An AI-powered sorting system in eastern China is separating garments faster than human workers, offering a scalable path forward for one of fashion's dirtiest logistics problems.
A textile recycling facility in eastern China has deployed an artificial intelligence system that sorts donated and discarded clothing at speeds no human workforce can match. The machine relies on computer vision and robotic automation to identify fabric types, colors, and garment conditions in real time, then routes each piece to the appropriate recycling stream. It is a practical solution to a problem that has long bottlenecked circular fashion efforts: the sheer volume of mixed textile waste arriving at processing centers makes manual sorting slow, expensive, and prone to error.
The fashion industry produces an estimated 92 million tonnes of textile waste every year, according to the United Nations Environment Programme. A significant portion ends up in landfills or incinerators because the sorting infrastructure cannot keep up with the inflow. China, as the world's largest textile producer and a massive consumer market, sits at the center of this challenge. Facilities there receive enormous daily volumes of used clothing from both domestic donation networks and international export streams. Human sorters, working under tight deadlines, typically manage to process a limited number of garments per minute while making frequent misclassification errors that contaminate recycling batches.
This is where the AI system changes the economics. By automating the identification and separation process, the machine reduces labor costs, increases throughput, and improves the purity of sorted material batches. Cleaner sorting means downstream recyclers can more effectively apply fiber-to-fiber recycling techniques, turning old garments back into raw materials suitable for new clothing production rather than downcycling them into lower-value products like insulation or rags.
Most consumers assume recycling clothing works the same way as recycling paper or glass. It does not. A single donated bag often contains a chaotic mix of cotton, polyester, nylon, elastane blends, and zippers or buttons made from metal and plastic. Each material requires a different recycling process. Fiber-to-fiber recycling, the most environmentally valuable approach, demands inputs sorted with a high degree of accuracy. Even small amounts of contamination from a wrong fabric type can ruin an entire batch.
Manual sorting has historically been the only viable method, and it is inherently limited. Workers must feel, stretch, and visually inspect each garment, sometimes checking care labels that may be missing or illegible. Speed suffers. Accuracy degrades over long shifts. The result is an industry bottleneck that has kept textile recycling rates dismally low globally, with the Ellen MacArthur Foundation estimating that less than 1% of material used to produce clothing is recycled into new clothing.
AI-powered sorting directly addresses this constraint. Computer vision systems trained on thousands of fabric samples can identify material composition in fractions of a second. Robotic arms or air-jet systems then physically separate the garments into designated streams. The machines do not fatigue, and their accuracy remains consistent across thousands of items per hour.
Market Implications for Startups and Investors
The deployment in China signals a broader trend. Venture capital has been flowing into textile recycling technology across Europe and North America, with companies like Tomra, Circ, and Worn Again Technologies developing their own approaches to automated sorting and chemical recycling. But China's combination of massive textile waste volumes, advanced manufacturing infrastructure, and government pressure to improve environmental standards creates a uniquely favorable environment for scaling these systems quickly.
For startups operating in this space, the opportunity is substantial. The global textile recycling market is projected to reach $5.6 billion by 2028, driven by tightening regulations in the European Union and growing corporate sustainability commitments from major fashion brands. Any company that can demonstrate reliable, high-speed sorting at commercial scale will find willing buyers among recycling facility operators looking to reduce costs and meet rising demand for recycled fibers.
The broader implication is that AI and robotics are finally reaching a maturity level where they can handle complex, high-variability physical tasks that were previously uneconomical to automate. Textile sorting is a particularly demanding use case because of the diversity of materials and conditions involved. If AI systems can solve this problem effectively, similar approaches will likely follow in other waste streams, from electronics dismantling to construction debris sorting.
As ABC News recently reported, the Chinese facility is already demonstrating that AI sorting can outperform human workers on both speed and consistency. The question now is how quickly this technology can be replicated across the thousands of textile recycling operations worldwide that still rely on manual processes. The infrastructure for circular fashion will not scale without it.
