The Marginal Battery Has No Cobalt: Why Recycling Is Becoming a Compliance Business

Lithium iron phosphate passed half of the world’s electric vehicle battery deployment in 2025 — more than 55%, up from nearly 50% in 2024, according to the International Energy Agency’s Global EV Outlook 2026. The recycling industry’s revenue model was built on the two metals that chemistry does not contain.

Both statements are true at once, and together they explain why our own forecast — USD 21.69 billion in 2025 to USD 78.83 billion by 2033, a 17.5% CAGR that recomputes from those endpoints — should be read as a revenue forecast for a business that is changing customer, and not as a bullish call on metals recovery. The tonnage will arrive. The value per tonne is what is being repriced.

What a tonne of feedstock is worth now

Nickel-manganese-cobalt black mass is a nickel and cobalt concentrate with some lithium in it. LFP black mass is a lithium ore with a large quantity of iron and phosphate attached. That is not a detail of process chemistry. It is the entire revenue model.

A 2026 review in iScience put the global recycling efficiency for LFP batteries at approximately 10%, with 90% of recoverable value left unutilised, and identified lithium as the principal value target while noting that iron and phosphorus recovery is typically neglected — and that the phosphorus-bearing residue is a disposal liability if mishandled, with soil acidification and eutrophication risk. The same review notes that lithium extraction dominates the industrial hydrometallurgical route, and that the residue is what gets left behind.

January 2026 trade price assessments put cobalt metal near USD 56,400 a tonne and lithium hydroxide near USD 18,400 a tonne. Cobalt is elevated; lithium is depressed, albeit recovering. And the cobalt number is not a demand signal. China’s imports of cobalt hydrometallurgical intermediates were roughly 1,879 physical tonnes in December 2025, down 65% month on month and 96% year on year, with Democratic Republic of Congo origin material at 1,162 tonnes, down 98% year on year. Small-batch pilot export approvals were granted in the DRC in mid-January 2026, and with a shipping cycle of close to three months, imports were expected to stay suppressed into April.

So the metal that makes nickel-cobalt recycling look solvent is being priced by an export-quota regime, on exactly the chemistry whose share of new deployment is falling. That is the worst of both worlds for a plant whose flowsheet was designed around it: the feedstock is disappearing and its price is being set by policy rather than by demand.

Europe’s headline target measures mass, and mass is the wrong denominator

Regulation (EU) 2023/1542 sets the schedule everyone in this industry is building against. Recycling efficiency by the end of 2025: 65% for lithium-based batteries, 75% for lead-acid and 80% for nickel-cadmium, with higher targets from the end of 2030. Recovery of cobalt, copper, lead and nickel: 90% by the end of 2027 and 95% by the end of 2031. Lithium recovery: 50% by the end of 2027 and 80% by the end of 2031. Recycled content from 18 August 2031: minimum 16% cobalt, 85% lead, 6% lithium and 6% nickel in active materials, rising again from 18 August 2036. The methodology for calculating and verifying those rates was published as Delegated Regulation (EU) 2025/606 on 4 July 2025.

Here is the trap in that schedule, and it is not a small one. Sixty-five percent by mass is easy on LFP. Copper, aluminium, steel casing, graphite and the iron-phosphate fraction are most of the mass, and most of them are already separated by mechanical pre-treatment before anything touches a leach. A plant can clear the mass bar comfortably while discarding the lithium, and the 10% recovery figure above is the evidence that most of them do.

The obligation that actually bites is the 6% lithium recycled content from August 2031, supported by the 50%-by-2027 lithium recovery rate. Both are dated after the current global construction wave. The industry is building plants to a 2019 specification against a 2031 test — and unlike a metals price, an obligation does not move with the cycle.

Nor is it a tonnage obligation. Article 8 requires the percentage of recovered cobalt, lithium and nickel to be documented per battery model, per year, per manufacturing plant. That is an audit requirement attached to a product, not to a commodity. It is the reason a recycler that can assay and certify output is selling something a miner cannot.

The feedstock the regulation presumes does not exist yet

Our own report contains the number that reframes the whole thesis: end-of-life electric vehicle batteries are estimated at only about 15% of recycler input, with the balance still coming from production scrap and portable electronics. The retired-EV wave that the capacity was financed against has not arrived.

The price evidence from 2025 shows what that scarcity does. As capacity outran collection, recyclers moved from collecting gate fees for taking scrap to paying for it — and then the balance began to turn. Price-reporting coverage in August 2025 quoted a European recycler saying the market now required a gate fee again, citing badly underutilised capacity. Scrap scarcity is not the same thing as scrap value. When nobody can fill a line, the price of a tonne of feed rises until the subsidy or the obligation arrives to make the difference up.

Li-Cycle is the archetype of what happens when it does not. The company filed for bankruptcy protection in Canada under the CCAA and in the United States under Chapter 15 on 14 May 2025, with a USD 10.5 million debtor-in-possession facility and a stalking-horse credit bid of at least USD 40 million from Glencore, its largest secured creditor. Glencore’s own site now records that Li-Cycle was acquired as of 8 August 2025. Hydrometallurgical capacity, built ahead of feedstock, financed against 2022 metal prices, ends up owned by the trading house that can afford to wait for the 2031 obligation to arrive.

The arithmetic of one plant, and why the grant is the business model

Take a current, fully financed project rather than a failed one. In April 2026 Altilium was awarded GBP 18.5 million from the UK’s DRIVE35 Scale-Up Fund for ACT3, a commercial refinery at Plymouth designed to process 24,000 EV batteries a year and produce roughly 5,200 tonnes a year of nickel mixed hydroxide precipitate, 8,000 tonnes a year of lithium sulphate — equal to about 1,000 tonnes of lithium carbonate equivalent — and 5,400 tonnes a year of graphite. Construction was to begin in summer 2026, with commissioning at the end of 2027. A follow-on ACT4 plant in Teesside is scoped at 150,000 batteries a year and 30,000 tonnes of cathode active material.

The arithmetic is worth doing, because it explains the structure of the financing. One thousand tonnes of lithium carbonate equivalent spread across 24,000 batteries is 41.7 kg of LCE per vehicle. At January 2026 lithium hydroxide prices near USD 18,400 a tonne, and treating LCE as broadly comparable for order-of-magnitude purposes, that is roughly USD 766 of contained lithium per vehicle.

Set that against what it costs to collect a pack, discharge it, dismantle it safely, and run a hydrometallurgical line with the effluent, the reagents and the phosphorus residue that comes with LFP. USD 766 a car is not a business on its own. It is a business with a grant, a processing fee and a regulatory obligation attached — which is precisely what ACT3 is.

Note also what the output mix reveals. A UK plant betting on 5,200 tonnes of nickel MHP a year is assuming that nickel-cobalt feedstock stays in the mix long enough to pay for lithium-selective chemistry. That is a reasonable hedge. It is also an admission about where the margin currently sits: in the nickel, not in the lithium the regulation is about to mandate.

Europe’s feedstock is Chinese, which is the second problem

The IEA’s 2026 assessment adds a supply-chain fact that European recyclers will have to live with. LFP was more than 10% of EU EV battery demand in 2025, similar to 2024 — and nearly all of it was imported from China, roughly 30% as cells shipped directly and close to 70% embedded in vehicles. Europe’s future recycling feedstock is therefore Chinese-built, Chinese-documented cells, at a time when the EU’s own chain-of-custody rules are tightening.

The United States is heading the other way, and this matters for anyone modelling a global recycling market as one pool. American LFP share almost halved in 2025 from an already low base, reflecting higher tariffs on Chinese imports and sourcing requirements tied to the EV tax credit available until the third quarter of 2025. At the same time more than 50 GWh of US cell capacity — among them LG Energy Solution and Ford — was reallocated to LFP production, largely for stationary storage, which reached about one-third of US battery deployment in 2025.

Follow that through and the two regions are not solving the same problem. Europe’s recycling stream will be dominated by vehicles it did not manufacture. America’s will skew toward storage packs: fewer, larger, professionally decommissioned and tracked from installation. That is a logistics advantage and a documentation advantage at the same time, and it points to larger average feedstock value per collection event even as the chemistry value per tonne falls.

So what is actually being bought and sold

Strip the sector back and there are three products with durable pricing power, and one without.

Processing and tolling. Original equipment manufacturers and gigafactories with production scrap need the material back as specification-compliant intermediate, not as cash. A tolling fee is a service revenue line with a cost-plus structure, and it is not exposed to the cobalt price at all.

Certified recycled content. This is the one that compounds. From 18 August 2028 documentation on recovered content must accompany industrial, EV and SLI batteries containing cobalt, lead, lithium or nickel; from 18 August 2031 the minimum percentages become binding, per model, per year, per plant. Whoever can assay and evidence plant-level output is selling a product no mine can supply, priced against a compliance requirement rather than a metal index.

Battery-grade intermediates. Mixed hydroxide precipitate, lithium sulphate, and cathode active material precursor to whoever is onshoring cells and needs non-Chinese origin for sourcing rules. That is where the volume in the table above actually goes.

And what has no durable margin: shredding. Mechanical pre-treatment is becoming a logistics business with a gate fee, competing on collection density and trucking distance. The 2025 gate-fee reversal was not an anomaly. It was the market discovering what shredding is worth when capacity is abundant.

The marginal battery on the world’s roads is now LFP, and it has no cobalt. An industry built to recover cobalt is being rebuilt around a lithium-selective leach, a processing fee and a compliance certificate. That is a smaller revenue pool per tonne than the 2021 story implied, and a considerably more durable one — because the obligation, unlike the price, does not move.

Note on sources: market sizing is Pheonix Research’s own, from the Global Battery Recycling Market report. Chemistry shares are from the IEA’s Global EV Outlook 2026. Regulatory detail is from Regulation (EU) 2023/1542 and Delegated Regulation (EU) 2025/606. Corporate facts are from Altilium’s April 2026 funding announcement, Glencore’s own Li-Cycle disclosure and Reuters’ report of the May 2025 filing. Metal prices are January 2026 trade assessments and are quoted as such. No third-party market-sizing estimate is used anywhere in this piece.