如果你想通过官网查看国产AI芯片公司的最新产品信息，恐怕要失望而归了：最新的芯片型号时常并 不会出现在产品菜单上，更不用说芯片的具体参数信息。 这是国产AI芯片公司过去几年力求低调的缩影。"这几年做芯片比较敏感，我们出来讲得也比较少。"一 位与大厂关联密切的芯片公司负责人坦言。 但近期，随着华为、百度等大厂接连公开宣布AI芯片的迭代路线图，阿里的平头哥芯片亮相央视《新 闻联播》，闷声做事的状态突然间被打破。未来三年，华为将发布四款昇腾AI芯片；按照百度的规 划，接下来两年有两款昆仑芯AI芯片即将上市。与百度仅披露芯片型号与用途相比，华为给出的信息 更为翔实，将互联带宽、算力、内存等外界关注的芯片参数悉数公开。 华为打响第一枪，百度最新接棒 在国产AI芯片阵营中，华为昇腾的头号地位毋庸置疑。以至于英伟达创始人黄仁勋也不得不将华为视 作"强大的竞争对手"，直言"低估华为这样的竞争对手是愚蠢的"。 华为在2018年发布首款昇腾AI芯片310，2019年迭代至昇腾910，其性能超过英伟达同时代的V100 GPU。但随着2020年美国制裁、台积电断供，华为自研AI芯片步入艰难时刻。从2020年开始的约五年 时间内，华为 ...

Core Insights - The openEuler Summit 2025, themed "Intelligent Leap Beyond Boundaries," was successfully held in Beijing, focusing on fostering innovation in operating systems and accelerating the development of the global open-source software ecosystem [1] Group 1: Community Growth and Development - The openEuler community has grown significantly, with over 2,100 member organizations and more than 23,000 global contributors, serving over 5.5 million users [1] - The cumulative installation of openEuler operating systems is expected to exceed 16 million by the end of 2025, maintaining a leading market share across various sectors including internet, communication, government, finance, public utilities, and energy [1] Group 2: Technological Advancements - The release of openEuler 24.03 LTS SP3 marks the beginning of a new five-year development path, emphasizing the embrace of supernodes and AI technologies [2] - Key features of the supernode-oriented operating system include global resource abstraction, unified memory addressing, and device pool management, aimed at enhancing computational efficiency [2] Group 3: Strategic Collaborations - Huawei, as a strategic donor and contributor, is collaborating with the openEuler community to develop supernode-supporting operating system plugins and deepen partnerships with global open-source organizations [3] - A joint development initiative was announced between the OpenAtom Foundation, Huawei, and China Telecom Tianyi Cloud to leverage their technological strengths and promote the openEuler ecosystem [5][6] Group 4: AI Solutions and Industrial Applications - The openEuler community introduced the Intelligence BooM full-stack open-source AI solution, which supports over 50 model fine-tunings and improves inference efficiency by 10%-30% [4] - The community is evolving embedded capabilities for new industrial automation, achieving microsecond-level response times in various commercial applications [4] Group 5: Globalization and Ecosystem Building - The openEuler community has established partnerships with global organizations like AMD, Arm, and Linaro to enhance the AI industry ecosystem and support openEuler's global expansion [6][7] - The establishment of the academic and educational committee aims to strengthen ties with academia and promote talent development in the system software field [7][8]

Core Insights - The article discusses the significance of operating systems in both consumer technology and industry digitalization, highlighting the rise of open-source operating systems like openEuler and OpenAnolis in China [1][2] - openEuler has rapidly gained market share, reaching approximately 50% in China last year and is expected to exceed 55% this year, positioning itself as the largest and fastest-growing open-source OS community in the country [1][2] - The development of openEuler is part of a broader trend in China's IT industry, which is forming a complete IT infrastructure stack, although it still lags behind global leaders [2] Industry Developments - openEuler originated from Huawei's self-developed server OS, EulerOS, and was officially open-sourced in late 2019, transitioning to community governance under the OpenAtom Foundation [2] - The openEuler community has become one of the most active open-source OS communities globally, with over 2,100 member organizations and more than 23,000 developers contributing [3] - The community's global download count has surpassed 5.5 million, indicating strong engagement and growth [3] Future Directions - AI is identified as a key focus for openEuler in the next five years, with 22 out of 25 newly incubated open-source projects being AI-related [5] - The concept of "super nodes" is emerging as a new norm in AI infrastructure, necessitating significant changes in operating systems to support advanced computing needs [6] - openEuler has released the world's first operating system designed for super nodes, with a planned community launch on December 30, 2025 [6] Strategic Collaborations - Huawei is collaborating with the openEuler community to enhance the operating system's capabilities, including support for super nodes and key features like unified memory addressing and low-latency communication [7] - Major chip manufacturers, including Intel, Arm, and AMD, have joined the openEuler community as donors, indicating strong industry support [10] - The community aims to deepen cooperation with global open-source organizations in AI and data, enhancing its international presence and solutions [10][11] Globalization Efforts - The openEuler community emphasizes a systematic approach to globalization, aiming to establish a robust framework for international collaboration rather than competing in isolation [11] - The strategy includes focusing on regions like the "Belt and Road" initiative to enhance global outreach and partnerships [11]

Core Insights - The openEuler community announced the upcoming release of the openEuler 24.03 LTS SP3 operating system, designed for AI computing infrastructure, featuring unified memory addressing, low-latency communication for heterogeneous computing, and global resource pooling [1] - The community currently has over 2,100 member organizations, more than 23,000 global contributors, and over 5.5 million users, with an expected cumulative installation of over 16 million systems by the end of 2025 [1] - Huawei has contributed support for the supernode operating system plugin code and has opened the Lingqu Interconnection Protocol 2.0, while also collaborating with Tianyi Cloud to deepen cooperation in AI and supernode fields [1] - Major chip companies including Intel, Arm, and AMD have joined the community, which has established partnerships with 15 global open-source organizations [1] Industry Applications - The openEuler embedded solution has been implemented in companies such as China Southern Power Grid and Phoenix, achieving microsecond-level response times [2] - The conference announced 23 leading commercial practice projects, covering both large-scale and innovative categories [2]

Investment Rating - The report maintains a "Recommended" rating for the computer industry [1] Core Insights - The report highlights the increasing consensus on supernodes as a foundational infrastructure for AI, driven by significant capital expenditures from major cloud service providers (CSPs) and advancements in GPU/ASIC technologies [5][8] - OpenAI has secured substantial power agreements totaling over $1 trillion, indicating a robust demand for AI computing power [5][14] - The report emphasizes the growth potential in various segments including servers, liquid cooling, power supply, and storage, with clear growth momentum identified [7][8] Summary by Sections Demand Side - CSPs are raising their capital expenditure forecasts, with Google increasing its 2025 capex to $91-93 billion, Meta to $70-72 billion, and Amazon to $125 billion, primarily for AI infrastructure [5][29] - OpenAI has signed significant power agreements with NVIDIA, AMD, and Broadcom, totaling 26GW and over $1.1 trillion in value [5][15] - Sovereign AI investments are projected to reach $1 trillion over the next five years, with major projects in the US, EU, and Saudi Arabia [5][25] Supply Side - The report notes the continuous iteration of chips, with supernodes becoming a consensus in AI infrastructure, as evidenced by new product launches from various manufacturers [34][5] - NVIDIA's upcoming GPU architectures are expected to enhance computational capabilities significantly, with the Blackwell Ultra and Rubin architectures set to launch in the coming years [36][39] - Major companies like AMD and Huawei are also advancing their AI chip offerings, with AMD's Helios solution and Huawei's Atlas series expected to drive further innovation [41][44] Growth Segments - The server market is experiencing substantial growth, with companies like Hon Hai and Wistron achieving significant increases in AI server shipments [7] - Liquid cooling technology is becoming essential in AI data centers, with companies reporting high double-digit revenue growth [7] - The storage chip industry is entering a period of severe shortage, driving up DRAM prices and increasing demand from CSPs [7] Multi-modal and Application Ecosystem - The report outlines the rapid evolution of AI models, with major updates from OpenAI and Alibaba, and a significant increase in token usage across platforms [7][8] - OpenAI's new applications and models are enhancing its ecosystem, with a notable increase in daily token usage [7][8]

Core Insights - The openEuler operating system is positioned to lead the development of foundational software in China, particularly in the AI era, with a focus on supernodes and global expansion [1][5][6] - By the end of 2025, openEuler is expected to surpass 16 million installations, maintaining a leading market share in key industries [1][5] - The community has established partnerships with 15 global open-source organizations, enhancing its international presence and collaboration [7][8] Group 1: Technological Advancements - The openEuler operating system supports advanced AI capabilities, enabling robots to perform complex tasks through powerful computational support [2][3] - The integration of openEuler with various hardware has improved adaptability, allowing for "plug-and-play" functionality, which enhances efficiency in AI applications across multiple sectors [3][5] - The system aims to create a unified ecosystem for robotic operations, facilitating seamless interaction between hardware and software [2][3] Group 2: Market Position and Strategy - openEuler has become a cornerstone for digital transformation across various industries, with a projected increase in market share for the current year [5][6] - The community has grown to include 2,100 member organizations and 23,000 developers, indicating significant scalability in government, finance, and energy sectors [5][8] - The strategic focus on global collaboration and innovation aims to position openEuler as a leader in the international open-source community [7][8] Group 3: Future Outlook - The next five years are critical for openEuler as it seeks to embrace supernodes and AI, aiming to provide a robust software foundation for the future [1][5][6] - The transition from a survival strategy to a leadership role reflects a significant shift in China's foundational software industry, emphasizing proactive engagement in global markets [6][8] - The community's commitment to open collaboration and technological advancement is expected to drive China's progress from a digital power to a digital stronghold [8]

Core Viewpoint - The conference on operating systems, themed "Intelligent Leap Without Boundaries, Open Source for a Better Future," successfully gathered industry leaders to promote the development of the openEuler operating system and accelerate the global open-source software ecosystem [2]. Group 1: Development and Growth of openEuler - The openEuler community has grown significantly over the past six years, with over 2,100 member organizations and more than 23,000 global contributors, serving over 5.5 million users [2]. - The cumulative installation of openEuler is expected to exceed 16 million sets by the end of 2025, establishing it as the preferred operating system for digital transformation in various industries [2]. - The community is set to embark on a new five-year development path, launching an operating system tailored for supernodes by the end of 2025, aiming to lead in the AI era and enhance globalization efforts [2][12]. Group 2: Strategic Importance of Basic Software - Academician Ni Guangnan emphasized the strategic nature of basic software, advocating for independent innovation, collaborative ecosystem building, and sustained long-term investment [3]. - The transition to supernodes is recognized as a mainstream trend in computing infrastructure, with operating systems playing a crucial role in connecting hardware and applications in the intelligent era [3]. Group 3: Collaboration and Ecosystem Building - The core of open-source is collaboration, and the future of the ecosystem relies on co-creation and sharing among hardware partners, software vendors, and global developers [5]. - Huawei's CEO highlighted the rapid transformation brought by AI technologies and the need for operating systems that can support supernodes, contributing to the openEuler community with key capabilities [6][10]. Group 4: Technological Innovations and Solutions - The openEuler community has introduced the Intelligence BooM full-stack open-source AI solution, enhancing inference efficiency by 10% to 30% through heterogeneous collaboration [16]. - In the new industrial automation sector, openEuler has evolved its embedded capabilities, achieving microsecond response times and successfully implemented in various well-known enterprises [16]. Group 5: Globalization Efforts - New donors to the openEuler community include major chip manufacturers like AMD, further strengthening the community's resources [18]. - The community has established deep technical cooperation with 15 global open-source organizations in areas such as AI, cloud computing, and embedded systems, enhancing its global presence [20].

Core Insights - The openEuler community has officially launched a new five-year development plan aimed at creating an operating system for supernodes by the end of 2025, focusing on AI and global expansion [2] - The openEuler community has grown significantly over the past six years, with over 2,100 member organizations, more than 23,000 global contributors, and a user base exceeding 5.5 million [2] - The cumulative installation of openEuler operating systems is expected to surpass 16 million units by the end of 2025, with applications in various sectors including internet, telecommunications, government, finance, public utilities, and energy [2] Collaboration and Partnerships - A joint development initiative was announced by the OpenAtom Open Source Foundation, Huawei, and China Telecom's Tianyi Cloud to promote technology co-construction, ecosystem building, and business collaboration in the fields of AI and supernodes [2] - New donors to the openEuler community include AMD, Inspur Cloud, and Digital China, joining major chip companies like Intel and Arm [3] - The openEuler community has established deep technical cooperation with global open-source organizations such as the Zephyr Embedded Technology Foundation and the Linux Foundation AI & Data Foundation, collaborating with 15 global open-source organizations in areas like AI, cloud, big data, HPC, and embedded systems [3]

Investment Rating - The industry investment rating is "Overweight" and has been maintained from the previous rating [5] Core Insights - The rise of Scale-up architecture is driven by the increasing demand for high-speed interconnects due to the growth of large model parameters and the need for unified computing units formed by connecting multiple GPUs [1][13] - The Scale-up switch market is expected to grow rapidly, with a projected market size of nearly $6 billion by 2025 and a 5-year CAGR of 26% from 2025 to 2030 [1][68] Summary by Sections Scale-up Supernodes: Breaking Through Single Card Computing Bottlenecks - The demand for larger parameter scales in large language models necessitates the expansion of computing clusters, which can significantly enhance model output accuracy and performance [13] - Cross-node communication requirements are heightened due to the use of tensor parallelism and expert parallelism technologies [17] Overview of Mainstream Supernode Systems - NVIDIA's GB200 NVL72 supernode connects 36 Grace CPUs and 72 NVIDIA Blackwell GPUs using NVLink technology, achieving high bandwidth and low latency [31][32] - Google's seventh-generation TPU Ironwood can scale up to 9216 chips, achieving peak performance of 4614 TFLOPS [35][36] - Amazon's Trainium2 server interconnects 64 chips through NeuronLink, ensuring low latency and high bandwidth communication [45][48] - Huawei's CloudMatrix384 supernode is designed for large-scale AI workloads, achieving 300 PFLOPS of BF16 computing power [49][52] Beneficiaries of Scale-up Network Demand - High-end switching chips and interface chips are expected to see increased demand due to the customized networks required for Scale-up architecture [3] - The integration of hardware and software in supernode systems is likely to enhance the value-added services of hardware manufacturers [56] - The demand for short-distance high-speed copper connections is strong, while optical interconnects are gradually being integrated into Scale-up networks [3] Focus Areas - Key companies to watch include: - Interconnect & Switching Chips: Sanken Communication, Lattice Semiconductor, and Wantong Development [4] - Network Equipment: ZTE, Ruijie Networks, and Unisplendour [4] - Optical Interconnect: Zhongji Xuchuang, New Easy, and Tianfu Communication [4] - Copper Interconnect: Ruikeda, Shenyu Co., and Wolker Materials [4]

Core Insights - The article discusses the current structure of the AI industry, highlighting a "pyramid" model where the majority of value is captured at the chip level, while models and applications capture significantly less value. The ideal structure proposed is an "inverted pyramid" where models generate ten times the value of chips, and applications generate one hundred times the value of models [2][14]. Company Developments - Baidu has launched two new chips, Kunlun M100 and Kunlun M800, aimed at enhancing performance in large-scale inference and training of multimodal models. The M100 is set to be released in early next year, while the M800 is a high-performance version of the M100 [5][6]. - Baidu's AI cloud business has shown significant growth, with non-online marketing revenue surpassing 10 billion yuan, reflecting a 34% increase, despite a slight decline in overall revenue due to weak advertising performance [17]. Technological Innovations - Baidu introduced "Tianchi 256 Super Node" and "Tianchi 512 Super Node," capable of supporting trillion-parameter model training. The 512 Super Node boasts a 95% performance improvement per card and an eightfold increase in single-instance inference performance [9][10]. - The emergence of super nodes is a response to the increasing size of models and the limitations of traditional server interconnects. Baidu's approach to super nodes indicates a strong underlying capability in chip design and communication technology [10][13]. Industry Context - The article references a Morgan Stanley estimate predicting that the generative AI industry will generate approximately $153 billion in total revenue by 2025, with Nvidia projected to account for $130.5 billion of that revenue [15]. - The current "pyramid" structure of the AI industry is acknowledged, but there are signs of a shift in value distribution, with companies like Baidu actively working to enhance their AI capabilities across the entire value chain [17][20].