|
Part I
The Mechanism
|
Cobalt.
Bloomberg ran the Q3 headline three weeks ago: “Cobalt Prices Plunge 31% as Oversupply Returns.” CNBC had a battery analyst on Tuesday explaining how LFP chemistry is making cobalt obsolete. If you’ve been reading this letter for more than a week, you know the drill. They’re watching the price chart. The machine is somewhere else entirely.
Twenty-one months ago, the Democratic Republic of Congo—which mines more than 70% of the world’s cobalt—suspended all exports. Cold. No notice. Then in October 2025, Kinshasa replaced the ban with a quota system capping exports at 96,600 tonnes a year. That’s less than half of what the country produced in 2024. The DRC didn’t just intervene in the cobalt market. It built an OPEC for battery metal in under a year.
The “oversupply” narrative is reading from a script that expired in February 2025. Prices did fall 31% in Q3—from an elevated base, after a 300% rally in hydroxide over the prior year. The correction was real. The structural shift underneath it is more real. And it just got a fresh data point.
Glencore’s Q3 production report, filed October 29, showed Mutanda—once among the world’s largest cobalt mines—produced zero cobalt for the entire first nine months of 2026. Not reduced. Zero. A marquee cobalt mine isn’t mining cobalt. That’s not a headline you saw on CNBC.
Just a few years ago, they laughed at him…
And said that what Elon was planning to do was impossible.
In 2018, the New York Post even called Elon “a total fraud”...
And it proves the media was 100% wrong about Elon.
|
Part II
The Diagram
|
Story off. Numbers on.
Start with the quota architecture. ARECOMS—the DRC’s strategic minerals regulator—sets the ceiling. Of the 96,600-tonne annual cap, 87,000 tonnes go to producers: CMOC gets 31,200 tonnes (27% of its prior-year Congolese output), Glencore gets 22,800 tonnes. The remaining 9,600 tonnes sit in a “strategic reserve” that ARECOMS allocates at its discretion. In June, ARECOMS forfeited all unused H1 quotas and reassigned them to the state reserve. Use it or lose it. The regulator is not bluffing.
Now look at what the producers are actually doing with those quotas. Glencore’s own-sourced cobalt production: 13,700 tonnes in nine months, down 52% year-on-year. Mutanda—once among the world’s top cobalt producers—reported zero. Glencore is retaining cobalt in hydrometallurgical solution rather than converting it to saleable hydroxide. They’ve pivoted the entire DRC operation toward copper cathode, which rose 37% year-on-year. The cobalt stays in the tank. Literally dissolved, unsold, unprocessed.
Meanwhile, Indonesia—the supposed swing supplier—is choking on its own input costs. HPAL plants need sulfuric acid. Sulfuric acid needs sulfur. Sulfur ships through the Strait of Hormuz. You see where this connects. Indonesian MHP production fell in Q1 2026, and the sulfur supply disruption has kept costs elevated all year. The alternative sources—pyrite, imported acid, calcium sulfate—are stopgaps, not solutions.
China refines 78% of global cobalt. It holds stakes in 15 of 19 Congolese mines. But the quota flipped the power dynamic. Hydroxide payable—the percentage of the metal price that the DRC intermediate commands—moved from around 55% to 100%. The margin shifted from Shanghai refiners to Kinshasa producers. China still processes the metal. The DRC now sets the terms.
This is the part that never made it into the “oversupply” headline.
|
Part III
The Weak Link
|
ARECOMS holds quarterly review authority over the quota. The chairman told Fastmarkets in August that quotas may be reduced if supply-demand imbalance deteriorates. Not increased. Reduced.
That’s the part the machines aren’t reading.
The Q3 price drop—hydroxide down 31% to around $38,000/tonne, metal easing about 11%—happened because the market saw DRC exports resuming and assumed the worst was over. Higher exports from the DRC did add supply. But the quota ceiling is structural, not temporary. And the forfeiture mechanism means any producer that doesn’t ship their allocation loses it to the state. The system ratchets. It doesn’t relax.
Supply disruption rumours out of the DRC hit the electronic trading platform on November 3. Refined cobalt futures went limit-up from around 200,000 yuan per tonne. I’ve seen this pattern enough times now. Fundamentals hadn’t improved—SMM said it plainly—but the market is a coiled spring sitting on a −80,000-tonne ex-quota deficit, and any match lights the fuse.
The machines reading Q3’s decline see weakness. They short it. They don’t know Mutanda produced zero cobalt for nine months. They don’t check ARECOMS communiqués. They have never, to my knowledge, tracked a hydroxide payable ratio.
The LFP substitution narrative is real at the margin—cobalt intensity per kilowatt-hour is declining. But batteries are 70%+ of total cobalt demand, and the absolute volume keeps growing because EV production keeps growing. You can reduce the cobalt per car and still need more total cobalt if you’re building enough cars. The sell-side loves the per-unit decline story. They forget to multiply by the number of units.
|
Part IV
The Chain Reaction
|
The sequence is mechanical.
ARECOMS has the lever and the stated willingness to pull it. If prices stay soft through Q4, the regulator’s incentive is to tighten—not loosen—the quota. Every tonne not exported at a premium is revenue Kinshasa leaves on the table. The DRC watched OPEC’s playbook. They learned the right lesson: cut into weakness, not into strength.
If the quota gets cut—even 10–15%—Chinese refiners face a feedstock scramble. Indonesia can’t backfill because acid costs make half the HPAL plants uneconomic at current cobalt prices. Recycled black mass added supply in 2026 but nowhere near enough to offset a further DRC cut. The refined deficit, already at −16,000 tonnes, blows out.
Where does the capital go? Not into the broad battery metal ETFs—those are diluted with lithium names still fighting their own oversupply narrative and nickel producers drowning in Indonesian NPI. The edge, if there is one, is in the DRC-exposed producers who received the largest quota allocations and have unhedged hydroxide inventory sitting on-site.
CMOC, with 31,200 tonnes of quota and the Kisanfu mine producing 80,000+ tonnes of hydroxide a year, is the single most leveraged play on a tightening quota regime. They produce far more than they can export. That stored inventory is an option on higher prices that the market values at approximately zero.
Glencore made a different bet—pivot to copper, let cobalt sit in solution, wait. If prices rip, they re-process. If they don’t, the copper revenue covers the carry. It’s the kind of trade a company makes when it expects the quota to last longer than the market thinks. I’ve made similar bets on timeline mismatch. They work until the market catches up. The question is always when, not if.
Financial layer says correction. Physical layer says structural deficit masked by a one-quarter price dip. ARECOMS has the valve. They’ve shown they’ll turn it. In this machine, the regulator is the mechanism—and the mechanism only turns one direction.
Sources: ARECOMS, Glencore Q3 2026 Production Report, Cobalt Institute, Benchmark Mineral Intelligence, Fastmarkets, Shanghai Metals Market (SMM), IEA, Reuters, Ecofin Agency
