Luxonis lands $14M Series A to scale ‘physical AI’ perception platform

Luxonis sits in the “boring but vital” layer of the AI stack: giving robots and machines the eyes and low‑latency perception they need to do useful work. The $14 million Series A is modest compared with cloud model rounds, but for a company already shipping OAK cameras and an open DepthAI SDK, it’s fuel to industrialize a platform that thousands of robotics and automation teams rely on. As more attention shifts to ‘physical AI’—agents that act in warehouses, farms and factories—companies like Luxonis effectively become the de facto perception layer for a large swath of the ecosystem.

Because Luxonis runs much of its inference on‑device, it sidesteps some of the bandwidth, latency and privacy headaches of cloud‑only AI. That’s critical for safety‑critical systems and for emerging markets with unreliable connectivity. It also creates a natural interface between edge and cloud models: edge perception feeding into higher‑order reasoning agents running in data centers.

In the AGI context, high‑quality, widely‑deployed perception hardware is what will let more advanced general models actually see and manipulate the world at scale. Luxonis doesn’t need to build AGI itself to matter; it just needs to be in the critical path whenever someone wants to put an intelligent system on wheels, tracks or manipulators.