|
Part I
The Mechanism
|
Cobalt just dropped 31% in a quarter. Bloomberg ran the headline: Congo's fading rally is testing its price-control efforts. The read on fintwit is simpler — the quota system is leaking, supply's back, rally's done.
None of that is what happened.
What happened is a pipe cleared. The DRC's export quota caps national cobalt shipments at 96,600 tonnes a year — less than half of what the country mined in 2024. But quota allocation is not quota execution. Q1 and Q2 allocations got stuck. Customs bottlenecks. Sampling delays. Logistics chokepoints on the single truck corridor from Kolwezi to Durban. By the time ARECOMS — Kinshasa's strategic minerals regulator — threatened to forfeit unused allocations on June 29, producers panic-shipped everything at once.
That's not supply returning to market. That's six months of backed-up inventory hitting the dock in eight weeks. A slug, not a flow. The annual cap hasn't changed. ARECOMS hasn't raised quotas. The structural deficit Fastmarkets modeled at 10,700 tonnes for 2026 hasn't moved.
What moved was price — and the market is now treating a plumbing event as a thesis change.
|
Part II
The Diagram
|
Numbers. Nothing else.
The DRC mined approximately 226,000 tonnes of cobalt in 2024. The export cap for 2026: 96,600 tonnes. Of that, 87,000 tonnes are allocated to registered producers on a pro-rata basis. The remaining 9,600 tonnes sit in ARECOMS' strategic reserve — discretionary, unreleased, under direct government control.
CMOC — the world's largest cobalt producer, minority-linked to CATL — received a 31,200-tonne allocation for 2026. That's roughly 32% of the national quota. The problem: CMOC produced 117,549 tonnes in 2025. The arithmetic is brutal. Three-quarters of what the biggest producer can mine stays in the ground or in stockpile.
Glencore made the opposite calculation. H1 2026 cobalt output: 10,200 tonnes, down 46% year-on-year. Total own-sourced copper production: up 15% to 397,000 tonnes. They said it publicly: their DRC assets are holding cobalt in solution as export quotas constrain shipments, while prioritising copper production. When the world's second-largest cobalt miner holds output in solution rather than shipping it, that is not a supply-neutral event.
Then there's the forfeiture mechanism. In late June, ARECOMS notified producers that all Q1 and Q2 allocations unused by June 30 would be seized and transferred to the national strategic stockpile. Roughly 20,000 tonnes — worth approximately $1.1 billion — reportedly faced forfeiture. The result: a stampede to ship before the deadline, which created the Q3 glut that every headline writer is now misreading as a trend reversal.
Indonesia is the only at-scale alternative. Expected output: 59,800 tonnes in 2026, up 21% year-on-year. But nearly every tonne comes as a byproduct of nickel HPAL processing. It doesn't respond to cobalt price signals. It responds to nickel economics and Indonesian ore quotas. If you're modeling Indonesian cobalt as the relief valve, you're modeling a valve that's bolted to someone else's pipe.
Fastmarkets' 2026 base case: a 10,700-tonne global deficit, with demand at 292,300 tonnes. That number assumes full quota execution. Every tonne that doesn't ship widens it.
|
Part III
The Weak Link
|
The payable is the signal everyone ignored.
Cobalt hydroxide payables — the ratio of intermediate price to refined metal price — hit 100% in February 2026. The long-run norm is 55–75%. A hundred-percent payable means refiners are paying full metal value for an intermediate product. That's a physical market screaming that it can't get enough material.
By mid-September, the CIF Asia MHP payable had crashed to 65.8%. Chinese battery makers — CATL's Brunp, CNGR — slashed what they'd pay for Indonesian MHP and DRC hydroxide. The Q3 slug gave them leverage, and they used it.
Here's what the payable crash actually means: Chinese refiners are pricing a one-time inventory flush as a permanent supply restoration. I've seen this move before. It's the commodity market equivalent of declaring the recession over because one GDP print bounced.
The structural picture hasn't changed. Benchmark Mineral Intelligence projects ex-DRC inventories falling to approximately one month of demand by Q4. The DRC banned copper and cobalt concentrate exports outright on June 29 — that's not just the quota on finished hydroxide, that's feedstock for every non-DRC smelter on the planet. And ARECOMS' chairman told Fastmarkets the regulator may reduce quotas further if the market needs rebalancing. Translation: Kinshasa has a ratchet, and it only turns one direction.
Meanwhile, 99% of global cobalt production is a byproduct. Copper mining. Nickel processing. There are essentially no primary cobalt mines left operating at scale on Earth. The metal that keeps your EV battery from catching fire has no dedicated supply chain. Every tonne depends on someone else's economics.
Ninety-nine percent byproduct. A single-country quota system. A forfeiture mechanism that punishes slow shippers. And a refining sector that just priced a slug as a new normal. That's not one weak link. That's a chain made of them.
|
Part IV
The Chain Reaction
|
The sequence from here is mechanical.
If Q4 allocations hit the same customs bottleneck — and nothing in the DRC's single-corridor logistics has been upgraded — producers face another forfeiture deadline. Ship or lose it. But this time, the easy inventory is gone. Glencore has already pivoted to copper. CMOC has sufficient finished inventory for near-term quotas but is building DRC-side stockpile, not expanding shipments. The Q3 flush was the one-time release. There isn't another slug behind it.
When the payable floor reasserts — and 65.8% is below the level that makes Indonesian MHP economically rational for most refiners — the first move is a scramble for hydroxide already on water. Second move: spot cobalt metal on the LME reprices. At $39,140 per tonne on October 2, the contract already sits 30% below its Q2 peak. That's the gap.
Where does capital go? Not into the lithium battery ETFs. LIT, BATT — those are diluted with lithium carbonate exposure, nickel plays, and companies whose cobalt revenue is a footnote. The edge, if it exists, is in the handful of assets positioned outside the DRC-China corridor. Electra Battery Materials is building North America's first cobalt sulfate refinery in Ontario — commissioning 2027, backed by C$17.5 million from the province. Jervois Global's Idaho Cobalt Operations had construction suspended and sit idled, waiting for exactly the price signal the market just gave. These aren't mines that need a decade of permitting. They need a sustained price above $20–25 per pound — roughly $44,000–$55,000 per tonne — to switch on.
The machines read a 31% quarterly drop and shorted the thesis. The physical market reads a 96,600-tonne annual cap against 226,000 tonnes of production capacity and a forfeiture mechanism that gets tighter every quarter. In my experience, when the paper market and the physical market tell you different stories, the physical market wins. It just takes one more forfeiture deadline to prove it.
