Xinterra's AI Materials Lab Turns T-Shirts Into Carbon Capture Devices
Source: a+ Singapore
Materials science has a speed problem: getting a new material from lab to market can take years of trial-and-error, which is exactly the wrong timeline for climate change. A Singapore startup spun out of A*STAR's Institute of Materials Research and Engineering thinks AI can compress that timeline — and its...

Materials science has a speed problem: getting a new material from lab to market can take years of trial-and-error, which is exactly the wrong timeline for climate change. A Singapore startup spun out of A*STAR's Institute of Materials Research and Engineering thinks AI can compress that timeline — and its first consumer-facing product is a T-shirt that captures carbon dioxide. Xinterra, founded during the pandemic when its scientists were locked out of their labs, now develops and commercialises sustainable materials at a pace traditional setups cannot match.
The flagship product is Cozterra, a textile treatment that captures CO2 on fabric, converts it into a harmless mineral, and washes away with regular detergent. The company's numbers are striking: in one day, 20 T-shirts treated with Cozterra absorb as much carbon dioxide as a mature tree. Early adopters include Singapore labels Zoda and Kemunna, plus Brazilian ethical fashion platform Malwee, whose Cozterra-treated T-shirt debuted at London Fashion Week. Xinterra's broader client base spans paint, pulp and paper, and commodity plastics — industries where hitting target performance at lower cost is the everyday challenge.
The founding story is quintessentially Singapore deep tech. Jatin Kumar, Xinterra's CTO, spent years as a scientist at A*STAR's IMRE working on polymers for applications from anti-terrorism to nanomedicine, and grew impatient with how long innovation took. The AI pivot was his escape hatch — using machine learning to leapfrog the experimental grind. When the pandemic shut the labs, he and his colleagues turned that frustration into a company built around AI-accelerated materials discovery.
The implications reach beyond fashion. Carbon capture has been criticised as costly and hard to scale; a distributed approach that turns everyday clothing into capture devices reframes the debate — Kumar's stated goal is to 'empower humans to become carbon capture agents,' weaning society off dependence on monolithic corporate or government action. With Cozterra expanding into polyester and three novel products in the pipeline, Xinterra is betting that AI-driven materials science becomes a core export of Singapore's deep tech ecosystem.
Why it matters for Singapore: Xinterra is a homegrown example of the AI-for-science direction Singapore is pushing — A*STAR research commercialised into a climate tech product with global reach, from a London Fashion Week debut to Brazilian retail. As Singapore positions itself as a hub for AI-driven deep tech, startups like this show the pipeline from public research to real-world product is working, and that AI's most valuable local application may be speeding up physical science, not just software.


