来 源 ： 内容来自nextplatform 。 用于 AI 训练和推理的集群架构正在推动数据中心基础设施支出空前的增长，但它们也对 HPC （高性能计算）架构产生了反射性的有益影响。这得益于为 AI 项目获取资金相对容易，以及 升级现有 HPC 系统以进行传统模拟和建模的需求。 公众号记得加星标⭐️，第一时间看推送不会错过。 本 周 ， SC25 超 级 计 算 大 会 的 第 一 天 以 Hyperion Research 公 司 的 传 统 早 餐 会 拉 开 序 幕 。 Hyperion Research 刚刚完成了 2024 年 HPC 市场的考察，目前正在整理 2025 年的数据，并 更新其对未来直到本十年末的预测。Earl Joseph 和 Mark Nossokoff 介绍了传统的 HPC 市场 及其人工智能增强（AI augmentation），以及它的本地部署（on-premises）和云部署模型； Bob Sorensen 则对量子计算市场进行了深入分析。我们稍后将单独跟进量子计算分析，现在先 聚焦于结合了 AI 增强的 HPC 市场和传统 HPC 市场。 从最高层面来看，根据 Hype ...

内容概况：量子科技作为事关国家信息安全的战略新兴领域，近年来在国家战略指引下，量子科技行业 市场规模增长迅速，同时随着量子科技的不断突破，应用场景逐渐向纵深拓展，量子科技产业呈现出较 好的发展势头。数据显示，中国量子科技行业市场规模从2015年的45.23亿元增长至2024年的131.04亿 元，年复合增长率为12.55%。2024年，量子科技在全球范围内迎来了重大突破，研究已经进入从理论 到应用的过渡期，多个量子计算平台取得了具有里程碑意义的新成果。量子比特的数量和质量得到大幅 提升；量子算力已经应用于计算流体动力学、金融、生物医药等领域。量子通信干线已经实现量子保密 通信的实际应用，不仅为金融、电力等行业提供了加密通信的基础设施，也为未来的量子互联网奠定了 基础。2025年3月，我国成功研制出自主创新的量子计算原型机"祖冲之三号"，这一重大突破使得我国 在超导体系量子计算领域跃居全球前列。在通信领域，我国科研团队近期取得了显著进展，成功构建了 300公里级的量子直接通信网络，这一里程碑式的成就验证了长距离高安全通信的可行性。量子科技发 展具有重大科学意义和战略价值，是一项对传统技术体系产生冲击、进行重构的 ...

Core Insights - The AI computing industry is transitioning into an "application dividend period," driven by the integration of artificial intelligence and high-performance computing, with a projected global market size of $234 billion in 2024 and expected to exceed $2.75 trillion by 2032, reflecting a compound annual growth rate of 36.8% [5][6][10] AI Computing Industry Overview - AI computing (Intelligent Computing) combines AI with high-performance computing, utilizing dedicated hardware and distributed architecture to provide efficient and scalable computing power for AI tasks [2][3] - The industry is characterized by a three-pronged model of "computing power + algorithms + data," which supports complex AI model training and inference, making it a critical infrastructure for the digital economy [2][3] Global and Chinese AI Computing Development - The global AI market is evolving from a "model dividend period" to an "application dividend period," becoming a core engine for the digital economy, with significant advancements in computing power architecture [5][6] - In 2024, the total global computing power is expected to reach 2207 EFLOPS, with intelligent computing power contributing 1610 EFLOPS, marking a year-on-year growth of 63.8% [5][6] - China's total computing power is projected to reach 280 EFLOPS in 2024, with intelligent computing power at 90 EFLOPS, indicating a significant gap between supply and future demand [6][10] AI Computing Industry Chain in China - The AI computing industry chain in China consists of upstream hardware (GPUs, NPUs), midstream service providers (telecom operators, cloud service providers), and downstream application sectors (internet, finance, manufacturing) [8][10] - The market for AI acceleration chips is expected to grow from 17.56 billion yuan in 2020 to 142.54 billion yuan in 2024, with a compound annual growth rate of 68.8% [9][10] Competitive Landscape - Major players in the infrastructure layer include Huawei, which leads in domestic chip production, and Inspur, which dominates the AI server market [10] - Telecom operators and cloud service providers, such as China Telecom, Alibaba Cloud, and Tencent Cloud, are key players in the computing service layer [10] Future Trends in AI Computing Industry - The industry is expected to evolve towards collaborative evolution, with a focus on efficient resource utilization and deep integration with the real economy [11][12] - The emphasis will shift from model performance to creating measurable business value through large-scale applications in key sectors such as manufacturing, finance, and healthcare [13]

Market Overview - The Shanghai Composite Index fell by 2.45% on November 21, with the largest declines in the comprehensive and non-ferrous metal sectors, down 5.50% and 5.26% respectively [1] - The computer industry also experienced a decline of 3.01% [1] - Overall, there was a net outflow of 128.99 billion yuan in the main funds across the two markets, with only the media sector seeing a net inflow of 1.48 billion yuan [1] Computer Industry Analysis - The computer industry saw a net outflow of 11.448 billion yuan, with 336 stocks in the sector; 37 stocks rose, including 3 that hit the daily limit, while 299 stocks fell, with 6 hitting the lower limit [2] - Among the stocks with net inflows, Rongji Software led with a net inflow of 267 million yuan, followed by Jiuqi Software and Wanxing Technology with inflows of 240 million yuan and 198 million yuan respectively [2] - The stocks with the largest net outflows included Huasheng Tiancheng, Zhongke Shuguang, and Haixia Innovation, with outflows of 762 million yuan, 475 million yuan, and 426 million yuan respectively [2] Top Gainers in Computer Sector - The top gainers in the computer sector included: - Rongji Software: +9.98%, turnover rate 39.35%, main fund flow 266.57 million yuan - Jiuqi Software: +9.96%, turnover rate 17.55%, main fund flow 240.39 million yuan - Wanxing Technology: +6.18%, turnover rate 22.44%, main fund flow 198.01 million yuan [2] Top Losers in Computer Sector - The top losers in the computer sector included: - Huasheng Tiancheng: -9.99%, turnover rate 26.70%, main fund flow -761.64 million yuan - Zhongke Shuguang: -4.19%, turnover rate 2.23%, main fund flow -474.65 million yuan - Haixia Innovation: -20.01%, turnover rate 34.07%, main fund flow -425.94 million yuan [3]

Core Insights - The integration of artificial intelligence with the real economy is accelerating, with computing power being a crucial component of new productive forces, driving industrial upgrades [1][6] - Domestic companies are adopting a dual-track development model of "open collaboration" and "full-stack ecosystem" to break single technology dependencies and build a diverse technology-driven ecosystem [1][7] Group 1: Technological Breakthroughs - The Chinese government's 20th Central Committee has set a clear direction for domestic computing power development, emphasizing the need for innovation in AI and digital technologies [1] - At the Changsha World Computing Conference, the scaleX640 supernode showcased by Inspur demonstrated significant advancements in computing power, achieving a 30%-40% performance improvement for trillion-parameter model training compared to traditional solutions [3][6] Group 2: Market Demand and Growth - The global demand for computing power is surging, with the AI computing power market expected to reach 141.98 billion yuan in 2024 and nearly 1.344 trillion yuan by 2031, reflecting a compound annual growth rate of 37.3% [6] - Traditional computing architectures face challenges such as insufficient computing power, slow data transmission, high energy costs, and instability, which are significant barriers to digital transformation for many companies [6] Group 3: Industry Strategies - Leading domestic companies are exploring differentiated paths to address the growing computing power demand and systemic technological bottlenecks, with Inspur's open architecture and Huawei's full-stack solution representing two distinct approaches [7] - The collaboration between Inspur and over 20 upstream and downstream companies aims to establish an open and inclusive AI computing industry ecosystem, highlighting the trend towards diversified technological routes in the market [7]

Group 1 - The A-share market experienced a decline, with the Shanghai Composite Index down by 2.05%, the Shenzhen Component Index down by 2.80%, and the ChiNext Index down by 3.31% as of 14:13 on November 21 [1] - Lenovo Group announced plans to launch a personal super intelligent agent at the Lenovo Innovation Technology Conference next year, which will learn user habits and predict intentions to enhance user experience [1] - Lenovo's CEO Yang Yuanqing expressed confidence that the market will not experience an AI bubble, predicting a shift towards the popularization of AI technology, making it accessible to a broader audience [1] Group 2 - Huatai Securities noted that generative AI is entering a new development stage dominated by AI agents, with a trajectory of "first B2B, then B2C, and finally terminal" [2] - The AI ETF (515070) tracks the CS AI Theme Index (930713), focusing on companies providing technology, foundational resources, and applications in the AI sector, often referred to as the "robot brain" creators [2] - The top ten weighted stocks in the AI ETF include leading domestic technology companies such as Zhongji Xuchuang, Xinyisheng, and Hikvision, indicating a strong presence in the AI industry [2]

Core Viewpoint - Zhongke Shuguang's stock performance shows a slight decline with significant financing activities indicating high market interest and potential volatility [1][2]. Financing Activities - On November 20, Zhongke Shuguang experienced a financing buy-in of 258 million yuan, with a net financing outflow of 136 million yuan, resulting in a total financing balance of 9.551 billion yuan [1]. - The financing balance of 9.533 billion yuan accounts for 6.52% of the circulating market value, indicating a high level compared to the past year [1]. - The company also reported a securities lending activity with 4,600 shares sold and a remaining balance of 18,000 shares, reflecting a securities lending balance of 17.99 million yuan, which is also at a high level [1]. Company Overview - Zhongke Shuguang, established on March 7, 2006, and listed on November 6, 2014, specializes in high-performance computing, general servers, and storage products, with 88.79% of its revenue coming from IT equipment [1]. - The company reported a revenue of 8.82 billion yuan for the first nine months of 2025, marking a year-on-year growth of 9.68%, and a net profit of 966 million yuan, reflecting a 25.55% increase [2]. Shareholder Information - As of October 10, 2025, the number of shareholders decreased to 323,500, while the average number of circulating shares per person increased to 4,521 [2]. - Cumulative cash dividends since the A-share listing amount to 1.922 billion yuan, with 1.083 billion yuan distributed over the last three years [3]. - Major institutional shareholders include Huatai-PB CSI 300 ETF and E Fund CSI 300 ETF, with notable changes in their holdings [3].

Core Insights - Computing power has become the core "new productive force" in the digital economy era, driving the development of advanced technologies [1][3] - The computing power industry needs to focus on the collaborative advancement of construction, application, and ecology to build a "three-dimensional computing" system, avoiding redundant construction and resource waste [1] Industry Developments - With the arrival of the artificial intelligence era, computing power is deeply integrated into daily life, supporting various applications from mobile usage to online browsing and gaming [3] - The company has established one of the earliest high-standard intelligent computing centers in Changsha, which provides a controllable computing foundation for local AI applications and ensures the stable operation of domestic AI software [3] - The introduction of super-node technology will help overcome performance bottlenecks in key industrial software, promoting continuous technological iteration and application deepening [3] Company Initiatives - The company is leveraging Changsha's talent advantages and comprehensive competitiveness to support breakthroughs in the world's first single cabinet-level 640-card super-node system, scaleX640 [3] - Future applications of next-generation AI technology will focus on the industrial sector, with the company aiming to provide robust technical support for high-end manufacturing by utilizing Changsha's rich application scenarios [3]

Core Insights - The World Computing Conference held in Changsha showcased the ScaleX640 super node by Inspur, which is the world's first single-rack 640-card super node, achieving a 20-fold increase in computing density and supporting million-card-level cluster expansion, thus injecting strong domestic computing power into the development of new productivity [1][3]. Group 1: Product Features - The ScaleX640 super node consists of a single immersion liquid cooling heat exchange module and two immersion liquid cooling dedicated computer cabinets, representing a significant breakthrough as it surpasses traditional solutions that would require multiple cabinets to achieve similar computing power [3]. - The device utilizes immersion phase change liquid cooling technology, allowing it to maintain a stable operating temperature below 35 degrees Celsius, which is lower than typical gaming PCs, and achieves a Power Usage Effectiveness (PUE) of 1.04, meaning 96% of electricity is used for computing, resulting in over 30% energy savings compared to traditional air cooling solutions [4]. Group 2: Industry Impact - The ScaleX640 super node adheres to the "AI computing open architecture" concept, initiated by Inspur in collaboration with over 20 industry partners, aiming to build an open and collaborative innovation system to overcome challenges in domestic AI development, such as insufficient software-hardware synergy and fragmented ecosystems [5]. - The super node supports multi-brand acceleration cards and is compatible with mainstream computing ecosystems, enabling users to flexibly select domestic computing power for rapid deployment of trillion-parameter model training and high-throughput inference, effectively addressing the practical pain points of application adaptation and software-hardware disconnection in domestic intelligent computing [5].

Core Viewpoint - Zhongke Shuguang's Q3 2025 report shows impressive performance with steady revenue growth and a significant year-on-year increase in net profit excluding non-recurring items [1] Group 1: Financial Performance - The company reported robust revenue growth and a substantial increase in net profit year-on-year [1] - The non-recurring net profit showed a high growth rate compared to the previous year [1] Group 2: Product Development - Zhongke Shuguang recently launched the scaleX640 super node, which significantly enhances single cabinet computing power density [1] - The new product supports multiple brand acceleration cards and mainstream computing ecosystems, potentially establishing a new paradigm for domestic intelligent computing [1] Group 3: Strategic Initiatives - The company is actively advancing the merger with Haiguang Information, aiming to create a leading domestic computing power industry chain [1] - The goal is to achieve deep integration and synergy from chips to services, which is expected to significantly enhance core competitiveness [1] Group 4: Investment Rating - The company maintains a "buy" rating based on its strong performance and strategic initiatives [1]