Huawei has unveiled a three-year plan to challenge Nvidia in AI chips, focusing on massive chip clusters, faster data links, and strong policy support.

At its Huawei Connect conference, Chairman Eric Xu detailed a strategy to link up to 15,488 Ascend AI chips using a new UnifiedBus protocol. Huawei claims this setup can move data up to 62 times faster than Nvidia’s upcoming NVLink144 technology. The goal is to offset weaker chip performance with scale and networking strength.

The announcement signals Huawei’s confidence in its local manufacturing supply despite US sanctions that cut it off from top chipmakers like TSMC. Bernstein analysts say the move marks progress toward a self-reliant Chinese semiconductor ecosystem.

China is pushing homegrown chip innovation as US tariffs and export curbs tighten. Huawei still lags in raw chip power—its next-gen Ascend 950 is said to reach only about 6% of Nvidia’s upcoming VR200—but hopes to bridge the gap by clustering up to a million chips and improving memory architecture.

While experts doubt whether Huawei can mass-produce its designs, the company believes its SuperPod cluster tech and high-speed interconnects could eventually rival Nvidia’s dominance in AI computing.

Unicorn Signals leverages advanced AI technology to provide you with powerful market predictions and actionable stock scans. Download the app today and 10x your trading & investing journey!