EdgeCortix develops energy-efficient AI accelerators based on its Dynamic Neural Accelerator architecture, which reconfigures data paths at runtime to keep compute units busy. The SAKURA-I and SAKURA-II chips ship as M.2 and PCIe cards for robotics, smart cities, defense, telecom and industrial customers that need to run vision and generative models locally. The MERA compiler and software framework imports models from PyTorch, TensorFlow and ONNX.
Focus: ai accelerators, reconfigurable npu.
Tags: ai accelerator · reconfigurable · computer vision · generative ai · pcie
More edge ai companies
i-PRO
Japanese camera maker, formerly Panasonic Security, with on-camera AI analytics
Idein
Actcast platform for running and managing AI applications on fleets of edge devices
Sony Semiconductor Solutions
Image sensors with on-chip AI processing and the AITRIOS platform for edge sensing