Core Viewpoint - The company reported strong financial performance for the first three quarters of 2025, with significant growth in both revenue and net profit, indicating a positive trend in its operations and market position [1][2]. Financial Performance - For the first three quarters of 2025, the company achieved operating revenue of 8.82 billion yuan, a year-on-year increase of 9.68% [1][2]. - The net profit attributable to shareholders reached 966 million yuan, reflecting a year-on-year growth of 25.55% [1][2]. - The non-recurring net profit attributable to shareholders was 757 million yuan, showing a substantial year-on-year increase of 70.17% [1][2]. - Quarterly breakdown shows revenue growth of 4.34%, 0.94%, and 27.51% for Q1, Q2, and Q3 respectively, with Q3 showing accelerated growth [2]. - Quarterly net profit attributable to shareholders was 186 million yuan, 542 million yuan, and 237 million yuan for Q1, Q2, and Q3 respectively, with Q3 growth at 15.07% [2]. Product Development - The company launched the scaleX640 super node, which is the world's first single-cabinet 640-card super node, during the World Internet Conference in November 2025 [3][4]. - The scaleX640 features a "one-to-two" high-density architecture, achieving high-speed interconnection and significantly enhancing computing power density by 20 times compared to traditional solutions [3][4]. - The product supports advanced AI applications, including training and inference for trillion-parameter models, and is positioned to capitalize on the growing domestic AI computing market [3][4]. Industry Positioning - The company is actively addressing challenges in the domestic AI computing landscape, such as high-end chip shortages and lack of standards, by collaborating with over 20 industry players to establish an "AI Computing Open Architecture" [4]. - The launch of the scaleX640 is seen as a strategic move to leverage opportunities in the domestic AI computing industry, which is expected to grow rapidly due to increased capital expenditure from major players like Alibaba and Tencent [3][4]. Earnings Forecast - The company adjusted its profit forecasts for 2025-2027, estimating net profits of 2.387 billion yuan, 2.950 billion yuan, and 3.631 billion yuan respectively, with corresponding EPS of 1.63 yuan, 2.02 yuan, and 2.48 yuan [5]. - The company is recognized as a leading provider of ICT infrastructure solutions in China, with a comprehensive layout in computing infrastructure [5].

Investment Rating - The report maintains a positive investment rating for the domestic computing power sector, emphasizing the potential of supernodes and aerospace computing as key growth areas [5]. Core Insights - Supernodes are expected to become a mainstream technology under cluster expansion, significantly enhancing training and inference capabilities, thus driving the development of the domestic computing power ecosystem [3][41]. - Aerospace computing is gaining attention from major tech companies and is anticipated to be a crucial direction for domestic computing power development [3][41]. - The report highlights key investment opportunities in domestic AI computing, including chip design, advanced wafer manufacturing, liquid cooling solutions, and AI servers [3][41]. Summary by Sections Market Review - During the week of November 3-7, the CSI 300 index rose by 0.82%, while the small and medium-sized board index fell by 0.59%, and the ChiNext index increased by 0.65%. The computer sector (CITIC) saw a decline of 2.08% [1][49]. Industry News - Major developments include the launch of the world's first single-cabinet 640-card supernode by Zhongke Shuguang, which integrates advanced cooling and power supply technologies [3][45]. - The report notes that two leading domestic computing power companies, Moer Thread and Muxi Co., are set to go public, which could catalyze further growth in the sector [3][29]. Company Dynamics - Fangzhi Technology plans to reduce its shareholding by up to 3% due to personal financial needs, without affecting the company's control [2]. - Suochen Technology is acquiring a 60% stake in Beijing Likong Yuantong Technology for approximately 192 million yuan, indicating active consolidation in the sector [2][47]. Investment Recommendations - The report suggests focusing on key players in domestic AI computing, including chip design leaders like Cambrian and Haiguang Information, as well as advanced wafer manufacturers like SMIC [3][41]. - In aerospace computing, it recommends monitoring core companies such as Putian Technology and Shunhao Co., along with satellite internet leaders [3][41].

Investment Rating - The report indicates an "Overweight" rating for the computer industry, reflecting a positive outlook on AI investments and growth potential in the sector [12]. Core Insights - North American tech giants are significantly increasing their capital expenditures, with a strong focus on AI deployment. Google, Meta, and Microsoft have all raised their capital expenditure forecasts for the upcoming years, indicating robust growth in AI-related services and infrastructure [14][16][17]. - Google Cloud's revenue grew by 34% in Q3, reaching $15.2 billion, driven by strong demand for enterprise AI products. The Gemini Enterprise AI platform has over 2 million users across 700 companies [14][15]. - Meta plans to invest over $60 billion in AI technology and infrastructure by 2028, with a revised capital expenditure forecast of $70-72 billion for the current fiscal year [16]. - Microsoft's cloud revenue reached $49.1 billion, growing 26%, with AI functionalities seeing significant user engagement. The company is increasing its capital expenditures to support the growing demand for Azure services [17][18]. Summary by Sections North American Tech Giants' Capital Expenditure - Google expects capital expenditures to be between $91 billion and $93 billion for 2025, up from a previous estimate of $85 billion. Significant growth is anticipated for 2026 [16]. - Meta's capital expenditure forecast for the current fiscal year has been adjusted to $70-72 billion, with a notable increase expected for 2026 [16]. - Microsoft reported a capital expenditure of $34.9 billion in Q3, with half allocated to short-term assets like GPUs and CPUs to meet Azure demand [17][18]. AI Developments and User Engagement - Google's Gemini App has over 650 million monthly active users, with query volume tripling since Q2. The company processes over 1.3 trillion tokens monthly, a 20-fold increase within a year [15]. - Microsoft's AI features have 900 million active users, with the Copilot series exceeding 150 million monthly active users [17]. Domestic AI Progress - Alibaba is investing heavily in AI, with over 600 million downloads of its Tongyi Qianwen model and the largest AI open-source community in China, serving over 20 million users [19][20]. - The launch of the world's first single-cabinet 640-card super node by Inspur demonstrates advancements in computing power and efficiency, with a 20-fold increase in computing density compared to similar products [22][25]. Investment Opportunities - The report suggests focusing on software companies showing signs of recovery and potential in AI agents, including Kingsoft Office, Deepin Technology, and others [26]. - Key players in computing power and custom AI agents include Cambricon, Hygon, Alibaba, Tencent, and various others across different sectors [27][28].

- The report predicts adjustments to the constituent stocks of major indices, including CSI 300, CSI 500, CSI 1000, CSI 2000, SSE 50, and STAR 50, based on publicly available index compilation rules and data [5][10][16][18][20] - CSI 300 Index is constructed by selecting the top 50% stocks based on average daily trading volume over the past year, followed by the top 300 stocks ranked by average daily market capitalization, while adhering to a 10% adjustment limit, priority for old samples, and a 20% buffer zone rule [5] - The adjustment prediction for CSI 300 Index involves calculating the average daily market capitalization and trading volume of A-shares over the past year, excluding stocks with suspension, violations, or financial reporting issues [5] - The report defines a "shock coefficient" to measure the price impact and duration caused by passive index fund rebalancing, calculated as: $ Shock Coefficient = (Passive Buy Amount - Passive Sell Amount) / Average Daily Trading Volume $ This coefficient is applied to assess the impact of adjustments on stocks [6][9] - CSI 500 Index is constructed by excluding CSI 300 constituent stocks and the top 300 stocks by average daily market capitalization over the past year, followed by removing the bottom 20% stocks by average daily trading volume, and selecting the top 500 stocks by market capitalization, adhering to a 10% adjustment limit, priority for old samples, and a 10% buffer zone rule [10] - CSI 1000 Index is constructed by excluding CSI 800 constituent stocks, the top 300 stocks by market capitalization, and stocks with insufficient liquidity (bottom 20% by trading volume), selecting the top 1000 stocks by market capitalization over the past year [16] - CSI 2000 Index is constructed by excluding CSI 800 and CSI 1000 constituent stocks, the top 1500 stocks by market capitalization, and selecting the top 2000 stocks by market capitalization over the past year [16] - SSE 50 Index is constructed by selecting the top 50 stocks by market capitalization and liquidity from the Shanghai Stock Exchange, adhering to adjustment rules similar to other indices [18] - STAR 50 Index is constructed by selecting the top 50 stocks by market capitalization from STAR Market, excluding stocks with delisting risks, major violations, or low liquidity (bottom 10% by trading volume) [20] - The report predicts adjustments to the STAR 50 Index, with two stocks, Aojie Technology-U and Shengke Communication-U, being added [20] - The shock coefficients for the predicted adjustments are calculated for each stock, with the highest coefficients observed for stocks such as Guangqi Technology and Ningbo Port in CSI 300, and Sheneng Shares and Suzhou Supor in CSI 500 [7][11][19][21]

Investment Rating - The industry investment rating is maintained as "Positive" [4] Core Insights - The release of the Shuguang super node marks a significant breakthrough in domestic computing power, showcasing advantages over traditional intelligent computing methods through high-speed interconnection technology, which enhances bandwidth and reduces latency [6] - The Shuguang Scale X640 super node, launched on November 6, 2025, is the world's first 640-card single-cabinet super node, designed for training trillion-parameter models, improving performance by 30%-40% compared to traditional solutions [6] - Domestic computing power and super nodes are expected to continue achieving breakthroughs, with the Shuguang X640 demonstrating top-tier performance, efficiency, reliability, and openness in the AI computing system [6] Summary by Sections Investment Recommendations and Targets - Recommended stocks include Haiguang Information (688041, Buy), Zhongke Shuguang (603019, Buy), and others that are either rated or not rated [3] Industry Overview - The report focuses on the computer industry in China, highlighting advancements in domestic computing capabilities and the competitive landscape [4][6]

Core Insights - The launch of the scaleX640 ultra-node by Zhongke Shuguang represents a significant advancement in computing infrastructure, addressing the bottleneck in AI scalability and deployment [1][2][3] Group 1: Product Features - The scaleX640 ultra-node is the world's first single cabinet-level 640-card ultra-node, designed on a leading open system hardware architecture [1] - It features a high-density architecture with a "one-to-two" design, achieving a 20-fold increase in computing density compared to traditional solutions [2] - The ultra-node supports MoE trillion-parameter model training and can ensure the deployment of ultra-large-scale clusters with 100,000 cards [2][3] Group 2: Technological Innovations - The ultra-node incorporates innovative technologies such as ultra-high-speed orthogonal architecture, immersion phase change cooling, and high-voltage direct current power supply [2] - It enables a comprehensive optimization of software and hardware, resulting in enhanced performance, efficiency, and reliability [2] Group 3: Industry Collaboration - Zhongke Shuguang has partnered with over 20 industry chain enterprises to launch the "AI Computing Open Architecture," aimed at reducing R&D barriers and promoting ecosystem collaboration [3] - The scaleX640 ultra-node exemplifies this collaborative approach, supporting multi-brand acceleration cards and mainstream computing ecosystems [3]

Core Viewpoint - The launch of the scaleX640 by Dawning Technology is expected to enhance the performance of trillion-parameter model training and inference by 30-40%, potentially accelerating the breakthrough of domestic computing chips in training scenarios [2][4]. Performance Enhancement - The scaleX640 achieves a 20-fold increase in computing density compared to traditional solutions, enabling a 30-40% performance improvement in MoE trillion-parameter model training and inference scenarios [2]. - The system has undergone a 30-day reliability test, ensuring support for large-scale cluster deployments of up to 100,000 cards, which helps mitigate performance degradation and cost increases during model training due to chip failures [2]. Open Ecosystem - The scaleX640 adopts an AI computing open architecture, supporting multiple brands of acceleration cards at the hardware level and being compatible with mainstream computing ecosystems at the software level [2]. - The open architecture is expected to facilitate the integration of domestic computing software ecosystems and unify supernode structures, potentially aiding rapid iterations of domestic supernodes [2]. Technological Innovation - The scaleX640 utilizes innovative technologies such as ultra-high-speed orthogonal architecture, ultra-high-density blades, immersion phase change cooling, and high-voltage direct current power supply, which may drive technological advancements in the upstream supply chain [3]. Catalysts for Growth - The successful implementation of the scaleX640 in domestic large model training scenarios is anticipated to accelerate the construction of local intelligent computing centers [4].

Investment Rating - The report assigns an "Increase" rating for the stocks mentioned, indicating a potential rise of over 15% relative to the CSI 300 index within the next 12 months [5][13]. Core Insights - The launch of the Shuguang scaleX640 marks a significant advancement in domestic computing power, achieving a 30-40% performance improvement in trillion-parameter model training and inference compared to traditional solutions [2][5]. - The scaleX640 enhances the cost-effectiveness of inference scenarios and is expected to accelerate the breakthrough of domestic computing chips in training applications [5]. - The report highlights the long-term stability of the scaleX640, which has undergone over 30 days of reliability testing, ensuring support for large-scale cluster deployments [5]. - The open architecture of the scaleX640 is anticipated to facilitate the integration of domestic computing software ecosystems and unify supernode structures, potentially leading to rapid iterations in domestic supernodes [5]. Summary by Sections Performance and Features - The scaleX640 boasts a 20-fold increase in computing density per cabinet and supports trillion-parameter model training, with a performance boost of 30-40% in inference scenarios compared to traditional methods [5]. - Innovative technologies such as ultra-high-speed orthogonal architecture, high-density blades, immersion phase change cooling, and high-voltage direct current power supply are employed in the scaleX640 [5]. Recommended Stocks - The report recommends the following stocks: - Cambrian-U (688256.SH) with a closing price of 1480.00 and an EPS forecast of 5.04 for 2025 [7]. - Haiguang Information (688041.SH) with a closing price of 239.68 and an EPS forecast of 1.34 for 2025 [7]. - SMIC (688981.SH) with a closing price of 124.85 and an EPS forecast of 0.46 for 2025 [7]. - Zhaoyi Innovation (603986.SH) with a closing price of 227.20 and an EPS forecast of 2.35 for 2025 [7]. - Shengke Communication-U (688702.SH) with a closing price of 119.31 and an EPS forecast of 0.04 for 2025 [7]. - The report also mentions Chipone Technology (688521) as a related stock [5].

Market Overview - On November 7, A-shares opened lower but rebounded, with the Shanghai Composite Index and Shenzhen Component Index both down by 0.16%, and the ChiNext Index down by 0.37%. In contrast, the North Star 50 Index rose nearly 1% [1][2] - The total market turnover for the half-day was 1.27 trillion yuan, slightly lower than the previous day, with over 2,300 stocks rising [2] Sector Performance - The basic chemical, petroleum and petrochemical, and retail sectors saw gains, while lithium battery, fluorine chemical, phosphorus chemical, and photovoltaic stocks experienced significant surges [2][5] - The fluorine chemical sector rose by 4.00%, while lithium battery-related stocks also saw substantial increases, with individual stocks like Dongyue Silicon Materials and Zhaoyuan New Energy hitting the daily limit [3][5] Notable Stocks - Key stocks in the lithium battery sector included: - Dongyue Silicon Materials: 20.04% increase - Zhaoyuan New Energy: 20.01% increase - Haineng Technology: 19.95% increase [6][10] - In the photovoltaic sector, stocks like Hongyuan Green Energy and Yijing Photovoltaic also saw significant gains, with Hongyuan Green Energy rising by 10.01% [7] Storage Chip Sector - The storage chip sector was active, with stocks like Demingli hitting the daily limit and reaching a new historical high of 271.85 yuan per share [11][12] - The supply-demand situation for storage chips is tight, with SK Hynix completing negotiations for HBM4 supply with Nvidia, leading to price increases [14][15] AI Sector - The AI application sector faced declines, with stocks related to operating systems, servers, and ChatGPT all underperforming. Notable declines included Kingsoft Office and 360, both dropping over 3% [16][17] - Concerns about high valuations in the AI sector have intensified, with discussions around the potential for an "AI bubble" emerging [16]

Group 1 - The article highlights the launch of the world's first 640-card super node, scaleX640, by Zhongke Shuguang, which features a high-density architecture design for high-speed interconnection and large-scale cluster deployment [2] - Hyundai Motor Group announced the opening of a new facility worth €150 million at its European Technology Center in Germany, focusing on innovation and engineering excellence in Europe [2] - SK Hynix is reportedly planning to restructure its production lines to enhance the competitiveness of its 1c DRAM process, with a focus on efficiency and precision in core equipment [2] - Panasonic is investing over 1.5 billion yuan to build an electronic materials factory to address opportunities in the generative AI server market, including new materials for 5G electronics and semiconductors [2]