Alibaba’s Hanguang 800

Company developing new AI inference chip for smartphones and servers.

September 3, 2025

Jon Peddie

The Hanguang 800, developed by Alibaba’s T-Head and unveiled in 2019, is a cloud-based AI inference NPU fabricated on TSMC 12 nm with 17 billion transistors. It delivers ~78,563 images/sec ResNet-50 throughput at ~500 IPS/W efficiency, targeting Alibaba’s internal workloads like search and recommendations. The chip features four cores with ring bus architecture and 192MB distributed memory. Used exclusively within Alibaba Cloud rather than sold commercially, it supports TensorFlow, MXNet, Caffe, and ONNX frameworks. Alibaba is reportedly developing a next-generation 7 nm successor for broader deployment. (Source: T-Head) In the fall of 2019, T-Head (a subsidiary of Alibaba) announced it

...

Enjoy full access with a TechWatch subscription!

TechWatch is the front line of JPR information gathering service, comprising current stories of interest to the graphics industry spanning the core areas of graphics hardware and software, workstations, gaming, and design.

A subscription to TechWatch includes 4 hours of consulting time to be used over the course of the subscription.

Already a subscriber? Login below

Alibaba’s Hanguang 800

Enjoy full access with a TechWatch subscription!

This content is restricted

Login