BI Foresight Briefings: from lab to fab. How AI is closing the gap between materials discovery and manufacture

By Sean Hargrave 28 Aug 26 Reading time: 2 mins

Google DeepMind’s GNoME model predicted 2.2 million new crystal structures in a single research programme. Two AI-materials start-ups have raised close to $750m between them in the past 12 months, one of them tripling its valuation to $2.6bn in July alone. AI is promising to revolutionise materials discovery and testing. Is it now just a case of when, not if?

Bassam El Said, who works on AI-assisted composite design at the Bristol Composites Institute, says AI now does in hours what used to take his research teams months. But he adds it will be, “another decade yet” before those discoveries reach an aircraft or a satellite. So, discovery has speeded up – but is industry not ready?

At the National Composites Centre (NCC), engineers have trained an AI model to monitor liquid resin infusion in real time, catching defects before they become critical. Materials Nexus, a London start-up, used machine learning to design a magnetic alloy with no rare earths in it, then had the sample independently validated at the Henry Royce Institute’s Sheffield facility. These are operational gains on the shop floor and in the lab, but new materials are not yet reaching production lines.

Dan Griffin, who leads the NCC’s resin infusion work, is careful about what has actually been proven. “We’ve demonstrated it on live infusions in controlled conditions,” he says, “and now we want to test it in an industrial setting.”

Controlled conditions and an industrial setting are not the same – which raises the obvious question of whether AI is actually shortening the road from discovery to deployment, or just making the first mile of a much longer journey faster?

The venture capital view and the regulatory view do not fully agree on where the real constraint sits, while Professor David Knowles, CEO of the Henry Royce Institute, is open about the UK’s current position in materials discovery but also its potential.

This BI Foresight briefing draws on interviews with:

• Bassam El Said, senior lecturer in Digital Design and Manufacture of Composites at the Bristol Composites Institute
• Dale Wyllie, senior payload systems engineer at the UK Space Agency
• Pooja Narayan, fast track lead for artificial intelligence at Airbus
• Dan Griffin, principal research engineer for automation and digital systems at the NCC
• Nina Gryf, senior policy manager at Make UK
• Professor David Knowles, CEO of the Henry Royce Institute
• Lawrence Lundy-Bryan, partner at Cloudberry VC
• Stephen Price, investment partner at the Clean Growth Fund
• Professor Martin Kuball, head of the Centre for Device Thermography and Reliability at the University of Bristol
• Emre Ozer, senior director of processor development at Pragmatic Semiconductor.

Read the briefing: From discovery to deployment: how AI is closing the gap between materials science and manufacturing

Sean Hargrave / Guest writer

Sean Hargrave is the former Innovation Editor of The Sunday Times. He has extensive experience freelancing on business and innovation topics for The Guardian, The Times, The Telegraph and Wired. After moving to the Oxford area he has extended innovation freelancing to helping the University of Oxford write about spinout companies as well as aiding Advanced Oxford research innovation opportunities for local and national policy makers. He also helps technology and digital marketing companies position themselves through white papers and thought leadership articles.