Nanoprotonic Devices for >240x Performance Analog AI Hardware

Default ARPA-E Project Image


Program:
Exploratory Topics
Award:
$500,000
Location:
Boston, Massachusetts
Status:
ACTIVE
Project Term:
09/18/2023 - 09/18/2025
Website:
TBD

Technology Description:

Eva will develop novel devices to build analog processors that could drastically improve the energy efficiency of training complex artificial intelligence (AI) models. Eva’s proposed technology—a new class of nanoprotonic programmable resistors—would reduce the programming voltage of the devices for integration compatibility with standard circuit drivers and overhaul the device structures with an encapsulant to enable monolithic integration. The resulting processors could outperform existing digital AI training hardware solutions by over 240 times.

Contact

ARPA-E Program Director:
Dr. William Horak
Project Contact:
Dr. Murat Onen
Press and General Inquiries Email:
ARPA-E-Comms@hq.doe.gov
Project Contact Email:
monen@evacorp.ai

Related Projects


Release Date: