Huawei unveiled its roadmap for the next generation of Ascend AI chips during the recent Huawei Connect event in Shanghai, China
In sum – what to know:
Doubled chip output – Huawei’s plan calls for 600,000 Ascend 910C units next year and up to 1.6 million dies across models.
Technology gap narrowing – Huawei’s 7nm Ascend chips still trail Nvidia’s 4nm GPUs, but advances in packaging and interconnects point to steady progress.
Chinese vendor Huawei Technologies is preparing to significantly expand production of its most advanced AI processors, according to a Bloomberg report. The Shenzhen-based company aims to manufacture about 600,000 units of its Ascend 910C chip in 2026, roughly twice the expected output for this year, the report said, citing people familiar with the plans.
Including other models in the Ascend line, Huawei could distribute up to 1.6 million dies next year, the report added.
The production push underscores Huawei’s growing role in China’s effort to reduce reliance on foreign technology. Partnering with Semiconductor Manufacturing International Corp. (SMIC), Huawei is using an enhanced 7-nanometer process to make the chips, while rivals like Nvidia rely on more advanced 4nm nodes from TSMC, according to the report.
Chinese cloud and internet players including Alibaba, Tencent, and DeepSeek are expected to be among the main customers, the report added.
For now, analysts expect Huawei’s chips to remain primarily used for inference rather than training the largest AI models.
Huawei unveiled its roadmap for the next generation of Ascend AI chips during the recent Huawei Connect event in Shanghai, China, outlining plans for three new chip series over the next three years.
The roadmap begins with the Ascend 950 family, which includes two chips: the Ascend 950PR, optimized for prefill inference and recommendation, and the Ascend 950DT, designed for decoding and training. Both chips share the Ascend 950 Die and deliver improvements in multiple dimensions, the company said.
The Ascend 950PR will hit the market in the first quarter of 2026, available in card and SuperPoD server formats.
The Ascend 950DT, set to be launched in the last quarter of 2026, targets training and decoding workloads with higher demands for memory and bandwidth. It integrates HiZQ 2.0 HBM, offering 144 GB memory, 4 TB/s memory access bandwidth, and a 2 TB/s interconnect.
Huawei will also introduce the Ascend 960 chip in the last quarter of 2027. Compared with the Ascend 950 chip, it will double compute power, memory, and interconnect capacity. The chip also introduces HiF4, a proprietary 4-bit precision format designed to deliver greater accuracy than standard FP4 solutions, promising a major boost to inference throughput.
The roadmap culminates in the Ascend 970, scheduled for the fourth 2028. While still under development, the company said the target is to double FP4 and FP8 compute power compared with the 960, double interconnect bandwidth, and increase memory bandwidth by at least 1.5 times.
“Generally, we will follow a 1-year release cycle and double compute with each release. Throughout this process, we will keep evolving our Ascend chips, making them easier to use, supporting more data formats, and increasing their bandwidth. The goal is to stay on top of ever-growing demand for AI compute,” Huawei’s rotating chairman Eric Xu said.
“Gaps still exist between Huawei and Nvidia in terms of individual chips and ecosystems, and we are working to close these gaps. But we have the capability to build SuperPoDs and SuperClusters – which are the source of our confidence,” Xu recently told Chinese media.
