Photo Credit: Joana Stock / Shutterstock.com
Huawei used a company event in Shanghai on Thursday to confirm two new AI chips arriving in 2027, the 960DT and the Ascend 960PR, extending its push to compete directly with Nvidia. The timeline is set. Rotating chairman David Wang laid it out, with the 960DT expected in the first quarter of 2027 and the Ascend 960PR following in the third quarter, as Reuters reported.
Central to the plan is Huawei’s UnifiedBus technology, which links large numbers of AI chips so they function as a single, far larger computing system. Individual chips cannot run the biggest AI programs alone, even the most capable ones. That constraint is the whole point. Wang said Huawei has already built 11 semiconductors around UnifiedBus, and the company’s largest connected setups, called superclusters, can reportedly handle up to a million AI processors working together.
Advertisement
Advertisement
On a smaller scale, Huawei says it has delivered more than 1,000 of these linked systems, known as supernodes, to over 370 customers worldwide. The rest stays vague. Wang did not say how many chips a single supernode holds, who the customers are, or how the shipment figures break down, leaving several gaps in the picture.
US export controls have restricted China’s access to the most advanced chips and manufacturing equipment, and that pressure has pushed Huawei to the front of China’s domestic AI chip supply. Nvidia remains the company to beat. Its hardware and software are used by developers across the industry, and closing that gap depends on more than silicon. Huawei needs developers building on its chips, and Wang said its AI chip ecosystem currently counts 5,270 monthly active developers. That number will need to keep climbing for Huawei to be treated as a genuine Nvidia alternative rather than a fallback option.
The chip rivalry between the US and China shows no sign of easing, and Huawei has now signaled it intends to remain a major participant through 2027. We are keeping an eye on how the developer numbers and supernode deployments progress as the launch windows approach.