Chinese tech giant Huawei Technologies revealed its new AI computing system, the CloudMatrix 384, at the World Artificial Intelligence Conference (WAIC) in Shanghai on July 26.
The system is seen by industry experts as a strong rival to Nvidia’s most advanced AI hardware, signaling Huawei’s ambition to expand its presence in the fast-growing AI market.
The CloudMatrix 384, which debuted publicly for the first time at WAIC, integrates 384 of Huawei’s latest 910C chips. According to semiconductor research group SemiAnalysis, it outperforms Nvidia’s current flagship system, the GB200 NVL72, which uses 72 B200 chips, on several key performance metrics.
The system’s high performance is attributed to Huawei’s innovative “supernode” architecture, which allows ultra-fast interconnection between chips. This design compensates for individual chip performance differences by leveraging more chips and advanced system-level engineering. Huawei’s CloudMatrix 384 is already operational on Huawei’s cloud platform, according to Huawei Cloud CEO Zhang Pingan.
Despite facing U.S. export restrictions, Huawei has emerged as China’s leading domestic supplier of chips vital to AI development. Nvidia’s CEO Jensen Huang acknowledged Huawei’s rapid progress earlier this year, citing the CloudMatrix as an example of their advancement.
Huawei staff at WAIC declined to provide further details on the system, and the company has not responded to additional questions.
The unveiling of CloudMatrix 384 marks a significant step in Huawei’s efforts to challenge Nvidia’s dominance in AI computing hardware both within China and internationally.
