US supercomputers train on billions of molecules to boost battery material discovery
AI foundation models, trained on supercomputers, are being used by University of Michigan researchers to predict molecular properties and suggest optimal battery materials, replacing traditional methods and accelerating next-generation battery development.