A 50,000-tonne polyester recycling plant doesn't need 50,000 tonnes of mixed textiles. It needs 50,000 tonnes of polyester.

That distinction — between volume and the right volume — is the one the circular economy keeps tripping over. And the clearest example of what happens when it's ignored is a company called Renewcell.

In 2022, Renewcell opened the world's first industrial-scale textile-to-textile recycling plant in Sundsvall, Sweden. It had spent twelve years developing the technology. It had raised nearly $200 million. H&M had backed it for seven years. Levi's had launched a product with it. Inditex, Tommy Hilfiger, Calvin Klein, Ganni, and Reformation had all signed agreements to buy its recycled material.

The technology worked. The fibre quality was, by one account, "probably the best I've ever seen in the textile market."

By November 2023, the company recorded zero sales in a single month. It filed for bankruptcy in February 2024.

The company was Renewcell. And the lesson it left behind is one the industry still hasn't fully absorbed.

What went wrong wasn't the chemistry

Renewcell's Circulose technology dissolved cotton waste back to cellulose and reformed it into dissolving pulp — a direct input for viscose and lyocell production. The process was proven. The output met brand specifications. The plant in Sundsvall, Sweden, had capacity for 60,000 tonnes per year.

But Renewcell's former Chief Commercial Officer, Tricia Carey, told the story plainly after the collapse. In an interview with Fashion Dive in March 2024, she said: "Brands really wanted this. We couldn't supply enough from the pilot line, and fibre producers really wanted this, and they signed offtake agreements. And we kept producing because we were like, 'They want it, they want it.' And then it was sort of, 'Okay, they want it, but not as much as they said they wanted it.'"

Offtake agreements — contracts in which brands commit to purchasing the recycled output — didn't convert to actual orders. The product cost nearly twice as much as competing wood pulp. Fibre producers' prices for Renewcell-based viscose were 50% higher than regular viscose. When budgets tightened, brands defaulted to virgin material.

H&M's patience ran out. It terminated seven years of support. Ten days after Renewcell's bankruptcy, H&M co-launched Syre — a new recycled-polyester venture — with a $600 million purchasing agreement.

Brands that had publicly championed Renewcell walked away. Writing in the Stanford Social Innovation Review in July 2024, Carey and industry consultant Robert Antoshak reflected: "Brands could have extended a lifeline but chose not to do so. Brands failed to follow through on their public support materially."

The fine print nobody reads

Here is the part that rarely makes it into the headlines.

Renewcell's technology could process cotton — but only cotton with up to 5% non-cotton content. A T-shirt blended with 10% polyester was unsuitable. A pair of running shorts with elastane? Out. The vast majority of products sitting in people's wardrobes? Incompatible.

This isn't unique to Renewcell. Every recycling technology has a specification — and the specifications are narrow.

Infinited Fiber Company, the Finnish cellulose recycler, requires feedstock that is at least 88% cotton. Syre, the polyester recycler backed by H&M's $600 million, processes polyester only — it cannot accept cotton or blends. Circ, which has announced a €450 million plant in France, is one of the few technologies that can separate polycotton blends — but it is still scaling toward commercial production.

Think of it like this: every recycling plant is a specialist surgeon. One does knees. Another does shoulders. A third is pioneering a technique for the hip. Each needs a very specific patient profile. But the system delivering patients to the hospital is sending everyone through the same emergency room door — unsorted, undiagnosed, and in no particular order.

This is why incineration still dominates. Incineration doesn't care what fibre type, what blend ratio, what dye was used, or whether there's a zip sewn in. It takes everything, at a fraction of the cost, with zero quality requirement on the input. It's the path of least resistance — and it's not going away.

Recyclers are fighting on two fronts. Below them, incineration sets a price floor that any alternative has to beat. Above them, virgin material sets a ceiling that recycled output has to undercut. A specialist surgeon, competing against someone solving the problem with a baseball bat. The fight is not fair — and it won't become fair until the cost of getting the right feedstock to the right recycler comes down. That's not a technology problem. It's a collection problem.

Billions flowing downstream. Almost nothing flowing upstream.

The scale of investment in recycling technology is genuinely impressive. Syre is building a $1 billion plant in Vietnam, targeting 3 million tonnes of circular polyester by 2032. Circ's French facility will target 70,000 tonnes annually from 2028. Re&Up's plant in Turkey aims for 200,000 tonnes. Worn Again Technologies has just launched its accelerator plant in Switzerland. Industry projections suggest recycling technologies will process over 8 million tonnes of waste annually by 2030, compared to less than 1 million today.

Meanwhile, the brands are signing agreements: Syre with Gap, Nike, and H&M. Circulose (the restarted Renewcell) with 11 brand partners. Circ with Selenis, Arvind, Madewell, Reformation, and C&A.

This is progress. Genuine progress. The lesson from Renewcell — lock in demand before scaling production — has been learned. Recyclers are now securing brand commitments before they build.

But there is a second lesson from Renewcell that has not been learned: who is responsible for getting the right products, in the right condition, in the right fibre composition, to the recycler in the first place?

Every one of these offtake agreements guarantees demand for the output. None of them guarantees supply of the input.

The sorting problem is a collection problem in disguise

To address the feedstock gap, the industry is investing heavily in sorting technology. NIR spectroscopy systems that identify fibre composition in seconds. AI-powered computer vision. RFID readers. A single automated sorting facility costs $5–8 million and can process 100 tonnes a day, distinguishing between 90 different material types.

This is clever and necessary work. But it exists, in large part, because of a failure upstream.

When a product enters the waste stream through a street container or a charity bin, nobody knows what it is. Not the brand. Not the fibre composition. Not the condition. Not the original owner. It arrives as anonymous mixed waste — and then millions of dollars of technology are deployed to try to untangle it. It is, quite literally, trying to unsort a tangled ball of Christmas lights: painstaking, expensive, and only partially effective. NIR accuracy drops below 60% on black textiles. Blends of three or more fibre types remain difficult to classify. Wet items can't be scanned at all.

Now consider what happens when a product is collected at the point of return in a store. The customer identifies the brand. The system records the product category. The condition is assessed. The data exists before the product enters any waste stream. That's not a replacement for downstream sorting — but it's a pre-sort that dramatically reduces the complexity and cost of everything that follows. And it means products can be routed to the right recycler, not just the nearest one.

A 50,000-tonne polyester recycling plant doesn't need 50,000 tonnes of mixed textiles. It needs 50,000 tonnes of polyester. The more the collection system knows about what's being collected, the more efficiently the recycling system can operate. Without that, even the most advanced plant is dependent on whatever comes through the door — just as Renewcell was.

The brand's role — and the hidden cost inside every offtake agreement

Here is where the current model breaks down most clearly.

When a brand signs an offtake agreement to purchase recycled cotton or polyester at, say, $2,000 per tonne, that price doesn't just cover what happens inside the recycling plant. It covers everything upstream: the collection from containers and charity bins, the transport to sorting facilities, the expensive NIR and AI systems to identify fibre composition, the manual labour to remove zips and buttons, the rejection of items that don't meet the spec, and the disposal of what can't be used. All of that cost is baked into the price the brand pays per tonne.

In other words, brands are subsidising the full cost of an inefficient upstream system they currently have no influence over. The messier the collection, the higher the sorting cost. The higher the sorting cost, the higher the price per tonne of recycled fibre. The higher the price, the harder it is for recycled material to compete with virgin — which is exactly the dynamic that undermined Renewcell.

But if brands played a more active role upstream — helping customers return products through retail partners, capturing basic product data at the point of collection, routing items toward the right destination before they become anonymous mixed waste — the pre-sorting complexity drops. The cost per tonne drops with it. And the recycler gets cleaner, more consistent feedstock, which means higher yields and better economics.

This isn't about brands becoming waste managers. It's about recognising that the price they pay for recycled fibre includes a sorting tax that better collection would reduce.

Retailers are already aggregators. They just haven't been asked to work in reverse.

Paper and cardboard justify a bin on the street corner. The material is bulk, consistent, and uniform. Glass is the same. But clothing is fundamentally different — dozens of fibre types, hundreds of blends, every item unique. France has nearly 48,000 textile collection points, the most mature system in Europe, and collects 32% of what's placed on the market. The volume is improving. But those are generic collection points. Everything goes into the same container — cotton, polyester, nylon, elastane, blends of all three — and the problem gets kicked down the road to a sorting facility that now has to separate it all.

That's where progress stalls. The cost and complexity of sorting acts as a bottleneck between what gets collected and what actually reaches a recycler in usable form. We're not solving the problem. We're building a dam and wondering why the water can't flow where it needs to go. And the industry response, so far, has been to pour money into bigger, better dams — automated sorting facilities at $5–8 million each, with NIR scanners that still can't reliably read black textiles or blends of three or more fibres. If we wait for those systems to reach scale and accuracy, the new generation of recycling plants may die on the vine — starved of usable feedstock at volume, exactly the way Renewcell did.

Brands have a simpler option. Bypass the sorting problem entirely by working with their customers at the point of return. Retailers already solve this problem in one direction — receiving products from dozens of brands, across categories, and distributing them to customers. They are equally capable of operating in reverse. Not as recycling depots, but as smart collection points where a customer buying a new pair of shoes hands over an old pair, and the product enters the system identified rather than anonymous. And this isn't asking retailers or brands to become waste sorters. It's asking them to link the return of an old product to the sale of a new one — which is a marketing, engagement, and loyalty opportunity, not a cost centre.

The retailer knows it's an ASICS running shoe. The brand already knows what that shoe is made of. That connection — between what the retailer collects and what the brand knows — is the data layer that makes intelligent routing possible. No molecular scanning required. No waiting for sorting technology to catch up. Just the information that already exists, captured at the right moment. The alternative is another seven-year cycle of lost investment — and the industry has already seen what that looks like.

Will this time be different?

There are reasons for optimism. The new generation of recyclers has learned from Renewcell's mistakes. Circulose is reopening its Swedish plant only after securing long-term brand commitments. Syre and Circ are building demand-side partnerships before scaling production. EPR regulation is creating financial mechanisms that shift end-of-life costs to producers.

The destruction ban, effective July 2026 for large companies, is closing one escape route. But it applies to unsold stock — not post-consumer products. The volumes these recycling plants need to operate at scale won't come from excess inventory sitting in warehouses. They'll come from the millions of garments already in people's wardrobes. And there is no legislation on the horizon that solves that collection problem. Are we going to rely on processing unsold stock to keep billion-dollar facilities running at economy of scale? That's not a viable long-term model.

If the industry waits for regulation to catch up — for post-consumer collection mandates, for enforceable EPR targets, for Digital Product Passports — it could be waiting years. And in the meantime, the new generation of recycling plants faces the same starvation risk that killed Renewcell. Not because the technology doesn't work, but because the feedstock isn't arriving in the right form, at the right volume.

There is one question that remains unanswered — and it's the same question that was unanswered when Renewcell went under: who is building the collection infrastructure that delivers the right products, identified and categorised, to the right recycling facility?

The recycling plants are being built. The offtake agreements are being signed. The sorting technology is being funded. But the first kilometre — from the customer's wardrobe to the collection point — remains the least invested, least structured, and most important part of the entire system.

It's also the simplest. A person walks into a store to buy a new pair of running shoes. They have an old pair. If the store makes it easy to hand them over — and captures what's returned — the product starts its journey to the right destination with data attached. Not as anonymous waste, but as an identified product with a brand, a category, and a condition that determines where it should go next.

That's not a technology breakthrough. It's an availability breakthrough. And it's the one the industry hasn't made yet.

The recycling technology is scaling. The brand commitments are growing. The policy environment is tightening. But the system still breaks at the same point it always has: getting the right product from the customer to the right destination. Until collection is as invested in as recycling, the industry risks building billion-dollar plants with the same feedstock problem that bankrupted the last one.

Sources: Renewcell bankruptcy and timeline: Business of Fashion (Feb 2024), Trellis/GreenBiz (Mar 2024), Fashion Dive (Feb & Mar 2024), WWD (Feb 2024), Bloomberg (Apr 2024). Former CCO Tricia Carey account: Fashion Dive interview (Mar 2024), Stanford Social Innovation Review (Jul 2024). Peer-reviewed analysis: Corvellec & Stowell, Journal of Circular Economy. Cost analysis: Systemiq (Sophie Herrmann). IVL Swedish Environmental Research Institute: Sandin Albertsson et al., Sustainability (Feb 2025). Feedstock purity: Renewcell (max 5% non-cotton), Infinited Fiber (min 88% cotton), Syre (polyester only), Circ (polycotton separation). Siptex sorting requirements via Vogue Business. Investment figures: Syre ($1bn Vietnam plant), Circ (€450m France), Re&Up (200,000t Turkey), Worn Again Technologies (Switzerland accelerator). Brand offtake agreements: International Fiber Journal (2025), Business of Fashion (Mar 2026). NIR sorting limitations: Dutch Research Centre for Circular Textiles (<60% accuracy on black textiles), Fibersort (blend limitations). Sorting facility costs: Global Textile Times ($5-8m per facility). Industry projections: 8m tonnes by 2030 vs <1m today, Global Textile Times. EU destruction ban: European Commission, effective July 2026.