AetherChip raises $120M Series B and launches plug-and-play edge AI module

Tech · 6 min read

AetherChip's new module pairs custom silicon with an SDK optimized for low-latency, low-power inference of moderate-sized transformer and vision models. The company claims orders-of-magnitude energy improvements compared with general-purpose CPUs and a simplified deployment experience for device makers.

The Series B will be used to expand manufacturing, deepen software support for quantized models, and subsidize partner integrations for robotics, smart cameras, and automotive cockpit systems. Sequoia led the round, joined by NVIDIA Ventures and several strategic hardware partners.

AetherChip emphasized developer ergonomics: the module ships with model templates and a cloud-based profiler that translates model op counts into expected latency and power budgets on real devices. The product launch is targeted at electronics OEMs and prototyping teams building new hardware-enabled UX experiments.