MIT's chip stacking breakthrough could cut energy use in power-hungry AI processes

MIT engineers developed a "memory transistor" that stacks logic and memory components directly on top of each other, potentially slashing energy consumption in AI chips. The nanoscale device combines computation and data storage in a single unit, eliminating energy-wasting data transfers between separate components that currently account for most AI energy usage. Built with indium oxide and ferroelectric materials, the transistor operates 10 times faster and uses half the voltage of traditional designs, offering hope for more sustainable AI processing.