners who control the concentrate the smelters are now paying a premium to access. Pure-play copper miners with unhedged production and high-grade deposits outside the strike zones. Companies whose concentrate is the scarce input every smelter on the planet is fighting over.
The old copper trade was: buy the metal, bet on demand. The new copper trade is: follow the midstream pain. Smelter economics have inverted. The processing layer that was supposed to be a cost center has become the chokepoint. And chokepoints, in my experience, don't resolve quietly.
Everyone is watching the copper price. Almost nobody is watching the smelter. That's usually when it matters most.
Sources: SMM (Shanghai Metals Market), ICSG, Mysteel, Mining.com, CRU Group, Fastmarkets, Reuters, Finimize
