Huawei Announces New AI Infrastructure Amidst Nvidia China Restrictions

Huawei Advances AI Infrastructure to Challenge Nvidia

Huawei, a leading technology corporation, has introduced novel AI infrastructure designed to significantly enhance computing capabilities. This development aims to strengthen the company’s competitive position against prominent chip manufacturer Nvidia.

Introducing SuperPoD Interconnect Technology

During the keynote address at the Huawei Connect conference held on Thursday in Shenzhen, China, Huawei revealed its new SuperPoD Interconnect technology. This technology enables the connection of up to 15,000 graphics cards.

The system supports Huawei’s own Ascend AI chips, facilitating a substantial increase in overall compute power.

A Competitive Alternative to NVLink

This new technology appears to directly challenge Nvidia’s NVLink infrastructure. NVLink is known for its ability to provide high-speed communication between AI chips.

Boosting Huawei’s AI Capabilities

Technologies of this nature are vital for Huawei to effectively compete in the semiconductor market, particularly against companies like Nvidia.

While Huawei’s AI chips currently offer less raw power compared to Nvidia’s offerings, the ability to cluster them together provides users with increased computational resources.

This enhanced compute power is essential for both the training and scaling of complex AI systems.

China's Restrictions on Nvidia Hardware

This announcement follows closely after China imposed a ban on domestic tech companies purchasing hardware from Nvidia. The restriction specifically includes Nvidia’s RTX Pro 600D servers, which were tailored for the Chinese market.

TechCrunch contacted Huawei to request further details regarding this new infrastructure.