Cornell researchers have developed a computing device that stores information electrically but reads it through tiny mechanical motion, an approach that could open a path toward more energy-efficient hardware for AI and scientific computing.