ASUS launches world's first USB AI accelerator with generative AI support

ASUS has launched the UGen300, the world's first USB edge AI accelerator with built-in generative AI support. The device features a Hailo-10H AI processor delivering 40 AI TOPS and 8 GB LPDDR4 memory, all in a compact USB-C device consuming just 2.5 watts.

The UGen300 enables running large language models (LLMs) and vision-language models (VLMs) locally on any machine with a USB-C port. The device supports Windows, Linux, and Android, and is compatible with TensorFlow, PyTorch, and ONNX out of the box. An M.2 version is also available.

At just 105 x 50 x 18 mm, the UGen300 is a plug-and-play solution that adds AI acceleration to machines lacking dedicated NPU or GPU power. The processor supplements the host device's CPU and frees up system resources.

For enterprises looking to run AI inference locally without cloud dependency, this is a noteworthy development. USB-based edge AI dramatically lowers the barrier to local AI processing, which is relevant for privacy, latency, and cost control.

Hailo CRO Max Glover called the launch proof that AI should be "accessible, affordable, and designed for anyone to build with."