Analog compute-in-memory chips built to run AI inference at the edge without the power bill of a GPU rack. Physics first, software second.
Analog AI accelerators for edge and embedded systems.