Photonic Chiplet for 20x Energy Efficient AI Compute

Photonic analog systolic array projecting 20x energy efficient matrix multiplication with throughput, accuracy and dynamic range comparable to digital tensor cores.

Product

Read about our photonic analog systolic array in our technical pitchbook. Request via pxshen@alumni.stanford.edu

We also maintain our own FDTD solver for photonic simulation and inverse design. Available to the community. An updated version will be released mid May 2026.

Honors

Nvidia Inception Program (2026)
Google Startup Program Ecosystem Tier (2025)

Team

Founder: Paul Shen | Stanford MS BS