Tearing apart the zinc smelter bottleneck

Zinc.

The sell-side consensus says zinc should be falling. Chinese property is still a wreck. Global construction spending is soft. Every second analyst note from the last six months has the word "surplus" in the title. One headline from three weeks ago literally reads "Zinc prices expected to decline in 2026 due to weak demand."

Meanwhile, zinc just touched $3,703.50 a tonne — a four-year high. The highest price since August 2022. If you're confused, it's because you're looking at the demand gauge. The failure is somewhere else entirely.

This is not a mining problem. Global mined zinc output surged 4.8% in 2025 after three consecutive years of contraction. New projects — Ivanhoe's Kipushi mine in the DRC and Boliden's restarted Tara mine in Ireland — are pouring concentrate into the system. The miners did their job.

The smelters didn't. That recovered ore flow translated into just 1.7% growth in refined zinc production. And all of that growth came from China. Western refined output actually contracted. The bottleneck isn't underground. It's in the processing layer — and it's getting worse.

Story off. Here's the engineering diagram of a smelter sector falling apart in sequence.

In October 2025, the ILZSG forecast a comfortable 271,000-tonne global zinc surplus for 2026. By April 2026, they revised that to a 19,000-tonne deficit. That is a 290,000-tonne swing in six months. The scale of the forecast error is genuinely remarkable. It happened because the models assumed recovered mine supply would flow through to refined output. It didn't.

The casualty list tells the story. Glencore's Kazzinc complex in Ust-Kamenogorsk, Kazakhstan — 167,000 to 190,000 tonnes per year of capacity — suffered a fatal explosion on May 5 and is still running at reduced output. Two weeks later, Nexa Resources suspended operations at Cajamarquilla in Peru, the largest zinc smelter in Latin America, after a fire. It restarted May 29 but lost weeks of production.

Those two plants alone account for roughly 520,000 tonnes of annual capacity. Then add Toho Zinc's Annaka smelter in Japan — permanently closed by March 2025. Young Poong's Seokpo refinery in South Korea — court-ordered shutdown for environmental violations. Western refined output didn't just slow down. It got hit by a sequence of structural and operational failures that the models never priced in.

And the treatment charges confirm it. The 2026 benchmark TC settled at $85 per tonne — a modest recovery from historic lows, but nowhere near enough to incentivize restarts. Spot TCs on imported concentrate in China have slumped to around minus $100 per tonne. Negative. Smelters are paying for the privilege of processing ore. I've been watching this market a long time and negative TCs still make me blink.

That flow diagram is the entire story. More ore in. Less metal out. The processing layer is where the deficit is actually forming. The analysts who called for surplus were reading the mine data and ignoring the smelter data. A classic mistake — like checking the reservoir and ignoring the cracked pipes.

Everyone looking at this market assumes there's a relief valve. Historically, there was: when the West ran short of refined zinc, Chinese smelters ramped exports and metal flowed into LME warehouses. That's the trade the machines are modeling. That's the equilibrium the forecasters assume.

The valve is closing.

Chinese refined zinc output grew 6.7% in 2025. Western output contracted. China is approaching what Citi analysts call refined zinc self-sufficiency — the tipping point where Chinese smelters produce enough to serve domestic demand without importing, and without generating reliable export surplus for the rest of the world.

Look at the numbers side by side. SHFE zinc inventory: nearly doubled to roughly 147,000 tonnes this year. China is well-supplied. LME stocks: below 100,000 tonnes and dropping. The Western market is running on fumes. These are not two views of the same market. They are two separate markets that the models still treat as one.

China briefly turned net exporter of refined zinc in Q4 2025 when the LME squeeze opened an arbitrage window. Some metal moved. Then it stopped. Outbound shipments over Q1 2026 totaled just 10,600 tonnes. The arbitrage has to be fat enough to pull Chinese metal west, and even then, the volumes are a rounding error against a 14-million-tonne global market. I've seen traders bet on Chinese relief flows three times in the last eighteen months. It worked once, briefly, and then the window shut.

And here's the part nobody is talking about: another Chinese smelter in central China is scheduled for maintenance in August. Output drops another 1,000 to 1,500 tonnes. A mine in southwest China is cutting concentrate production. These are small volumes individually. But in a market running a 19,000-tonne deficit on paper — and probably deeper in reality after the Kazakhstan and Peru incidents — every tonne matters. The margin for error is gone.

The sequence from here is mechanical. If you've watched any physical commodity squeeze — and we've had several in the last four years — you can already see the choreography forming.

We watched a preview of this in October 2025. LME on-warrant zinc dropped to 24,425 tonnes. Cash-to-three-months spread hit $323 — the widest since at least 1997. Six separate entities held long positions whose combined total entitled them to at least 300% of available stock. The bears who had been sleepwalking into shorts got eviscerated. It took weeks for Chinese metal flows to partially relieve the pressure, and even then, it was temporary.

The conditions now are arguably worse. LME stocks are below 100,000 tonnes and draining. The Kazzinc complex is still running at reduced capacity with no announced timeline for full restoration. Another smelter in central China goes into August maintenance. The ILZSG's 19,000-tonne deficit estimate was made before those additional curtailments. The actual deficit is likely deeper.

Where does capital go? Not into the diversified mining majors. Their zinc exposure is diluted and their forward hedges mean they'll sit out a spot spike. The edge — if there is one — is in mid-cap producers with unhedged output tied to spot pricing, smelting capacity that's actually running, and reserves in jurisdictions that aren't currently on fire, on strike, or in court. Ivanhoe's Kipushi just set a production record. Vedanta's Gamsberg Phase 2 is ramping up this month. Those are the operations adding real tonnes to a market that needs them.

The sell-side says surplus. The smelter layer says deficit. The LME backwardation says deficit. The casualty list says deficit. When every piece of physical evidence disagrees with the forecast, I know which one I trust. It just takes longer to play out than you want it to — and the October squeeze proved that even the relief rallies can be violent enough to rearrange your risk budget overnight.