AI material discovery startup CuspAI reportedly raising $400M round

CuspAI Ltd., a startup working to speed up material discovery, is reportedly in the process of raising a $400 million funding round.

The Financial Times reported today that the term sheet has been signed but the transaction is still being finalized. According to the paper, Amazon.com Inc. founder Jeff Bezos’ Bezos Expeditions and Kleiner Perkins are among the participants. The investment will reportedly value CuspAI at $2.6 billion, more than five times what it was worth after its last funding round.

UK-based CuspAI is led by Chief Executive Officer Chad Edwards (pictured, right) and Chief Technology Officer Max Welling (left). Edwards previously co-founded a quantum processor startup called Cambridge Quantum. Welling is an AI professor at the University of Amsterdam. CuspAI’s advisers include Nobel Prize laureate Geoffrey Hinton and Yann LeCun, two of the world’s most foremost AI researchers.

Hardware projects sometimes require scientists to develop new materials. For example, a chipmaker working on a new interconnect might need a conductor that withstands heat better than the existing options on the market. Developing a material that meets a project’s requirements can take years.

CuspAI is working to speed up the process. According to the Financial Times, the company offers an artificial intelligence platform that can generate material designs and test them in simulations. Its initial customer roster includes ASML Holdings NV, Meta Platforms Inc. and Hyundai Motor Co.

In April, CuspAI open-sourced a material simulation toolkit called kUPS. The company says that the software can speed up some research tasks by a factor of 49.

Simulations enable researchers to check whether a material design meets a project’s requirements faster than real-world testing. Under the hood, the algorithms that power a simulation vary based on project goals. Testing how well an optical cable transmits light requires a different algorithm than a simulation that evaluates a car part’s durability.

Historically, combining multiple simulation algorithms in the same material evaluation was technically challenging. CuspAI says that its kUPS toolkit eases the workflow. It also speeds up certain related workflows such as debugging faulty evaluations.

Simulations often model chemical phenomena at different scales. For example, an evaluation of a material’s conductivity might produce data about particle interactions at both the atomic and molecular levels. According to CuspAI, kUPS stores such multi-resolution measurements in a data format optimized for graphics processing units. It says that the format enables GPUs to run multi-resolution simulations more efficiently than other technologies.

The AI ecosystem’s largest players are also working to streamline material research projects. Their work could create more competition for CuspAI in the long term.

Earlier this month, Anthropic PBC revealed that its researchers are exploring ways of using Claude to analyze data from NMR, or nuclear magnetic resonance, machines. Chemists use such systems to study the properties of new materials. An NMR machine works by directing electromagnetic radiation at a rotating material sample and measuring how its magnetic field changes. Those changes can reveal information about the material’s composition and structure.

Historically, turning raw NMR material measurements into a form that lends itself to analysis was a mostly manual process. Anthropic says Claude can automate the task across some chemistry projects. Furthermore, the company plans to apply the model series to additional chemistry use cases.

Photo: CuspAI