Hailo-8 M.2 AI Acceleration Module

Hailo-8 M.2 AI Acceleration Module is a plug-in neural-network inference accelerator for edge computers. Its Hailo-8 processor delivers up to 26 TOPS and executes low-latency inference locally, allowing cameras, robots and industrial systems to process multiple model or video streams without sending every workload to a data centre.

The module comes in M.2 Key M, B+M and A+E formats. Key M uses a four-lane PCIe Gen 3 interface, while the smaller variants use two lanes. Hailo provides a Dataflow Compiler and support for TensorFlow, TensorFlow Lite, ONNX, Keras and PyTorch models on Linux or Windows hosts. Extended-temperature versions cover -40 to 85 degrees Celsius, and the commercial modules carry CE and FCC certification.