## Huawei Unveils Powerful AI Chip Clusters to Challenge Nvidia Amidst US Sanctions **News Title:** Huawei touts ‘world’s most powerful’ AI chip cluster as Nvidia’s China challenges mount **Report Provider:** CNBC **Author:** Evelyn Cheng **Date:** Published September 18, 2025 This news report details Huawei's significant advancement in artificial intelligence (AI) computing infrastructure with the announcement of its new "Atlas 950 SuperCluster" and a more advanced "Atlas 960" version. These systems are powered by Huawei's in-house Ascend chips and are designed to compete directly with U.S. rival Nvidia, particularly in the context of U.S. sanctions aimed at limiting China's access to advanced semiconductors. ### Key Findings and Conclusions: * **New AI Computing Systems:** Huawei has announced new computing systems, the Atlas 950 SuperCluster and Atlas 960, designed to power AI applications. * **In-House Ascend Chips:** These systems are built upon Huawei's proprietary Ascend chips, reflecting a strategic push for self-reliance in semiconductor technology. * **Challenging Nvidia:** The announcement directly challenges Nvidia's dominance in the AI chip market, especially in China, which is facing U.S. restrictions on advanced chip imports. * **"Double Compute" Goal:** Huawei aims to "double compute" capabilities with each annual release of its Ascend chips through the end of 2028. * **Overcoming Sanctions:** Huawei is using system design to overcome technical limitations imposed by U.S. sanctions, by grouping large numbers of less efficient, homegrown chips. * **Ambition for AI Leadership:** The move underscores China's and Huawei's ambition to become world leaders in AI. ### Key Statistics and Metrics: * **Atlas 950 Supernode:** * Supports **8,192 Ascend chips**. * The full Atlas 950 SuperCluster will use **more than 500,000 chips**. * **Atlas 960 (slated for launch in 2027):** * Supports **15,488 Ascend chips per node**. * The full supercluster would have **more than 1 million Ascend chips**. * **Comparative Performance Claims (Huawei):** * Atlas 950 supernode is claimed to deliver **6.7 times more computing power** than Nvidia's NVL144 system. * Atlas 950 supercluster is claimed to have **1.3 times the computing power** of Elon Musk's xAI Colossus supercomputer. * **Chip Efficiency vs. Quantity:** Research firm SemiAnalysis found that while each Ascend chip delivers about one-third the performance of an Nvidia processor, Huawei achieves an advantage by using **five times as many chips**. * **Current Deployments:** Huawei has deployed **more than 300 of its Atlas 900 A3 supernodes** to over 20 customers across telecoms, manufacturing, and other industries. ### Significant Trends or Changes: * **Increased Emphasis on Self-Reliance:** Huawei's announcement aligns with the Chinese government's growing emphasis on self-reliance in chip technologies. * **Strategic Response to Sanctions:** Chinese companies, including Huawei, are increasingly turning to aggregating numerous less powerful chips to achieve comparable AI computing capabilities in response to U.S. restrictions. * **Escalating Competition:** The move intensifies the competition between Huawei and Nvidia, with China actively promoting homegrown alternatives. ### Notable Risks or Concerns: * **Exaggeration of Capabilities:** George Chen, partner and co-chair of digital practice at The Asia Group, cautioned that Huawei might exaggerate its technical capabilities. * **U.S. Pressure on Nvidia:** China's actions, including extending a probe into Nvidia over alleged monopolistic practices and reportedly ordering local tech giants to stop tests and orders of specific Nvidia chips (RTX Pro 6000D), are increasing pressure on the U.S. chipmaker. * **Nvidia CEO's Disappointment:** Nvidia CEO Jensen Huang expressed disappointment regarding the reported ban on certain chips. ### Material Financial Data: * No specific financial figures were provided in this report. However, the announcement of new, high-capacity AI computing systems implies significant investment in research and development and manufacturing capabilities. ### Important Recommendations: * While no explicit recommendations are made, the report highlights the strategic importance of computing power for AI and Huawei's ambition to lead in this domain. ### Contextual Information: * The news was announced on Thursday, September 18, 2025, at Huawei's annual Huawei Connect event in Shanghai. * The Atlas 900 AI Cluster was announced two years prior at the same event. * The U.S. has been actively trying to restrict China's access to advanced semiconductors for AI development. * Recent trade talks between the U.S. and China included discussions on resolving issues related to TikTok, owned by ByteDance. In summary, Huawei's unveiling of its Atlas 950 and 960 AI superclusters, powered by its Ascend chips, marks a significant step in its strategy to challenge Nvidia's dominance. This move is driven by China's push for technological self-reliance and is occurring amidst increasing U.S. pressure on Chinese access to advanced semiconductors. While Huawei claims superior performance, the effectiveness and actual capabilities of these systems compared to Nvidia's offerings will be closely watched, especially given past observations about Huawei's strategy of using a larger quantity of less powerful chips to achieve its computing goals.
Huawei touts ‘world’s most powerful’ AI chip cluster as Nvidia’s China challenges mount
Read original at CNBC →A person walks past a display of an Atlas 900 AI cluster at the Huawei stand during the World Artificial Intelligence Conference at the Shanghai World Expo and Convention Center in Shanghai on July 28, 2025.Hector Retamal | Afp | Getty ImagesBEIJING — Chinese telecommunications giant Huawei announced Thursday new computing systems for powering artificial intelligence with its in-house Ascend chips, as it steps up pressure on U.
S. rival Nvidia.The company said it plans to launch its new "Atlas 950 SuperCluster" as soon as next year.The U.S. has sought to cut China off from the most advanced semiconductors for training AI models. To cope, Chinese companies have turned more to grouping large numbers of less efficient, often homegrown, chips together to achieve similar computing capabilities.
Huawei announced it would roll out three new versions of its Ascend chips through the end of 2028, with the aim to "double compute" capabilities with each year's release.The chips form the basis of Huawei's AI computing infrastructure, in which a supercluster is connected to multiple superpods, which, in turn, are built from multiple supernodes.
Supernodes, which form the base, are built on Ascend chips, using system design to overcome technical limitations imposed by U.S. sanctions.Huawei said its new Atlas 950 supernode would support 8,192 Ascend chips, and that the Atlas 950 SuperCluster would use more than 500,000 chips.A more advanced Atlas 960 version, slated for launch in 2027, would support 15,488 Ascend chips per node.
The full supercluster would have more than 1 million Ascend chips, according to Huawei.It was not immediately clear how the systems compared with those powered by Nvidia chips. Huawei claimed in a press release that the new supernodes would be the world's most powerful by computing power for several years.
In a speech Thursday, Eric Xu, vice chairman and rotating chairman of Huawei, claimed that its forthcoming Atlas 950 supernode would deliver 6.7 times more computing power than Nvidia's NVL144 system, also planned for launch next year.Xu even predicted that Huawei's product would "be ahead on all fronts" compared with another Nvidia system planned for launch in 2027 — and claimed the Atlas 950 supercluster would have 1.
3 times the computing power of Elon Musk's xAI Colossus supercomputer."Huawei's announcement on its computing breakthrough is well timed with recent increasing emphasis by the Chinese government on self-reliance on China's own chip technologies," said George Chen, partner and co-chair, digital practice, The Asia Group.
While he cautioned that Huawei might exaggerate its technical capabilities, Chen pointed out that the Chinese company's ambition to be a world AI leader "cannot be underestimated."Research firm SemiAnalysis found in April that Huawei's self-developed CloudMatrix system was able to perform better than Nvidia's — despite each Ascend chip delivering only about one-third the performance of an Nvidia processor.
Huawei built its advantage by having five times as many chips."Computing power has and will continue to be the key for AI," Rotating Chairman Xu said Thursday in a statement, translated by CNBC. He was speaking at the opening of the company's annual Huawei Connect event in Shanghai. The event runs through Saturday.
Two years ago at the same event, Huawei announced its Atlas 900 SuperCluster. The company currently sells a "Atlas 900 AI Cluster" with "thousands" of Ascend chips.Huawei said Thursday it had deployed more than 300 of its Atlas 900 A3 supernodes to more than 20 customers in telecoms, manufacturing and other industries.
Rising pressure on NvidiaHuawei's announcement comes as China promotes homegrown alternatives to Nvidia. Earlier this week, the two countries wrapped up trade talks in Spain that included a path toward resolving the long-contested U.S. operations of social media app TikTok, owned by Beijing-based startup ByteDance.
In another aggressive signal, China on Monday announced it was extending a probe into Nvidia over alleged monopolistic practices.Pressure has only risen since on the U.S. chipmaker. Its shares fell more than 2% Wednesday after the Financial Times, citing sources, said China has ordered local tech giants to stop tests and orders of the Nvidia RTX Pro 6000D chip.
Nvidia CEO Jensen Huang told reporters he was "disappointed" to hear the news of the reported ban. He's previously described Huawei as a "formidable" competitor.
