Huawei announced the launch of its own SuperPoD Interconnect technology, designed to connect up to 15,000 GPUs, including Ascend chips, into a cluster. The presentation took place on September 18, 2025, in Shenzhen, at the Huawei Connect conference.
The main goal of the development is to ensure high data transfer rates between devices and enable scalable computing power for running large AI models. Essentially, Huawei has unveiled a response to Nvidia's NVLink, which has long been used in the industry for similar tasks.
Experts note that Ascend's performance is currently inferior to Nvidia's solutions, but the Chinese company is focusing on combining chips into powerful clusters. This approach should bridge the gap and offer customers a comparable level of capabilities for training neural networks.
Notably, the release took place just one day after Chinese authorities restricted access by domestic corporations to Nvidia products, including RTX Pro 6000D servers. These measures are aimed at accelerating the development of China's semiconductor industry and reducing dependence on American technology.
According to Chinese government officials, local manufacturers are already capable of producing chips that are as performant as Nvidia's specially scaled-down versions for the Chinese market. Huawei's new initiative confirms Beijing's commitment to technological independence in the strategically important field of artificial intelligence.
Analysts emphasize that this technology not only closes the gap created by restrictions on Nvidia equipment imports but also forms the foundation for creating a domestic high-performance computing ecosystem. This will allow China to advance more rapidly in AI and reduce its vulnerability to external factors.
Thus, Huawei is strengthening its position amid the global competition for leadership in artificial intelligence while simultaneously demonstrating its willingness to resist sanctions pressure.
Earlier, it was reported that Nvidia invested $5 billion in Intel as part of a strategic alliance aimed at developing AI technologies, confirming the intensifying competition between the industry's largest players.