Innatera designs neuromorphic processors that run spiking neural networks next to sensors, keeping always-on pattern recognition in the microwatt range. Its Pulsar chip combines a spiking neural network engine, a conventional CNN accelerator and a RISC-V core so devices such as wearables, radar sensors, smart buildings and audio products can detect events locally before waking a main processor. The company ships the Talamo SDK, which uses PyTorch to train and map models.
Focus: neuromorphic processors, sensor ai.
Tags: neuromorphic · spiking neural networks · sensor fusion · risc-v · always-on
More edge ai companies
Axelera AI
Metis AI processing units with digital in-memory computing for edge vision workloads
AAEON
Embedded boards and BOXER edge AI systems built on Intel, NVIDIA Jetson and Hailo
Aetina
Taiwanese builder of NVIDIA Jetson and GPU edge AI systems for industrial and medical vision