The end of a battery's first life is a decision point rather than a destination. Four options sit in front of a retired pack, and the waste hierarchy ranks them: reuse in the same application, repurpose into a gentler one, remanufacture by replacing failed components, or recycle to recover materials.
Which one happens depends less on engineering than on diagnostics cost, disassembly labour and the price of a new cell.
What retirement actually means
Traction batteries are commonly considered retired at around 80 per cent capacity retention. That figure is a warranty convention, chosen because a vehicle losing a fifth of its range has lost a meaningful part of its usefulness. It is not a physical threshold, and nothing happens to the cells at that point.
A pack at 80 per cent may have years of service left in an application that cycles gently and does not care about volume or mass. It may also be close to a degradation knee, in which case its remaining life is short and unpredictable. Telling those two cases apart without expensive testing is the central problem of the second life industry.
This is why the state of health data requirements in the EU Battery Regulation matter commercially rather than only administratively. A documented degradation history changes the cost of assessing a pack from a laboratory exercise into a data lookup, which changes what a repurposer can afford to pay.
Why second life has been harder than expected
The concept is sound. The obstacles are practical and they compound.
Diagnostics cost. Establishing remaining useful life needs testing time on equipment, per pack or at best per module. When cells are cheap, that cost is a large fraction of the value of the asset being tested.
Disassembly labour. Packs are designed for assembly, structural integrity and crash safety, not for taking apart. Adhesives, welded joints and structural bonding all serve the vehicle and work against the repurposer. Automated disassembly exists in pilot form and struggles with the variety of designs in the field.
No standardisation. Module formats, voltages, communication protocols and mechanical interfaces differ by manufacturer and by model year. A second life integrator cannot build one product; they build one product per donor pack family.
Liability. Who is responsible when a repurposed pack fails is not always cleanly settled, and insurers price that uncertainty.
The falling floor. New cell prices have fallen substantially over the last decade. Second life competes against new cells, so every price reduction in new product compresses the margin available to a business built on avoided cost. Several second life ventures have been squeezed by exactly this.
Where it does work, the pattern is consistent: low cycle count duties, behind-the-meter applications, buffering for EV charging sites, telecom and backup power, and situations where the donor packs come from a captive fleet so the design variety problem disappears.
The scale ceiling has been rising. In June 2025 Redwood Materials, the largest lithium-ion battery recycler in North America, launched its Redwood Energy unit and deployed a 63 MWh array built from second-life EV packs with Crusoe at an AI data centre in Sparks, Nevada, the largest second-life deployment announced to that point. It fits the pattern exactly: a steady donor stream, stationary duty, and a recycler standing downstream for when the packs finally fade.
The recycling routes
Before any metallurgy, packs are discharged, dismantled at least to module level, and shredded, usually under inert atmosphere or in solution to manage the reactivity and the electrolyte. Casings, current collector foils and separator material are separated mechanically, leaving black mass: a powder containing the active materials, lithium, nickel, cobalt, manganese and graphite. Black mass is the traded intermediate, and it is increasingly regulated as a waste stream with restrictions on cross-border movement.
| Route | Method | Recovers | Loses | Position |
|---|---|---|---|---|
| Pyrometallurgy | High temperature smelting | Nickel, cobalt, copper as an alloy | Lithium to slag in many configurations, aluminium, graphite, electrolyte | Established, tolerant of mixed and dirty feed, energy intensive |
| Hydrometallurgy | Leaching with acid, then solvent extraction and precipitation | Lithium, nickel, cobalt, manganese at high purity | Graphite often, depending on configuration | Dominant in new capacity, needs cleaner and better sorted feed |
| Direct recycling | Recovering cathode material as a compound and relithiating it | The cathode structure itself, avoiding re-synthesis | Sensitive to chemistry mixing and contamination | Pilot and demonstration scale, best theoretical value, least proven |
Hydrometallurgy has taken the lead in new investment for one main reason: it recovers lithium, and lithium recovery is both economically and regulatorily necessary. It also produces battery-grade precursors directly rather than an intermediate alloy needing further refining.
Direct recycling is the most attractive on paper, because re-synthesising cathode material from recovered metals discards the energy and value embedded in the crystal structure. It is also the most demanding, since it needs feedstock sorted by chemistry and free of contamination. Mixed feedstock is the enemy of direct recycling, and mixed feedstock is what the waste stream naturally produces.
The feedstock reality nobody advertises
Most material entering recycling plants today is production scrap from cell factories, not end-of-life vehicle packs.
The reason is arithmetic. Electric vehicle sales at scale are recent, and vehicles last well over a decade, so the retirement wave has not arrived. McKinsey projected in March 2023 that production scrap would remain the primary recycling feedstock until around 2030, when end-of-life volumes overtake it, and Harper and colleagues calculated in Nature in 2019 that the one million electric vehicles sold in 2017 will on their own produce roughly 250,000 tonnes of pack waste when they retire. Meanwhile every new gigafactory generates scrap, and early-ramp yields make that a large stream. Production scrap is also excellent feedstock: chemistry is known, it is clean, it is unformed or lightly formed, and it arrives in predictable volumes under a commercial contract rather than through a collection system.
Two implications follow. Recycling capacity planning based on projected end-of-life volumes will run ahead of actual availability for some years. And as manufacturing yields improve, the scrap stream shrinks, which is good for the factories and awkward for recyclers who built capacity around it.
The 2025 shakeout made that concrete. Northvolt filed for bankruptcy in Sweden in March 2025, removing both a scrap source and its in-house Revolt recycling programme. Li-Cycle entered creditor protection in Canada in May 2025, and Glencore completed the purchase of its main assets in August 2025. Plants sized for a feedstock wave that was still years away ran out of money before it arrived.
Why LFP changes the economics
Recycling economics were built on nickel and cobalt. Remove both and the recoverable value falls sharply, while the process cost does not.
Lithium iron phosphate contains lithium, iron and phosphate. Iron and phosphate are cheap. So the value case rests on lithium recovery, on the cost of the process itself, and on whatever producers pay under extended producer responsibility obligations. When lithium prices are high, LFP recycling looks viable. When they fall, it depends on regulation to function.
Since LFP has taken the majority of stationary storage and a large share of mass-market vehicles, this is not an edge case. It is the coming mainstream of the waste stream, and it is why the regulatory framework does the heavy lifting rather than the metals market.
Direct recycling is particularly attractive for LFP for this reason: recovering the cathode material intact preserves value that leaching to base metals destroys. Whether it scales is one of the more consequential open questions in the sector.
What the regulation requires
Regulation (EU) 2023/1542 sets collection targets for portable batteries of 63 per cent by the end of 2027 and 73 per cent by the end of 2030, material recovery targets of 90 per cent for cobalt, copper, lead and nickel by the end of 2027 rising to 95 per cent by the end of 2031, and for lithium 50 per cent by the end of 2027 rising to 80 per cent by the end of 2031. From 18 August 2031, new batteries in scope must contain minimum recycled shares of 16 per cent cobalt, 85 per cent lead, 6 per cent lithium and 6 per cent nickel, with higher figures from 2036. Extended producer responsibility obligations and the battery passport, mandatory from 18 February 2027, sit on top.
The methodology for calculating and verifying recycling efficiency and material recovery was fixed by Commission Delegated Regulation (EU) 2025/606, published in July 2025. Other delegated acts are still arriving, so check the current implementation status before relying on any specific date.
The structural effect is worth understanding regardless of the dates. Recycled content requirements create demand for recycled material independent of its spot price against virgin material, which is precisely the mechanism that makes low-value chemistries recyclable. Producer responsibility puts the collection and processing cost on the party that chose the design, which over time pushes design for disassembly in a way that voluntary initiatives have not.
Informational and educational content only. Not professional, financial, legal, or engineering advice.