As leaders gather for COP31 in Türkiye, Nvidia said Wednesday that startups in its Inception program are using vision AI and robotics to sort mixed waste that would otherwise go to landfills.

The company said less than 7% of materials entering the global economy come from recycled or reused sources, and that municipal solid waste accounts for more than 2 billion tons of trash a year, a figure that could nearly double by 2050. “A more circular economy relies on making better use of what we already have,” Josh Parker, head of sustainability at Nvidia, said in the post. “AI can help us understand what’s in a waste stream, identify what can be recovered and automate the work of sorting it.”

Taiwan’s AIRECO uses Nvidia Jetson Orin NX modules on a Tainan pilot line, including robotic arms that pick targets off a conveyor; the company said recognition accuracy on the test line tops 95% and that a full line planned for next year aims to raise annual capacity from 18,000 to 48,000 tons while cutting operating costs by 85%, according to Nvidia. EverestLabs AI, a U.S. startup, has about 150 computer-vision robots in North America and Australia and estimates its systems have identified about 40 billion objects, recycled roughly 25,000 tons of material and avoided about 30,000 metric tons of carbon dioxide emissions, the post said.

U.K. startup Greyparrot’s Analyzer, mounted above belts moving up to 4 meters a second, has more than 300 systems in over 20 countries and has made over a trillion bounding-box detections, Nvidia said, reducing the time needed to check 15 tons of waste from 375 hours to six. Facilities using the data report efficiency gains of 10% to 30%, and at packaging company Amcor the tool helped raise food-grade plastic yield by 12%, the post said. Berlin-based reverse.fashion sorts used textiles at 3 meters a second with more than a dozen TensorRT-optimized models on Jetson modules; cofounder Karsten Pufahl estimated the system sorts in an hour what previously took a week. Those performance figures are the companies’ claims as reported by Nvidia.