The launch marks Huawei's most prominent public demonstration yet of its ambitions in the artificial intelligence space, particularly in China’s fast-growing and increasingly competitive AI market. The CloudMatrix 384 drew significant attention at WAIC, attracting a steady stream of visitors and industry professionals to the company’s booth throughout the three-day event.
Though Huawei representatives on site declined to provide technical briefings on the system, the machine itself has been under close global scrutiny since it was first announced in April. Semiconductor experts and research groups have pointed to the system as evidence of China’s accelerating capabilities in AI hardware—even as the country faces restrictions on access to U.S.-made chips and systems.
According to analysis by Dylan Patel of SemiAnalysis, the CloudMatrix 384 may outpace Nvidia’s GB200 NVL72 system on several performance metrics. Patel argues that while Huawei’s individual chips may lag behind Nvidia’s in raw power, the company has compensated through innovative system-level engineering. The CloudMatrix 384 employs a “supernode” architecture that integrates 384 of Huawei’s latest Ascend 910C chips, designed to maximize bandwidth and processing efficiency through high-speed interconnection.
In contrast, Nvidia’s GB200 NVL72, currently considered the U.S. chipmaker’s most advanced system-level solution, integrates 72 B200 chips. Yet, the sheer scale of Huawei’s configuration—and its ability to harness more processing units effectively—may give it an edge in specific computing tasks.
Notably, Huawei's AI ambitions are not purely theoretical. In June, Huawei Cloud CEO Zhang Pingan confirmed that the CloudMatrix 384 is already operational within Huawei’s cloud infrastructure, offering clients access to powerful AI capabilities through the company’s platform.
Despite ongoing U.S. export restrictions that limit Huawei’s access to some cutting-edge semiconductor technologies, the company has emerged as China’s leading domestic alternative in the AI hardware space. Even Nvidia CEO Jensen Huang acknowledged Huawei’s rapid progress in a May interview with Bloomberg, singling out the CloudMatrix system as a sign of how fast the Chinese firm is advancing.
While Huawei has yet to release full technical documentation or comment in depth on the system’s commercial deployment, its showcase at WAIC signals the company’s growing confidence—and its intent to become a major player not only within China but potentially in the global AI infrastructure market.