Technology and Science17 September, 20263 min read
Huawei Moves Up Ascend 960DT AI Chip Launch to Q1 2027, Targets Nvidia
This article is an automated aggregation of coverage from news outlets. Our editorials offer in-depth analysis curated by the NewsCord team.
Huawei plans two AI chips in 2027: 960DT in Q1 and Ascend 960PR in Q3. Chips linked as superclusters via UnifiedBus to boost AI data-center performance.
Huawei accelerates Ascend 960
Huawei said at its Huawei Connect conference in Shanghai that it will move up the launch of its Ascend 960DT AI chip to the first quarter of 2027, with the companion Ascend 960PR expected in the third quarter of 2027. Reuters reported that rotating chairman Eric Xu said Huawei cannot produce enough artificial-intelligence computing equipment to meet demand in China and is limiting overseas sales, while David Wang said the 960DT would be ready in the first quarter of 2027. Huawei also framed its push as a response to U.S. export controls restricting China’s access to Nvidia’s most advanced chips and semiconductor manufacturing equipment, and Reuters quoted Xu linking the chip push to China’s effort to reduce reliance on foreign technology.
In the same Reuters account, Xu said, "Since we don't have enough capacity to even satisfy the demand in China, we don't have a plan to expand into the international market in a fully-fledged way," as Huawei detailed its next-generation roadmap. The Tech Portal similarly quoted David Wang announcing, "Ascend 960DT will be available in Q1 2027, three quarters ahead of the company’s original roadmap," tying the accelerated timeline to Huawei’s challenge to Nvidia.
UnifiedBus and superclusters
Huawei’s strategy, as described by Reuters, is to compensate for limits on individual Chinese chips by connecting large numbers of processors so they can work together as one larger computing system. Reuters said Huawei unveiled a new computing architecture called Peerium, and that its UnifiedBus technology connects processors, memory, storage and networking equipment inside those systems. Reuters reported that Huawei said its new Ascend 960 supernode would connect as many as 4,096 AI processors, and that multiple supernodes could be linked into clusters supporting as many as 1 million processors.
In a separate account, TradingView said Huawei has developed 11 chips around UnifiedBus for its supernode and supercluster systems, and that Wang said the company has already shipped more than 1,000 supernode systems to more than 370 customers. TradingView also quoted Wang saying the company plans to launch the 960DT in the first quarter of 2027 and the Ascend 960PR in the third quarter of 2027, positioning the interconnect strategy as a way to expand domestic AI-computing capacity.
Demand, developers, and stakes
Huawei said it is already seeing demand outstrip supply in China, with Reuters quoting Eric Xu on the lack of capacity and on limiting overseas sales while Huawei tested its Ascend 950DT and expected many developers to begin training next year. Reuters also reported that Huawei said more than 5,200 developers were active each month on software for its Ascend chips and that more than 40 AI models had been trained directly on its computing platform. In the same Reuters piece, Xu said, "I think Ascend market share should be bigger than Nvidia's," while adding that Huawei did not provide data to support the estimate.
The Tech Portal added that Huawei’s AI chip software ecosystem now has 5,270 monthly active developers and that Huawei has shipped more than 1,000 multi-chip “supernode” systems to more than 370 customers. Quartz reported that Eric Xu said, "Since we don't have enough capacity to even satisfy the demand in China, we don't have a plan to expand into the international market in a fully-fledged way," and it also said Huawei planned to release the Ascend 970 in 2028 and the Ascend 980 in 2029.