Key takeaways
- An estimated 7 million tons of plastic film and flexible packaging winds up in landfills because recycling systems haven’t been built to handle it.
- Training AI to improve sorting will increase the volume of plastic feedstock that consumer products brands can reuse in packaging.
- The technology requires recyclers, haulers and packaging producers to rethink existing collection, sorting and recycling processes.
Google parent Alphabet’s research division X is collaborating with Dow Chemical to address a thorny recycling challenge: identifying films and other flexible plastics and separating them from landfill-bound waste streams so they can be recycled and reused.
Their approach was announced April 30 at Trellis Group’s Circularity conference. It will use real-world materials data from Dow’s plastics recycling ventures to improve an artificial intelligence-powered system for detection that was developed by X’s moonshot for circularity.
The goal is to encourage higher collection and recycling rates for films and flexible plastics, so they can be turned into new types of packaging used by consumer products companies, according to X and Dow executives.
“We’re working toward more precise recycling to restore the value,” said Rey Banatao, director and project lead for X’s moonshot for circularity. The mission: “Let’s demonstrate you can make a high-performance material again.”
An estimated 7 million tons of multi-material, film or flexible packaging is tossed annually in the U.S. alone, but the current recycling rate is less than 5 percent because most curbside recycling programs can’t handle it, Banatao said. One big sticking point is the inability of recycling equipment to recognize and separate clear films from other materials on conveyor belts. There is also uncertainty over the most effective options for collection.
X is building a database using generative AI and sensors to identify thousands of pieces of plastic at the molecular level every minute — including an additive often used in films and flexible packaging that can’t be easily detected.
The venture is running a pilot at an Oregon recycling plant, where sensors scan pieces of plastic, identify their makeup at a molecular level and sort the materials into specific categories for processing. Dow’s data will expand the materials that can be detected.
X and Dow created a proof-of-concept at Dow’s packaging lab (Dow Pack Studios) that recognizes multi-layered packaging made of plastics, paper and metals such as bags used to keep granola, potato chips or pet food fresh. The next phase of work will take place at recyclers, collection and sortation facilities and other locations where the approach can be put to the test in real-world situations, said Jill Martin, global sustainability fellow at Dow.
“We’re very focused on the film and flexible plastics unlock, and how we make these streams more recyclable,” Martin said. “We understand the materials we want to have.”
Partnership goal: boost plastic recycling rates
One of Dow’s corporate sustainability goals is to recycle 3 million tons of plastics waste annually by 2030. That requires increasing the sorts of plastic that can be processed.
It bought plastics recycler Circular in 2024 for an undisclosed sum to support that commitment. Circulus has two U.S. facilities with expertise in film recycling — in Ardmore, Oklahoma, and Arab, Alabama — which currently can handle 50,000 metric tons a year.
Dow’s work with X will benefit companies across the system, informing new processes for baling, collection, sortation and recycling. It will also help packaging design decisions of consumer brands, by suggesting materials that can be more easily sourced or recycled.
“The better we can understand, the better we can understand the customer applications,” said Haley Lowry, global sustainability director at Dow.
Dow and X declined to provide details about where their first pilot will take place, in part because they want more companies to offer funding and feedback for the experiments. “Brands need to share their voice on what their needs are, too,” Lowry said.
For example, Dow is collaborating with consumer products company Procter & Gamble on technology to convert the sorts of plastics that can be identified by the X project into recycled polyethylene with “near-virgin quality.” P&G plans to use this recovered material in its packaging. It’s unclear whether P&G will actually be involved with the new project, but the work is likewise focused on bringing down technical barriers to using more post-consumer resin.
AI and recycling
X’s moonshot for circularity is one of several ventures pitching AI-powered identification systems as essential for boosting plastics recycling rates.
Another is Amazon-backed startup Glacier, which disclosed $16 million in Series A backing April 28. The round was led by Ecosystem Integrity Fund; Amazon’s Climate Pledge Fund was among the return investors.
Glacier is building AI-enabled recycling robots that are made of readily available components, making them more cost-effective and smaller than existing options. The technology can detect more than 30 types of materials including polyethylene terephthalate (PET) plastic, aluminum cans, toothpaste tubes and cat food tins. It works at a sorting rate of up to 45 items per minute. The robots have been deployed at facilities in Chicago, Detroit, Los Angeles, Phoenix, San Francisco and Seattle.
“After seeing Glacier’s technology in action at our other facilities, it became clear they offered a faster, safer and more accurate way to recover valuable materials back into the circular supply chain,” said Sal Coniglio, CEO of waste management company Recology, which has installed the robots in multiple locations.
