While the attention of the world is focused on AI technologies that create text, images, and movies, a startup founded by a former senior researcher from DeepMind is trailblazing the development of GenAI technology to drive innovation in physical materials.
Orbital Materials, led by Jonathan Godwin, a key contributor to DeepMind’s material research, is pioneering an AI-driven platform to explore a wide range of materials, from batteries to carbon dioxide-capturing cells.
Taking inspiration from AI systems like AlphaFold, renowned for predicting protein structures from amino acid sequences, Godwin established Orbital Materials with the vision of applying similar techniques to the field of materials science.
Traditionally, the discovery of new materials has relied on tedious trial and error methods in laboratories, often requiring years of experimentation. Recognizing the need for a new approach, Godwin advocates for the integration of AI experts with materials scientists to bridge the gap between virtual concepts and tangible applications.
Developing a new material, whether with AI assistance or not, is a complex process involving the identification of physical and chemical structures to achieve specific properties such as lightweightness and rigidity. Subsequent steps include determining the processes, like melting or evaporating, to create these structures reliably. The material then undergoes extensive testing under various conditions to ensure its suitability for intended applications.
While AI cannot solve all challenges in materials design, it can expedite the process and cut costs by using computational methods to predict properties and processes leading to desired materials.
Orbital Materials stands out with its custom AI model, Linus, tailored for materials science. This model, akin to AlphaFold’s success, drives the startup’s research in New Jersey. Scientists can input natural language instructions to prompt Linus to generate a 3D molecular structure meeting specific criteria, refining it iteratively for the desired outcome.
Despite its capabilities, Linus may occasionally produce impractical materials. Nevertheless, Orbital Materials has successfully developed a cost-effective, efficient carbon dioxide capture filter using this technology, with plans for further revelations soon.
Based in London with a 13-member team, Orbital aims to advance materials to proof of concept or pilot demonstration stages before partnering with external manufacturers. A recent Series A funding round of \\(16 million led by Radical Ventures and Toyota Ventures, totaling approximately \\\)21 million, will primarily support the expansion of Orbital’s data science and wet lab teams.
Drawing a parallel to AlphaFold’s impact on drug discovery, Godwin envisions Orbital Materials’ technology accelerating the design and commercialization of advanced materials at an unprecedented pace.
