Huawei, the leading technology company, has introduced a new AI infrastructure designed to enhance computing capacity and strengthen its position against chipmaker Nvidia.
During its Huawei Connect conference in Shenzhen on Thursday, the company revealed the SuperPoD Interconnect, a technology capable of connecting as many as 15,000 graphics cards—including its own Ascend AI processors—to significantly ramp up computational performance.
This solution appears positioned as an alternative to Nvidia’s NVLink platform, which enables rapid data transfer between AI accelerators.
Innovations like these are vital for Huawei as it seeks to challenge leading semiconductor firms such as Nvidia. Although Huawei’s AI processors currently lag behind Nvidia’s in raw performance, grouping them together can provide users with the greater computing resources needed for advanced AI training and scaling.
This announcement follows closely after China imposed a ban on local tech firms purchasing Nvidia’s products, including the RTX Pro 600D servers that were developed specifically for the Chinese market.
TechCrunch has contacted Huawei to obtain further details.
