Neurophos develops optical processing units (OPUs) designed to deliver AI compute at unprecedented speed and energy efficiency. By performing massive in-memory matrix multiplications with photonic hardware, the company aims to achieve high compute density in a compact footprint, targeting teraflop-scale performance per watt. Based in Austin, Texas, Neurophos positions its OPUs as a disruptive approach to AI inference and training hardware that could outperform traditional GPU racks while consuming far less power.
No recent deals for this company.