You've probably heard of Nvidia, which just overtook Aramco to become the world's leading company. third largest company by market capitalization. But that high stock valuation speaks to a big problem: Demand for chips that can run power-hungry AI applications far outstrips supply. That's an especially big problem for the military, which is looking to run complex AI programs in environments where it won't be possible to call back to large groups of enterprise clouds due to electromagnetic interference, remoteness, etc.
DARPA is funding research into new chips to do the job. On Wednesday, the defense research agency announced an $18.6 million grant to EnCharge AI, a California company founded in 2022 by Princeton computer science professor Naveen Verma.
The grant is part of DARPA's $78 million Optimum Processing Technology Inside Memory Arrays (OPTIMA) program, which aims to develop new types of chips that can run AI applications using fewer computing resources, also saving power and size. .
Verma and his team have created a new type of chip for the type of AI called large-scale convolved neural networks. Current chips typically transport data back and forth between external memory and the Integrated memory located in the chip processor., so data is effectively stored in one place but processed elsewhere. That results in wasted energy and delays, like having to constantly consult a reference book for answers you can't memorize.
On the other hand, Verma's chip does not in-memory computing. His processor uses โmetal-oxide-metal (MOM) finger capacitors instead of current-domain calculations, based on bitcell transistor transfer functions,โ he wrote in a document 2019. This has increased power efficiency by a factor of 16 and performance by a factor of 5. That could allow much smaller devices, such as phones and laptops, to run much more complex AI programs without needing to call back. large groups of servers. "The future is to decentralize AI inference, freeing it from the data center and bringing it to phones, laptops, vehicles and factories," Verma said in a Press release.
Several other anonymous institutions also participate in the OPTIMA program.
