Huawei on Thursday unveiled the Ascend 960 SuperPoD, which it described as the world’s first AI supernode to use near-packaged optics (NPO) technology and said can support large-scale training and inference for models with up to 10 trillion parameters, a Huawei official said at Huawei Connect 2026.
Why Ascend 960 SuperPoD matters
The SuperPoD, or supernode, links multiple physical computing nodes through high-bandwidth, low-latency interconnects so they can operate as a single logical computer. Huawei presented the system as a response to rising demand for greater computing scale and lower latency as AI models grow larger and more complex.
Huawei also accelerated the rollout of its next-generation Ascend 960 AI chips after development progressed faster than expected, media reports said. Those reports said the new chips deliver roughly double the computing performance of the previous generation. The Ascend 960DT is expected to be ready in the first quarter of 2027, three quarters earlier than originally planned, while the Ascend 960PR is scheduled for the third quarter of 2027, one quarter ahead of prior timelines.
Analysts cited by the company and press observers say efficiently linking large numbers of processors is increasingly important as model sizes expand, and SuperPoDs are designed to address that challenge.
China’s broader push to scale AI computing
The announcement coincides with broader efforts in China to expand AI computing infrastructure. The Ministry of Industry and Information Technology said China’s intelligent computing capacity reached 2,185 EFLOPS by the end of June, up 177 percent year-on-year. The ministry’s recent industry development plan calls for the orderly deployment of intelligent computing clusters with 10,000, 100,000 or more accelerator cards and for greater adaptation to domestically developed computing chips.
At the corporate level, competition in the AI chip and system space is placing greater emphasis on system-level computing: JD Cloud and Chinese GPU maker Moore Threads recently announced plans to build a 100,000-GPU computing cluster, illustrating a push to deploy domestically developed AI chips at much larger scale.
An Atlas 950 SuperPoD was also displayed by Huawei at the 2026 World AI Conference and High-Level Meeting on Global AI Governance in Shanghai on July 19, 2026, where attendees viewed the system, according to event materials and images.
