Tearing apart the nickel cost floor

Nickel.

The sell-side has spent six months calling it an oversupply story. LFP batteries are "killing nickel demand." Bloomberg runs the surplus chart. CNBC hosts nod along. The word "glut" appears in every other research note.

Meanwhile, the world's largest nickel mine is shut down. Not because of a strike. Not because of a geological failure. Because the Indonesian government slashed its mining permit by 70% and the operation ran out of legal tonnage by late May.

The International Nickel Study Group doesn't throw around deficit forecasts for fun. In April, they revised their 2026 outlook from a 261,000-tonne surplus to a 32,000-tonne deficit. If you've been trading metals long enough, you know what those INSG revisions look like in the rearview mirror. They're always late. They're always conservative.

The narrative says surplus. The numbers say the valve just closed.

Indonesia controls more than 50% of global nickel output. In February, the Energy Ministry set the 2026 RKAB mining quota at 260–270 million wet metric tonnes. The 2025 quota was 379 million. That's a 29–31% cut.

Indonesian smelters need 343–345 million tonnes of ore to keep their furnaces and autoclaves running. The quota gives them 270 million. Do the arithmetic.

PT Weda Bay Nickel — Eramet, Tsingshan, and Antam's joint operation, the largest nickel mine on the planet — received an initial allocation of 12 million tonnes for 2026. It produced 42 million in 2025. That's not a trim. That's amputation. By late May the quota was exhausted and the mine went into maintenance mode. Workforce reduced. Active extraction ceased.

Vale Indonesia suspended mining after delays in its own plan approval. Nickel Industries received 14.3 million wet metric tonnes, up from 9.0 million — one of the few operations that got more. The policy isn't cutting uniformly. It's reshaping which operators get ore access and which don't.

Then there's the chemistry problem. Indonesia's dominant processing route is High Pressure Acid Leach — HPAL. It consumes enormous quantities of sulphuric acid. Indonesia imports roughly 75–80% of its sulphur from the Middle East. Hormuz disruptions and a Chinese sulphuric acid export ban drove granular sulphur from under $600 per tonne to around $1,000. Some HPAL plants hold only one to two months of sulphur inventory. That's not a buffer. That's a countdown.

Ore grades in many producing areas have fallen below 1.5% nickel content. That's the threshold where acid consumption per unit of nickel recovered starts climbing on a fixed chemical curve that operational efficiency cannot fix. Lower grades, higher acid costs, tighter quotas. Every input to the HPAL cost equation moved in the wrong direction simultaneously.

The July 1–31 supplementary quota application window just closed. On July 10, Director General Tri Winarno stated explicitly: no comprehensive quota increase. Only targeted allocations for undersupplied smelters. The market spent June selling nickel on speculation of a quota raise to 360 million tonnes. The government said no. The deadline passed.

Here's the part nobody's pricing in.

LME nickel sits at roughly $16,965 per tonne this morning. Bernstein's July cost-curve update puts 75th-percentile C1 cash costs at $17,870. The 90th percentile is $18,650. The spot price is below the marginal cost of making the stuff.

I've seen this movie before. The market prices in a supply response that hasn't arrived. Traders read a rumor about quota easing, sell the metal, and move on to the next headline. Meanwhile, the actual cost curve says a quarter of global production is losing money at this price. That doesn't persist. Mines that lose money either shut down — deepening the deficit — or they get bailed out by higher prices. Either way, price goes up.

LME stocks sit at about 267,522 tonnes. Sounds comfortable if you don't run the consumption math. Global nickel demand runs roughly 3.3 million tonnes per year — about 9,000 tonnes a day. The entire LME stockpile covers roughly thirty days of global consumption. And a meaningful share of that on-warrant metal is Russian-origin, which much of the Western buyer base won't touch.

The LFP narrative is real but overstated. Yes, Chinese EV makers are shifting lower-end models to lithium iron phosphate chemistry, which contains no nickel. But NMC cathodes still dominate premium EVs, long-range vehicles, and aerospace and defense batteries. High-nickel NMC cathodes run 60–80% nickel content. The demand hasn't disappeared. It's bifurcated. And stainless steel — still the largest single source of nickel demand — grew in both 2025 and 2026.

The market is pricing in a quota increase that the Indonesian government explicitly denied three weeks ago. The application window just closed. The gate is shut.

The algorithms see price weakness and short more. They don't call Jakarta. They don't check sulphuric acid inventories at HPAL plants on Halmahera. They have never, to my knowledge, looked up the delivered cost of granular sulphur through the Strait of Hormuz.

If the quota holds — and three separate official statements say it will — the sequence is mechanical.

HPAL plants that can't source ore or afford acid start idling. Indonesia's MHP output — the battery-grade nickel intermediate the EV supply chain depends on — contracts. The 32,000-tonne deficit INSG projects becomes the floor, not the ceiling. That estimate doesn't account for the full extent of HPAL production cuts that haven't hit the data yet. It rarely does in the first pass.

Sulphide-process producers are the quiet beneficiaries. Their metallurgy generates leaching acid internally from the sulphur in the ore body itself. They don't import acid. They don't care about Hormuz. They don't need Indonesian mining permits. When the HPAL cost floor rises, sulphide operations become the low-cost marginal tonne — and they reprice accordingly.

I watched this exact mechanic play out in nickel during the 2022 LME squeeze — different trigger, same cost-curve physics. The HPAL complex was supposed to keep prices capped forever. Then input costs moved, and the "cheap nickel" thesis evaporated in about six weeks. Someone I know learned that lesson at about $40,000 a tonne. It stuck.

Where does capital go? Not into Indonesian NPI producers hedged at lower prices. Not into broad-base mining ETFs. The edge — if there is one — sits with sulphide-route producers holding unhedged production and non-Indonesian ore bodies. Companies whose cost structure is insulated from the exact variables now compressing margins across the HPAL complex: sulphuric acid, ore-grade degradation, and RKAB politics.

The market is trading nickel like the supply problem has been solved. The world's largest mine is in maintenance mode. The quota window just closed. The cost curve says a quarter of production is underwater. Markets can stay wrong about cost floors for a while. But geology and chemistry don't negotiate. The floor always reasserts itself. It's just a question of how much damage happens on the way down.