Huawei unveils AI cluster breakthrough to supercharge China's tech drive

Huawei has developed powerful AI supernodes using its own chips, aiming to boost China's computing power and reduce reliance on foreign technologies like Nvidia

Huawei Technologies has unveiled new hardware it claims can deliver world-class computing power without relying on Nvidia’s advanced chips — a move that could ease China’s supply constraints in artificial intelligence, the South China Morning Post reported.

 

The Shenzhen-based telecom giant said it had developed what it calls the “world’s most powerful” supernode computing cluster using domestically produced chips, boosting China’s push for self-reliance in AI technology.

 

“Huawei is seeking to build a ‘supernode + cluster’ computing solution using chip manufacturing processes available in China to meet the growing compute needs,” said Xu Zhijun, Huawei’s deputy chairman and rotating chairman, at the company’s annual Connect Conference in Shanghai, as quoted by the news report.   

 

Plans for upgraded AI chips

 

Xu also shared Huawei’s roadmap for its Ascend AI chips over the next three years, including the Ascend 950PR slated for release in the first quarter of 2026. These plans run alongside AI chip developments from Nvidia and AMD.

 

Xu’s announcement comes amid Beijing’s push for Chinese tech giants to stop buying Nvidia’s China-tailored chips, which were designed under US export restrictions. Nvidia CEO Jensen Huang commented on the new stance, saying he was “disappointed with what I see, but they have larger agendas to work out between China and the United States, and I’m patient about it".

 

Supernodes: High-powered AI clusters

 

Huawei, which is sanctioned by the US, is taking a leading role in enabling China’s AI ambitions without using American technology. A supernode is a high-performance cluster that combines multiple AI accelerators through ultra-fast interconnects.

 

On Thursday, Huawei unveiled the Atlas 950 SuperPoD and Atlas 960 SuperPoD, capable of supporting up to 8,192 and 15,488 in-house Ascend processors, respectively. The company claims these systems lead the world in processor count, total computing power, memory capacity, and interconnect bandwidth, the news report said.

 

“Computing power was the key to artificial intelligence in the past, and it will remain so in the future – especially for China’s AI development,” Xu said.

 

Huawei also introduced the Atlas 950 SuperCluster and Atlas 960 SuperCluster, which can scale to 500,000-1 million processors, making them the “largest AI compute clusters” globally.   

 

China seeks to boost domestic AI chip production

 

China aims to triple its AI chip production in 2026 to reduce reliance on Nvidia, the Financial Times reported last month.

 

Huawei plans to start production at a dedicated AI chip plant by the end of this year, with two additional facilities expected to open in 2026. While these plants are designed to support Huawei, ownership details remain unclear. Huawei told the FT that it does not plan to own the plants itself.

 

Combined, these three potential plants could exceed the production capacity of China’s top chipmaker, Semiconductor Manufacturing International Corporation (SMIC). SMIC intends to double its manufacturing capacity for 7-nanometer chips next year, with Huawei as its largest customer.

 

Beijing is fast-tracking domestic AI chip development, aiming to rival Nvidia’s China-specific H20 chips, which have raised security concerns.