Tearing apart the Zambia smelter energy crisis

Zambia.

Every copper analyst this week is watching Chile. The strike risk at Escondida, the Codelco output miss, the usual checklist. What they're not watching — because it doesn't show up on the Bloomberg terminal summary — is that Africa's third-largest copper-producing nation is quietly throttling its smelter capacity. Not because of a labor dispute. Because the river is running low.

Zambia runs on hydropower. Roughly 85% of its national grid is fed by the Kariba Dam on the Zambezi and the Kafue Gorge system downstream. The rains failed again this season — the third consecutive below-average rainfall year across the Zambezi basin. Lake Kariba's water level is sitting at levels that constrain turbine output. The dam authority has been rationing power to industrial users since March. The smelters are first in line to get cut.

The mainstream frame is that Zambia is a developing-world infrastructure problem — chronic, sad, and irrelevant to the price of copper in London. That frame is wrong on the last part. Zambia produces roughly 800,000 tonnes of refined copper per year in a normal functioning year. That's about 3.5% of global mine output, but more importantly, it's refined copper — cathode, ready for the wire mill. Not concentrate. Not a processing step away. Cathode.

When the smelters go dark, that's not a mine output story. It's a finished-metal story. And LME doesn't warehouse concentrate. It warehouses cathode. The machine is about to feel this at the delivery point, not the mine gate.

Numbers. Strip everything else out.

The Copperbelt smelters — Mopani, Konkola, and the newer Chinese-operated Chambishi complex — are all electricity-intensive. A copper smelter needs roughly 500 kWh per tonne of output at the flash furnace stage, and then more again through the electrorefining circuit. Eight to twelve hours of load-shedding per day doesn't halve your output — it does worse than that. It disrupts the thermal cycling of the furnace, forces cold restarts, and accelerates refractory wear. The damage compounds. A smelter running at 60% power availability doesn't produce 60% of its normal cathode. It might produce 40%, and it spends the rest of the time managing equipment stress.

The Kariba reservoir level is measurable. ZESCO — the Zambian state utility — publishes outage notices. The International Copper Study Group publishes monthly output data with a six-week lag. None of this is secret. It's just not on the terminal summary.

The 180,000 to 220,000 tonne annualized cathode shortfall from Zambia alone is roughly equal to four weeks of total LME warehouse stock at current inventory levels. That is not background noise. That is a primary input into the physical supply balance, and it is not in the consensus model.

Here's what makes this worse than a simple supply disruption: there is no emergency bypass. When a Chilean concentrator goes offline, the global refining system has slack — smelters in China, Japan, and Europe can absorb additional feed if the concentrate shows up. The Zambian problem runs in the opposite direction. The concentrate is in the ground. The smelting capacity exists. What's missing is the electricity to run it, and you cannot import that through a pipeline.

The mining operators have been trying to build emergency diesel generation capacity for two years. Konkola's parent company announced an agreement with a Tanzanian independent power producer in late 2024. That project is not yet commissioned. Mopani — now majority-owned by a Zambian government vehicle after the Glencore exit — has been in financing discussions for a captive solar-plus-storage installation since 2023. Also not commissioned. In the interim, they're buying diesel gen at spot rates and running it for the highest-priority process steps only. Refractory maintenance is getting deferred. I've heard the phrase "running on fumes" used sincerely in an operations context before. This qualifies.

The wet season in the Zambezi basin runs November through April. Lake Kariba does not refill meaningfully before November at the earliest, and after three consecutive drought years, a single average season won't restore operating levels — the dam engineers will run conservative discharge protocols to rebuild the buffer. The curtailment the Copperbelt smelters are operating under has a minimum runway of five months, and more likely eight to ten. This is not a disruption. It is a constraint with a calendar attached to it, and the calendar is public information.

Meanwhile, the CFTC positioning data shows managed money net long copper at levels that look bullish but are largely crowded into the Chile-and-China thesis. They have the right direction. They have the wrong mechanism. That gap matters when the trade resolves — the crowded longs will be right for the wrong reasons, and the unwind when something cracks in their thesis could be violent even as the underlying physical supply remains tight.

The sequence runs like this. It's not fast — physical supply gaps never are — but it's mechanical once it starts.

The dangerous moment is a Chile headline resolution. If Escondida reaches a deal, the paper longs — built on strike risk — unwind fast. Copper sells off. The algorithms see weakness and add short. For a few days or a week, the price chart looks like the bears were right. Then the ICSG monthly revision lands, the LME warrant holders look at actual deliverable stock, and the physical buyers who pulled back come back in simultaneously. That's the sequence. I've watched it set up in palladium, in nickel, in aluminum during the Russian sanctions. The details change. The choreography doesn't.

Where does capital flow if this reads correctly? Not the diversified miners — BHP and Rio have enough non-copper exposure to dilute the thesis. The edge is in producers with Zambian and DRC-adjacent assets whose stock is being priced on headline copper sentiment, not on the specific Copperbelt energy dynamic. First Quantum Minerals' Zambian operations are the most direct expression of this — though First Quantum comes with its own Panama overhang that complicates the story. Ivanhoe's Kamoa-Kakula complex in the DRC doesn't have the same ZESCO exposure, which is exactly why it will outperform when Zambia's cathode shortfall becomes consensus knowledge.

The physical market is already pricing the shortage. The paper market is pricing a different story in the same direction. Those two will eventually be looking at the same data. They usually are — just never at the same time.