_updates.jpg)
Chinese telecommunication giant Huawei has announced cutting-edge AI chip and computing plans in a bid to counter the dominance of US tech giant Nvidia.
After years of halt, Huawei ramps up its computing ambitions by declaring to launch the world’s most powerful computing clusters “Atlas 950 SuperCluster” as soon as next year.
Eric Xu, Huawei's current rotating chairman said in annual Huawei Connect conference, "We will follow a 1-year release cycle and double compute with each release.”
According to Huawei officials, Atlas 950 supernode would be capable of supporting 8,192 Ascend chips. On the contrary, the to-be-launched Atlas 950 SuperCluster would use more than 500,000 chips.
The more advanced form of computing cluster Atlas 960 version would support 15,488 Ascend chips per node. It is expected to be launched in 2027. The Atlas 970 version will be launched in 2028.
Moreover, the tech firm is also planning to introduce new computing power supernodes that would use more than 1 million Ascend chips, leading to chips interconnected at high speeds.
According to Wang Shen, data center infrastructure practice lead at tech research firm Omdia, "Huawei is leveraging its strengths in networking, along with China's advantages in power supply, to aggressively push supernodes and offset lagging chip manufacturing.”
The announcement of innovative chip plans came on the heels of China’s restriction on tech firms, blocking them from purchasing Nvidia AI chips.
In the recent investigation, China accused Nvidia of violating antimonopoly law.
