The Defense Department's largest research organization has joined forces with a Princeton-led initiative to develop advanced microchips for artificial intelligence (AI).

Source: Princeton

Professor Naveen Verma will lead the project, aiming to create AI chips that are more energy-efficient and suitable for diverse environments, from laptops to hospitals and even low Earth orbit. The Defense Advanced Research Projects Agency (DARPA) has granted an $18.6 million fund to support the project, breaking through key barriers in AI chip design.

The initiative, part of DARPA's OPTIMA program, seeks revolutionary advances in AI computing, and the funding will explore the chip's speed, compactness, and power efficiency. The Princeton-led project will collaborate with EnCharge AI, Verma's startup based in Santa Clara, Calif. The researchers have reimagined the physics of computing, introducing in-memory computing and analog computation to enhance efficiency and density.

The approach leverages the intrinsic physics of devices, particularly carefully designed capacitors, to achieve precise computation. The goal is to unlock AI from data centers, enabling its deployment in a wide range of applications.