Nvidia recently surpassed Aramco to become the world’s third-largest company by market capitalization, highlighting the high demand for chips capable of running energy-intensive AI applications. This demand poses challenges for the military, necessitating the development of new chips for complex AI programs in remote environments. DARPA has awarded an $18.6 million grant to EnCharge AI, founded by Princeton computer science professor Naveen Verma in 2022, as part of the $78 million OPTIMA program.
Verma’s team has invented a new chip for large-scale convoluted neural networks that utilizes in-memory computing, leading to increased energy efficiency and throughput. This innovative approach could allow smaller devices like phones and laptops to run more complex AI programs without reliance on large server clusters. The future of AI inference involves decentralization, as stated by Verma in a press release. Other institutions are also involved in the OPTIMA program.
