Tearing apart the DRC cobalt chokepoint

Cobalt.

The sell-side narrative is tidy. EV makers are switching to LFP cathodes. Cobalt is being "designed out." Every battery conference since 2023 has run some version of the same panel: "The Post-Cobalt Future." Bloomberg ran a piece last month about the commodity nobody needs anymore. Clean charts. Downward arrows. Case closed.

Meanwhile, in the physical world, the Democratic Republic of Congo just put 75% of the planet's mined cobalt supply under a quota lock — and the refineries in China that depend on it are running out of feedstock.

The "cobalt is dead" thesis has a problem. NMC cathodes still dominate every premium EV platform, every long-range cell, and every aerospace and defense battery contract on earth. LFP is cheaper. It's also heavier, lower-density, and useless at the high end. The market didn't stop needing cobalt. It stopped paying attention to it. There is a difference.

The DRC's regulatory body — ARECOMS — didn't just restrict exports. It built a quota architecture that controls how much cobalt leaves the country, who ships it, and on what timeline. This is not an embargo. It's a valve. And someone in Kinshasa just tightened it while the analysts were busy writing eulogies.

Numbers. The DRC capped 2026 cobalt exports at 96,600 tonnes. Of that, 87,000 tonnes are distributed pro rata to qualifying producers. The remaining 9,600 tonnes sit in a discretionary reserve controlled by ARECOMS — released or withheld at the regulator's judgment.

Monthly cap: 7,250 tonnes. In theory, strict use-it-or-lose-it — though ARECOMS has granted extensions and rollovers on a case-by-case basis, including carrying Q4 2025 quotas into Q1 2026. The discretion cuts both ways.

Read that again. CMOC, the single largest cobalt producer on the planet, mined nearly 118,000 tonnes of cobalt last year. Its 2026 export allocation is 31,200 tonnes. That means roughly 73% of what it pulls out of the ground has no permitted exit route from the DRC. The company maintains guidance of 100,000 to 120,000 tonnes of production for 2026 anyway — because the ore still has copper in it, and copper pays the bills. The cobalt piles up on-site.

Glencore, meanwhile, has pivoted to what it calls a "copper-first" strategy in the DRC. Translation: they're mining the same polymetallic ore but deprioritizing cobalt recovery. Effectively removing tonnes from the exportable supply even within the quota.

On the demand side, Chinese refineries that process DRC hydroxide into battery-grade material are already reporting feedstock deficits. Fastmarkets estimates the shortfall at 10,700 tonnes for 2026. That's not a rounding error in a market this small. It's the difference between running a refinery and idling one.

The market went from oversupplied to structurally gated in less than a year. Not because demand surged. Because one government closed a valve and the entire downstream chain seized.

Here's what nobody is modeling. The quota system has a use-it-or-lose-it structure. Q4 2025 allocations had to be fully shipped by April 30, 2026. Any unused quota evaporates. It doesn't roll forward. It doesn't get redistributed. It just disappears from the exportable supply.

Fastmarkets flagged a "massive discrepancy" between allocated quotas and actual export volumes earlier this year. Meaning: producers were given quota but couldn't physically ship the material in time. Logistics in the DRC are not a spreadsheet exercise. You're moving drums of hydroxide on unpaved roads through provinces where the security situation changes by the week. The quota might say 7,250 tonnes per month. The roads, the ports, and the trucking capacity say something else.

I've watched this exact setup before — not in cobalt, but in Indonesian nickel after the 2020 ore export ban. Government sets a hard cap. Market models the cap as a ceiling on supply. But the actual delivered volume undershoots the cap because the logistics between mine mouth and port can't keep pace. The effective supply is tighter than the quota implies. Every time.

And the financial layer? Still asleep. The "cobalt is dead" narrative is so deeply embedded that most commodity desks don't even have a cobalt analyst anymore. They folded it into the lithium team two years ago. I know because I've called three of them and gotten voicemail. The LME cobalt contract is thinly traded, physically settled, and invisible to the algo-driven funds that move everything else. There is no liquid futures market for short sellers to pile into, which means there's also no mechanism for a violent short squeeze. The pain here is quieter. It shows up as Chinese refinery margins compressing, then as cathode precursor price hikes, then — three months later — as an earnings miss at a battery cell maker nobody was watching.

The weak link isn't the DRC's willingness to enforce the quota. It's the market's refusal to believe it matters. Every month that passes with actual shipments undershooting the theoretical cap is another month the downstream inventory buffer thins. And cobalt doesn't have a deep buffer to begin with. This isn't copper with 270,000 tonnes sitting in LME sheds. The entire global cobalt market is roughly 310,000 tonnes. One country controls three-quarters of it. One regulator decides how much leaves.

The sequence from here is mechanical. It doesn't require a crisis. It just requires the quota to hold and the logistics to keep underperforming it.

The first place you'll see it is in Chinese refinery margins. They're already thinning. When hydroxide feedstock costs $53,800 per tonne and your contracted cathode precursor sale price was set when hydroxide was $30,000, the math breaks. Some refiners will eat the loss. Others will cut runs. Either way, less battery-grade cobalt sulphate enters the market.

The second-order effect hits the automakers. Every long-range EV platform using NMC 811 or NMC 622 cathodes needs cobalt. There is no near-term substitute for energy density at the top end. BMW, Mercedes, and the premium Chinese marques can't swap to LFP without redesigning the pack, the cooling system, and the vehicle floor. That's a three-year engineering cycle. The cobalt cost increase lands on their BOM in the next quarter.

Where does capital go? Not into the broad battery ETFs. Those are stuffed with lithium juniors, graphite explorers, and cell manufacturers who are on the wrong side of this input cost squeeze. The edge — if there is one — sits with the vertically integrated players who secured DRC offtake agreements before the quota hit. Companies whose cobalt supply is contractually locked, not spot-dependent. They're the ones whose margins widen as the market tightens.

There's also the secondary supply angle. Cobalt recycling from spent EV batteries is growing but still covers less than 10% of demand. The recycling thesis needs another five years of end-of-life battery volume before it changes the supply picture. Until then, primary DRC material is the market.

The consensus says cobalt is a sunset commodity. The DRC says it's a strategic asset with a hard export ceiling. When the consensus and the sovereign with the ore body disagree, I know which one I've learned to take seriously. It just tends to take the market longer than it should to figure that out.