The spot price of the two monomers that nylon 6,6 is built from spiked as much as 150% this year, caught in petrochemical supply swings. Epoch Biodesign, a London startup six years into its work, says the number should not matter to anyone who buys its output.
Epoch Biodesign designs enzymes with AI that break nylon 6,6 back into its original monomers, adipic acid and hexamethylenediamine (HMDA), at ambient temperature and over 90% recovery.
A $12 million round backed by lululemon and a memorandum of understanding (MoU) with INVISTA, a world-leading nylon 6,6 producer, moved the company from pilot to a demonstration plant, with a 20,000-tonne monomer facility targeted for 2028.
DuPont chemists first made nylon 6,6 in the 1930s, and the material has been in continuous production since. It shows up in athletic wear, airbags, climbing rope, carpets, and engine covers.
Recycling it has never worked at scale. Heat breaks the polymer chain in the wrong places. Solvents that could work cost more than virgin resin. Most nylon 6,6 ends up buried or burned.
The chemistry that stopped recycling
Nylon 6,6 is tough precisely because of the bonds that hold it together. Amide groups along the polymer chain form hydrogen bonds with neighboring chains, packing the material into a crystalline structure that shrugs off heat and wear. Those same bonds defeat most recycling.
Mechanical recycling shreds the polymer, shortening the chains and lowering fiber quality. Each cycle downgrades the material until it becomes carpet backing or filler. Solvolysis, the chemical route, uses high heat and solvents to break the chains, but it is energy-intensive and leaves byproducts that limit what can be done with the output.
Epoch took a different path. Instead of running the full machinery of a living microbe, it deploys only the enzymes, sourced from industrial suppliers that already make them at scale. Each enzyme in the cascade attacks a specific bond in the nylon 6,6 chain. Working together, they progressively break the polymer down into its original building blocks: adipic acid and hexamethylenediamine (HMDA).
The process runs at ambient temperature and recovers both monomers at purity, according to the company. From those monomers, a maker can spin brand-new, virgin-quality polymer and yarn. The enzyme cascade is the whole difference. A fermentation route has to keep a microbe alive, feed it, and handle its metabolic waste. A pure-enzyme route is closer to a chemical plant: controlled, continuous, and easier to scale.
The AI part matters more than the enzyme part. Designing one enzyme that cuts a single specific bond in a synthetic polymer is painstaking chemistry. Designing a cascade that handles a whole waste stream is a search problem. Its platform uses AI to design enzymes that deconstruct materials at the molecular level, which is a different discipline from engineering a bacterial strain for fermentation.
The economics hinge on a simple trade. Nylon 6,6 resists breakdown because its monomers are locked into a tight crystal lattice. Crack that lattice efficiently and the monomers come out whole, worth more than the waste you started with. Miss by a single bond and you get degraded fragments that sell for less than the energy you spent. Enzyme selectivity is the entire margin, which is why the company's pitch rests on AI rather than on brute-force chemistry.
The material itself is worth the effort. Nylon 6,6 has a higher melting point, more rigidity, and better heat resistance than nylon 6, the cheaper sibling used in most commodity textiles. Those properties make it the default choice for airbags, engine components, and premium sportswear. It is also exactly what makes the waste hard to process: the same engineering that gives the polymer its performance gives it its resistance to recycling.
As we wrote in July, AI-designed enzymes have started reaching industrial scale elsewhere. Scindo raised £4 million to bring enzyme production to market, a sign that the field is moving past the lab phase. Epoch's bet is that the same toolset can crack a polymer as stubborn as nylon 6,6.
Mixed waste is the actual problem
Nylon 6,6 is hard to recycle on its own. Most discarded textiles make it harder. A garment that is 60% nylon and 40% elastane is unrecyclable through standard channels, because no process separates the two chemistries cleanly. Coated fibers, laminated fabrics, and compounded automotive plastics are worse.
Its pitch is that the enzymes handle the entire bale. The cascade processes blended textiles, multi-layer laminates, coated fibers, and mixed automotive plastics, sorting the chemistry at the molecular level rather than sorting the waste first. Conventional mechanical and chemical recycling cannot address those streams at all.
The difference matters in the real supply chain. Collection systems sort textiles by rough category, not by exact polymer blend. A sorting line that has to separate every 60/40 nylon-elastane mix would be impossibly expensive, which is exactly why most blended garments are currently unrecyclable. The claim is that the enzyme cascade removes the need for that fine-grained sorting.
The feedstock pool is large precisely because the sorting problem is unsolved. Most of the textile waste in the world is blended. A process that accepts the whole bale unlocks a waste stream measured in millions of tonnes, while a process that demands pure nylon 6,6 input fights over a tiny, already-separated fraction. The economic argument is inseparable from the technical one.
That is where the buyers come from. lululemon's Align and Wunder Train leggings use nylon, and the company backed the latest round. Inditex, Zara's parent, participated in the earlier Series A. Leitmotif, a deep-tech investor with automotive ties, joined the 2026 round. Engine covers, airbags, and interior trim are the automotive end of the same polymer.
The partners are not charity. Brands need recycled content to meet their own sustainability commitments, and nylon is one of the hardest materials to source recycled. A drop-in polymer that does not force a supplier change is worth a great deal to a supply-chain team. Automotive OEMs face the same squeeze from EU end-of-life vehicle rules that push for more recycled material in new cars.
From pilot to demonstration plant
The company's stated path runs through two physical steps before the 2028 plant. The first is a demonstration site near Imperial College London, sized to validate commercial-scale output rather than laboratory yields. The second, which it says is nearing completion, is its largest nylon 6,6 biorecycling facility to date. Both sit between the multi-tonne pilot and the 20,000-tonne commercial target.
That staged approach is the standard shape of enzyme-process scale-up, and the standard failure point is the middle stage. A cascade that recovers 90% of monomers in a lab beaker can fall to 60% when the input is a real bale of mixed post-consumer textiles, because real waste is dirty, damp, and inconsistent. The demonstration facility exists to find that number before the company commits the larger capex.
Enzyme production itself is a second scaling problem. The company sources enzymes from industrial suppliers rather than growing its own microbes, which removes one variable. But the volume of enzyme needed for 20,000 tonnes of monomer is not what today's suppliers ship in a year. Either it grows into a meaningful buyer of enzymes or it brings production in-house, and both paths change the cost structure of the business.
The money that paid for scale
Epoch closed a $12 million strategic round on March 25, 2026. Participants included lululemon, KOMPAS VC, Happiness Capital, Extantia, and Leitmotif. Total funding now exceeds $50 million, following an $11 million seed in 2022 led by Lowercarbon Capital and a Series A in early 2025 led by Extantia.
The money is being spent on a specific step: moving from a pilot facility to a demonstration site near Imperial College London, built to validate commercial-scale output. The company says its second and largest nylon 6,6 biorecycling facility is nearing completion. A full production plant is targeted for 2028, designed to make 20,000 metric tonnes of monomer per year.
How the funding came together
2025 · Series A led by Extantia, with Inditex among the participants
2026 · $12 million strategic round with lululemon, KOMPAS VC, Happiness Capital, Extantia, and Leitmotif
The economics only started to look right recently. Because the company starts from waste textiles instead of fossil feedstocks, its costs are insulated from the swings that hit virgin chemical markets. When the spot price of the two monomers rose as much as 150% this year, that insulation became a selling point rather than a theory. Nathan told reporters the volatility made the pitch considerably easier to make in 2026 than it was in 2022.
"Our AI-engineered enzymes take these materials apart at the molecular level, producing virgin-quality chemical building blocks without the price volatility and environmental impact of fossil-derived products."— Jacob Nathan, Founder and CEO, Epoch Biodesign
The investor mix is itself the analysis. Fashion brands put in money because they want the output; climate funds because they want the carbon math; automotive investors because they want the feedstocks. Three different buyers paying for the same demonstration facility is exactly the signal a scale-up story needs.
Scale is the whole game. Multi-tonne output proves the chemistry. Multi-kilotonne output proves the business.
What could break the model
The biggest risk is the gap between the demo and the plant. Going from a demonstration facility to a 20,000-tonne commercial plant is not a linear step. Enzyme supply has to scale, process engineering has to hold at industrial volumes, and feedstock has to arrive reliably, sorted, and cheap.
Virgin-quality is a strong claim and a hard one to prove. That is what the INVISTA relationship is really testing. Under the memorandum of understanding signed in February, technical assessment and polymer qualification are underway, with application performance testing planned as the next phase. Until those tests pass, virgin-quality is an assertion, not a specification.
Feedstock is the quieter risk. The technology's advantage is mixed waste, but mixed waste is also hard to collect and inconsistent. A bale of blended textiles varies by season, region, and brand. The chemistry must absorb that variance, or the process economics shift with every shipment.
There is also the question of who pays for collection. Recycling economics depend on the waste being cheaper than the feedstock it replaces. If textile collection and sorting infrastructure costs keep climbing, the input price rises, and the advantage over virgin nylon narrows. Its own position is that the monomer price volatility of 2026 has widened the gap in its favor. That spread can close just as fast as it opened.
The biology is the least likely failure point. That is worth saying plainly: after six years, the enzymes are the proven part. The unproven part is everything around them.
The counter-argument
Enzymatic recycling is not the only route, and it has not won yet.
Mechanical recycling already operates at scale, and it is cheap. Its problem is degradation, but for many applications a downgraded fiber is good enough. Solvolysis is further along industrially for some polymers, and chemical companies have been funding it for a decade.
Biological recycling also carries a long record of overpromising. The field has produced years of press releases and few commercial plants. Its answer to that history is the investor list and an MoU with a company whose heritage traces to the invention of nylon 6,6.
The three routes differ in what they preserve and what they cost:
| Route | Output quality | Energy demand | Mixed-waste handling |
|---|---|---|---|
| Mechanical | ◐ Downgraded fiber | ✔ Low | ✗ Sorting required |
| Solvolysis | ✔ High purity | ✗ High heat | ◐ Partial |
| Enzymatic | ✔ Virgin-quality | ✔ Ambient | ✔ Whole bale |
There is also the price question in reverse. If virgin nylon stays cheap and precursor volatility eases, the incentive to pay a premium for waste-based monomer shrinks. Its economics depend on that spread staying wide. Nothing guarantees it will.
Watch the 2028 target the way you would watch any capital project. Enzyme plants have a habit of slipping by quarters, not months, and the company is running on a $50 million war chest against a scale-up that, by its own description, needs a second facility and then a third. The funding table above covers the early stages, not the endgame.
The strongest counter-argument is the record of the category itself. Enzyme recycling of PET, the plastic in bottles, spent more than a decade moving from lab to plant, and even the commercial successes required a partner willing to absorb years of capex. Nylon 6,6 is a harder polymer than PET. The same patience, the same capital, and the same skepticism are warranted here.
Key signals to track
INVISTA qualification milestones: when polymer qualification passes and application testing begins, the virgin-quality claim becomes a specification.
The 2028 plant: whether the 20,000-tonne monomer facility hits schedule and how much of its output is contracted.
Brand offtake: whether lululemon and Inditex convert their investments into purchase agreements.
The monomer price spread: how long the precursor premium holds, because that spread pays for waste feedstock.
Nylon 6,6 is a material the world will keep using. The question is whether it can be made, and remade, without a fossil feedstock at the start of the chain. The company has the enzymes, the money, and one of the industry's largest producers watching closely. The demonstration plant will answer the rest.
