Core Insights - Huawei is making significant advancements in AI computing power, showcasing a comprehensive AI computing architecture at the Huawei Connect Conference [1][3] - The company aims to establish a sustainable computing ecosystem based on its proprietary Ascend chips and Kunpeng CPUs, positioning itself as a key player in the AI market [5][8] Group 1: AI Computing Strategy - Huawei's AI computing strategy is centered around the Ascend chip series, with a roadmap for iterative upgrades over the next three years, aiming for a doubling of computing power annually [3][5] - The upcoming Ascend 950 series is designed to enhance training efficiency and support various low-precision data formats, indicating a focus on optimizing AI workloads [3][4] Group 2: Supernodes and Clusters - Huawei introduced new supernode products, Atlas 950 SuperPoD and Atlas 960 SuperPoD, which support 8192 and 15488 Ascend cards respectively, positioning them as leaders in global computing power [6][7] - The Atlas 950 SuperCluster and Atlas 960 SuperCluster are set to exceed 500,000 and 1 million cards, respectively, further solidifying Huawei's dominance in AI infrastructure [6] Group 3: Competitive Landscape - Despite NVIDIA's current dominance in the AI chip market, Huawei is rapidly developing its Ascend and Kunpeng chips to compete effectively within the Chinese AI landscape [5][6] - Huawei's approach includes a focus on system-level engineering and innovation, aiming to create a competitive edge in the evolving AI computing sector [7][9]

Core Viewpoint - Huawei is leading a new paradigm in AI infrastructure with its innovative supernode interconnection technology, emphasizing the importance of computing power in artificial intelligence development [2][8]. Summary by Sections Ascend Chip Development - Huawei has committed to the monetization of Ascend hardware and plans to open-source its CANN compiler and virtual instruction set interface by December 31, 2025 [2]. - The Ascend 950 series, including Ascend 950PR and Ascend 950DT, is set to significantly enhance computing power and efficiency compared to previous models [3][4]. - The Ascend 950 chip will support new data formats and achieve a computing power of 1 PFLOPS for FP8 and 2 PFLOPS for FP4, with interconnect bandwidth increased to 2 TB/s [5][6]. Upcoming Chip Releases - The Ascend 950PR chip is targeted for Q1 2026, focusing on inference prefill and recommendation scenarios, while the Ascend 950DT will be released in Q4 2026, emphasizing inference decode and training [6][7]. - The Ascend 960 chip, expected in Q4 2027, will double the specifications of the Ascend 950, enhancing performance for training and inference [7]. - The Ascend 970, planned for Q4 2028, aims to further upgrade performance metrics across various specifications [7]. Supernode Products - The Atlas 950 supernode, based on the Ascend 950DT, will support 8192 Ascend 950DT chips, achieving FP8 computing power of 8 EFLOPS and interconnect bandwidth of 16 PB/s, set to launch in Q4 2026 [9][10]. - The Atlas 960 supernode, launching in Q4 2027, will support 15488 cards, with FP8 computing power reaching 30 EFLOPS and interconnect bandwidth of 34 PB/s [11]. Interconnection Technology - Huawei has developed a new interconnection protocol, "Lingqu," to support large-scale supernodes, enhancing reliability and bandwidth while reducing latency [18][19]. - The Atlas 950 SuperCluster, consisting of 64 Atlas 950 supernodes, will achieve a total FP8 computing power of 524 EFLOPS, launching alongside the Atlas 950 supernode in Q4 2026 [20]. Future Directions - Huawei aims to continue evolving its supernode and cluster products to meet the growing demands for AI computing power, with a focus on both AI and general computing applications [12][21].

Core Viewpoint - Huawei's chip development has entered a new phase, focusing on AI computing power and advanced chip architecture to compete with global leaders like NVIDIA [1][2][3]. Group 1: Chip Development and Innovations - Huawei has introduced the Atlas 950 SuperPoD and Atlas 960 SuperPoD, which support 8192 and 15488 Ascend cards respectively, aiming to maintain the world's leading computing power [3]. - The company plans to release the Ascend 950PR in Q1 2024, adhering to a "one generation per year, doubling computing power" strategy [7]. - The Ascend 950 series, 960 series, and 970 series have been outlined for future development, with significant enhancements in performance and capabilities [8][21][24]. Group 2: Performance Metrics - The Atlas 950 SuperNode is expected to achieve 8 EFlops in FP8 computing power and 16 EFlops in FP4, with a memory capacity of 1152 TB and interconnect bandwidth of 16.3 PB/s [35]. - The Atlas 960 SuperNode will have a total throughput of 4.91 million TPS for training and 19.6 million TPS for inference, with FP8 computing power reaching 30 EFlops [42]. - The Atlas 950 SuperPlus cluster will integrate 64 Atlas 950 SuperNodes, achieving a total computing power of 524 EFlops [58]. Group 3: Strategic Positioning - Huawei acknowledges a short-term performance gap in single-chip capabilities compared to NVIDIA but aims to leverage system architecture to create supercomputers that outperform at the cluster level [5][30]. - The company emphasizes the importance of interconnect technology for large-scale supernodes, introducing the UnifiedBus interconnect protocol to enhance reliability and bandwidth [54]. - Huawei's strategy includes the development of general-purpose computing supernodes, with the TaiShan 950 supernode set to replace traditional database servers [49][50].

Core Insights - The article emphasizes the critical role of computing power in the development of artificial intelligence (AI) and highlights Huawei's advancements in this area through the launch of powerful supernodes and clusters [1][6]. Group 1: Huawei's Supernodes and Clusters - Huawei introduced the Atlas 950 SuperPoD and Atlas 960 SuperPoD, which support 8192 and 15488 Ascend cards respectively, claiming to be the world's strongest supernodes in terms of scale, total computing power, memory capacity, and interconnect bandwidth [1][2]. - The Atlas 950 SuperCluster and Atlas 960 SuperCluster were also launched, with computing power exceeding 500,000 cards and reaching one million cards, respectively, positioning them as the strongest computing clusters globally [1][2]. Group 2: Ascend Chips and AI Strategy - The Ascend chips are foundational to Huawei's AI computing strategy, with a roadmap that includes the Ascend 950 series, Ascend 960, and Ascend 970 series, with the Ascend 950PR chip expected to launch in Q1 2026 [2][6]. - The article notes that the Ascend 910C chip, part of the Atlas 900 supernode, can achieve a maximum computing power of 300 PFLOPS, maintaining its status as the largest supernode globally [2]. Group 3: Innovations in Interconnect Technology - Huawei faced significant challenges in designing the Atlas 950 and Atlas 960 supernodes, particularly in achieving long-distance, high-reliability connections and high bandwidth with low latency [4][5]. - The company implemented innovative solutions to enhance interconnect reliability by introducing high-reliability mechanisms at various protocol layers and achieving a 100-fold increase in optical interconnect reliability [5]. Group 4: Future of AI Infrastructure - The concept of supernodes is redefining AI infrastructure, enabling a new paradigm where multiple machines function as a single entity capable of learning, reasoning, and inference [3]. - Huawei's mixed supernode architecture is positioned as a solution for next-generation generative recommendation systems, supporting high-dimensional user features and low-latency inference [3].

Core Viewpoint - Huawei has made significant advancements in self-developed chips and AI infrastructure, showcasing its commitment to enhancing computing power for artificial intelligence applications in the coming years [1][3][5]. Group 1: Chip Development - Huawei plans to release multiple Ascend chips, including the 950PR, 950DT, and Ascend 960 and 970, with the 950PR set to launch in Q1 2026 featuring Huawei's self-developed HBM technology [1]. - The company has emphasized the importance of computing power as a key driver for artificial intelligence, particularly in the context of China's AI development [3]. Group 2: AI Infrastructure - The deployment of over 300 Cloud Matrix 384 supernodes marks a new norm in AI infrastructure, with Huawei focusing on a "supernode + cluster" computing solution to meet the growing demand for computing power [3]. - Huawei introduced the Atlas950 SuperPoD and Atlas 960 SuperPoD supernodes, supporting 8192 and 15488 Ascend cards respectively, positioning them as global leaders in key performance metrics [5]. Group 3: Technological Innovations - The launch of the TaiShan 950 SuperPoD marks Huawei's entry into general computing supernodes, aiming to replace various large and small machines as well as Exadata database systems [5]. - Huawei has overcome significant challenges in interconnecting large-scale supernodes, introducing the UnifiedBus protocol, with plans to open up the Lingqu 2.0 technology specifications in the future [5]. Group 4: Competitive Positioning - Despite facing challenges due to U.S. sanctions and limitations in chip manufacturing capabilities, Huawei leverages its extensive experience in connectivity technology to maintain a competitive edge in supernode capabilities [6].

Core Viewpoint - Computing power is crucial for artificial intelligence and will continue to be a key factor for China's AI development, as stated by Huawei's rotating chairman Xu Zhijun [3]. Group 1: Huawei's Chip Development - Huawei plans to launch the Ascend 950PR chip in Q1 2026, the Ascend 950DT in Q4 2026, the Ascend 960 chip in Q4 2027, and the Ascend 970 chip in Q4 2028 [3]. - The company has introduced the Atlas 900 A3 SuperPoD, with over 300 units deployed, serving more than 20 customers, and will release the Atlas 950 SuperPoD in Q4 2026, supporting 8192 Ascend cards with a computing power of 8 EFLOPS FP8 / 16 EFLOPS FP4 [5]. Group 2: Supernode Infrastructure - Huawei is developing the Atlas 960 SuperPoD, expected to support 15488 Ascend cards with a computing power of 30 EFLOPS FP8 / 60 EFLOPS FP4, planned for Q4 2027 [7]. - The company announced the Atlas 950 SuperCluster and Atlas 960 SuperCluster, with computing power exceeding 500,000 cards and reaching one million cards, respectively [9]. Group 3: General Computing and Open Technology - Huawei revealed the roadmap for the Kunpeng chip, with the Kunpeng 950 chip set to launch in Q4 2026 and the Kunpeng 960 chip in Q1 2028 [9]. - The company introduced the TaiShan950 SuperPoD, the world's first general computing supernode, expected to launch in Q1 2026, which aims to replace various large and small machines as well as Exadata database integrated machines [9]. - Xu Zhijun announced that the supernode technology will be fully open, including the Lingqu 2.0 technology specifications, which encompass foundational, firmware, and operating system reference design specifications [9].

Core Insights - Huawei has unveiled multiple self-developed chip advancements at the 2025 Global Connectivity Conference, including the launch timeline for the Ascend 950PR chip in Q1 2026, which utilizes Huawei's self-developed HBM technology [2] - The concept of "super nodes" has emerged as a new norm in AI infrastructure, with over 300 Cloud Matrix 384 super nodes deployed, emphasizing the critical role of computing power in the future of artificial intelligence in China [4][5] Group 1 - Huawei's newly released Atlas950 SuperPoD and Atlas 960 SuperPoD super nodes support 8,192 and 15,488 Ascend cards respectively, showcasing global leadership in key metrics such as card scale, total computing power, memory capacity, and interconnect bandwidth [7] - The company has also introduced super node clusters, Atlas950 SuperCluster and Atlas960 SuperCluster, with computing power exceeding 500,000 cards and reaching one million cards [7] - Huawei is confident in providing sustainable and ample computing power for the long-term rapid development of artificial intelligence, leveraging the strongest computing power globally through super nodes and clusters [8] Group 2 - Huawei has pioneered the introduction of super node technology into general computing, launching the world's first general computing super node, TaiShan 950 SuperPoD, which aims to replace various large and small machines as well as Exadata database integrated machines [8] - The company has overcome significant challenges in interconnect technology for large-scale super nodes, introducing the UnifiedBus interconnect protocol, with plans to open the Lingqu 2.0 technical specifications in the future [8] - Despite facing restrictions from U.S. sanctions that limit access to TSMC for chip production, Huawei emphasizes its strengths in connectivity technology, which has been developed over 30 years, enabling the achievement of super nodes with tens of thousands of cards [8]

Core Viewpoint - Huawei has made significant advancements in self-developed chips, planning to release multiple models in the coming years, emphasizing the importance of computing power for AI development in China [3][5]. Group 1: Chip Development - Huawei plans to launch several Ascend chips, including the 950PR in Q1 2026, utilizing self-developed HBM technology [3]. - The company has introduced the Atlas950 SuperPoD and Atlas960 SuperPoD, which support 8192 and 15488 Ascend cards respectively, showcasing global leadership in key performance metrics [7]. Group 2: AI Infrastructure - The concept of "super nodes" has become a new norm in AI infrastructure, with over 300 Cloud Matrix 384 super nodes deployed [5]. - Huawei's super node and cluster solutions, such as Atlas950 SuperCluster and Atlas960 SuperCluster, exceed 500,000 and 1 million cards in computing scale, respectively [7]. Group 3: Connectivity and Collaboration - Huawei has made breakthroughs in large-scale super node interconnection technology, introducing the UnifiedBus protocol, which will be open-sourced in the future [7]. - The company aims to build a robust foundation for AI computing needs globally, leveraging over 30 years of experience in connectivity technology [8].

Core Insights - Huawei's Ascend AI Summit 2025 will take place in Shanghai on September 18, 2025, focusing on the latest technological achievements and innovations in Ascend's hardware, software, and industry solutions, along with sharing successful business cases and insights from partners [1] Group 1: New Technologies - The summit will present significant advancements in Ascend's foundational hardware, software, and industry solutions, showcasing breakthroughs in ultra-node technology, comprehensive open-source software, and innovations across the entire technology chain including pre-training, post-training, and inference solutions [2] Group 2: New Directions - The summit will introduce the "Ultra-Node Development Report," which defines ultra-nodes and outlines their core characteristics, analyzing global industry evolution, stability challenges, and technological ecosystem patterns to provide clear directions for industry development [3] - Practical applications of Ascend ultra-nodes in various sectors such as internet, telecommunications, finance, energy, and manufacturing will be shared, with guest speakers from Midea discussing real-world applications and value [3] Group 3: New Value - The summit will highlight large-scale expert parallel practices across six major industries including telecommunications, government, education, finance, large models, and electricity, with a focus on innovative practices in digital finance by Postal Savings Bank [4] - New products including a large model application integrated machine and 2025 Ascend hardware partner products will be showcased, demonstrating the latest achievements and innovative directions in the Ascend ecosystem [4] - An Ascend MVP appointment ceremony will take place, inviting top industry experts and scholars to provide professional guidance and advice for the development of Ascend AI [4] Group 4: Event Highlights - The Huawei Connect 2025 will feature three major Ascend-themed forums focusing on open-source frameworks and intelligent cluster operations, showcasing how Ascend technology drives AI computing towards greater efficiency and openness [5] - The exhibition area will present the latest Ascend products and solutions, interactive demos, industry case studies, and open speeches, allowing attendees to experience advanced AI technologies and their business applications firsthand [5]

9月13日上午，以"元生有AI 万物盛开"为主题的2025年服贸会"数字开物·奇点π对"主题沙龙活动在北京 成功举办。来自中国信通院人工智能研究所、来画科技、思必驰、浩云长盛、优刻得、曙光数创、 ByteBridge、沨呵智慧、算力自由等业内专家与贯穿AI与算力全景生态的优秀企业代表出席活动并做精 彩分享。 活动现场，与会嘉宾共同探讨了大模型技术的演进趋势、包括AI陪伴、AI办公等AI应用的落地实践， 以及智算中心的基础设施创新、智算中心的热管理革命、GPU集群的精益化管理、服务于中小创业者的 AI算力云发展现状以及AI浪潮下全产业链的协同发展路径等前沿议题。本场活动由中国国际服务贸易 交易会、2025年服贸会ICT展指导，首都会展集团、数字开物主办，科智传媒承办。 01 前瞻洞察：大模型技术持续演进，Agent应用进入规模落地阶段 在主旨演讲环节，中国信通院人工智能研究所平台与工程化部主任曹峰带来题为《大模型技术和应用发 展现状及趋势》的分享。 聚焦AI应用，北京来画科技文化有限公司董事、合伙人白宇进行了《每个人的AI伙伴，无边界的沟通 桥梁》的分享。 白宇表示，当前AI应用无需执着于自建大模型，利用开源模 ...