Chinese tech giant Huawei has unveiled its once-secretive AI chip development program, an announcement that could reshape expectations across the fast-moving, high-stakes global industry.<\/p>\n
On September 18, Huawei chairman Eric Xu announced a three-year roadmap<\/a> for the company\u2019s Ascend series of processors at the Huawei Connect 2025<\/a> conference in Shanghai.<\/p>\n Coming soon after the Chinese government decided to block Chinese companies from using Nvidia processors<\/a>, Huawei\u2019s announcement will heighten competition between Chinese and American AI efforts while signaling a significant reduction in China\u2019s dependence on American technology.<\/p>\n Explaining the roadmap, Xu told the audience that, following the launch of the Ascend 910C earlier this year, Huawei will release three new chip series \u2013 the Ascend 950, 960, and 970 \u2013 in 2026, 2027, and 2028, respectively.<\/p>\n \u201cGenerally, we will follow a one-year release cycle and double compute with each release,\u201d he said. That pace of development, if achieved, is comparable to that of leading US chipmakers Nvidia and AMD.<\/p>\n Xu also said Huawei will offer proprietary high-bandwidth memory (HBM) solutions for variants of the Ascend 950, revealing that, along with Chinese memory maker CXMT, Huawei has been working hard to escape dependence on HBM imported from Korea \u2013 a target of US sanctions.<\/p>\n Xu\u2019s keynote address, titled \u201cGroundbreaking SuperPoD Interconnect: Leading a New Paradigm for AI Infrastructure,\u201d outlined Huawei\u2019s plans \u201cto create a new computing architecture, and develop computing SuperPoDs and SuperClusters, to sustainably meet long-term demand for computing power.\u201d<\/p>\n \u201cSuperPoDs,\u201d Xu continued, \u201chave become the main form factor of products for large-scale AI infrastructure. They are becoming the new norm. A SuperPoD is a single logical machine, made up of multiple physical machines that can learn, think, and reason as one. As computing demand continues to grow, so will SuperPoDs.\u201d A SuperPoD functions as a single computer, while SuperClusters connect multiple SuperPods.<\/p>\n In March 2025, Huawei launched its Atlas 900 A3 SuperPoD, which incorporates up to 384 Ascend 910C chips. According to Xu, it is currently the largest SuperPoD in the world.<\/p>\n \u201cMaybe you\u2019ve heard about the CloudMatrix384,\u201d Xu said. \u201cIt\u2019s a cloud service instance that Huawei Cloud has built on top of our Atlas 900 A3 SuperPoDs. Since its launch, we\u2019ve deployed more than 300 Atlas 900 A3 SuperPoDs to serve over 20 customers in sectors like ISP [Internet Service Provider], telecoms, and manufacturing.\u201d<\/p>\n This squares with the assessment of Dylan Patel and his colleague at SemiAnalysis, who found that Huawei\u2019s Ascend 910C AI processor is significantly more impressive when used in the company\u2019s CloudMatrix 394 rack-scale AI data center solution \u2013 a complete system consisting of 384 Ascend 910C processors, along with servers, networking, storage, power management and cooling.<\/p>\n In their estimation, the CloudMatrix 394 \u201ccompetes directly\u201d with Nvidia\u2019s top-end GB200 Grace Blackwell Superchip.<\/p>\n