ZHANGJIAGANG, China — In an industrial park in Zhangjiagang, a small metropolis on China’s east coast, a big buzzing and hissing machine feeds on piles of used garments and kinds them.
The novelty? It makes use of synthetic intelligence to kind them by composition at excessive velocity, providing a glimpse into how AI may play a job in decreasing the impression of artificial textile waste.
The Fastsort-Textile machine, named certainly one of Time journal’s Finest Innovations of 2025, was created by DataBeyond, a Chinese language AI recycling firm based in 2018.
“We are able to make full use of textile waste and cut back the quantity that’s incinerated which will probably be an important assist to recycling sources,” DataBeyond CEO Mo Zhuoya mentioned.
Artificial textiles are derived from fossil fuels and are a low-cost, common choice for trend manufacturing. Altogether they account for round 70% of worldwide textile manufacturing, in response to a report from Amsterdam-based nonprofit Circle Financial system, which analyzes methods to cut back textile waste.
Textile waste is a significant world pollutant, with China because the main contributor. China led world textile exports at $142 billion, greater than double that of the European Union, in response to the World Commerce Group’s 2025 Key Insights and Tendencies report.
Fastsort-Textile is getting used solely in a single location in China: Shanhesheng Environmental Expertise Ltd., a textile recycling facility in Zhangjiagang that put in the machine in 2025.
The gear makes use of an AI scanner to learn the composition of such textiles and kinds them by fibers, after which they are often recycled.
Fastsort-Textile kinds by way of 100 kilograms (220 kilos) of garments in two to a few minutes , in comparison with round 4 hours for one employee to do the identical factor. The machine can course of two tons per hour, whereas two individuals would wish two days and at diminished accuracy, in response to evaluation by Shanhesheng.
The AI scanner measuring 5-by-2 meters (16-by-6.5 ft) works with a sequence of conveyer belts. Staff load stacks of textiles onto belts that transfer them by way of the scanner, which emits a pointy hiss whereas studying the textiles’ composition. A stay video feed shows the studying on the scanner’s aspect.
It takes lower than one second to precisely learn one merchandise’s materials composition, which is ready in response to prospects’ desired benchmarks.
After the scanning course of, the textiles are transported to nylon and polyester sorting areas for recycling. Objects beneath the benchmark are sorted into a distinct space primarily for incineration or landfill, which is the place textile air pollution wreaks its most harm.
“This form of factor saves cash on labor prices, it saves time. When individuals kind supplies, they’ll’t inform precisely if it’s 80 or 90% polyester. This machine not often makes errors,” Shanhesheng Gross sales Supervisor Cui Peng mentioned.
Beforehand, as much as 50% of the processed textiles had been deemed unrecyclable and despatched to landfills or incinerated. That quantity is all the way down to 30% with the Fastsort-Textile machine, Gross sales Director Li Bin mentioned.
“Now, although machines are already able to sorting, individuals’s vitality is proscribed,” he mentioned. “Individuals can’t work for twenty-four hours straight, so robots might take over the roles ultimately. The final word purpose is a ‘darkish manufacturing unit’ with the robots operating 24 hours.”
