每经AI快讯，12月2日，中信证券研报指出，太空算力正从概念逐渐走向现实，海外科技巨头已纷纷布 局。我们认为太空算力的能源供给将高度依赖光伏技术，目前太空中使用的三结砷化镓价格非常昂贵， 而成本较低的晶硅电池抗辐照性能较差，无法长时间应用于空间领域。钙钛矿作为兼具高效率和低成本 的下一代光伏技术，有望成为重要的太空能源解决方案。2025年以来，钙钛矿产业化进展加速，效率、 稳定性纷纷取得突破，行业内多条GW级产线相继投产，建议关注钙钛矿产业投资机会。 ...

证券研究报告|宏观点评报告 [Table_Date] 2025 年 12 月 01 日 哪些行业可能成为轮动的"下一站"？若市场继续追逐弹性，可能会选择"今年表现强势，但尚未明显修 复"的行业。从 SW 一级行业来看，电力设备在今年以来涨幅靠前，上涨 40.85%，但在 11 月 21 日至今仍下跌 1.22%，有望吸引资金关注。基础化工今年以来上涨 28.25%，而在 11 月 21 日至今仍下跌 1.39%，同样可能成为 轮动方向。而若市场避险情绪升温，则可能会选择"今年上涨不多，且尚未明显修复"的行业，如消费和红利品 种。此外，11 月以来的政策空窗期已基本进入尾声，市场可能提前博弈 12 月政治局会议和中央经济工作会议的政 策，消费、地产等行业有望迎来阶段性反弹。 港股科技和红利品种的调整幅度，已基本来到历史底部经验值。科技方面，恒生科技在 10 月 3 日至 11 月 21 日下跌 19.26%，从 2024 年 10 月以来的经验来看，其每段跌幅基本处于 20%-30%范围内；红利方面，港股通红 利低波指数在 2024 年 10 月以来的下跌分为两类，一是小幅下跌，跌幅处于 5%-7%附近；二是大 ...

Investment Rating - The report maintains a "Recommended" investment rating for the computer industry [2][19]. Core Insights - The acceleration of space computing deployment is creating a new competitive landscape in the AI industry, with significant developments from both China and the United States [4][14]. - The global AI industry is entering a resonance period, driven by the rapid expansion of AI infrastructure and the commercialization of downstream applications [5][18]. Summary by Sections Recent Developments - The Beijing Municipal Science and Technology Commission announced plans for a space data center construction, aiming to build a system with over GW power in low Earth orbit by 2035 [4][14]. - China's "Trisolaris Computing Constellation" successfully launched its first 12 satellites, marking a significant milestone in space computing infrastructure [4][14]. Competitive Landscape - NVIDIA and its startup partner Starcloud launched the first space AI server equipped with the H100 chip for a three-year testing mission [4][14]. - Google is exploring the integration of its TPU chips into satellites to utilize solar energy for power [4][14]. Key Company Performance - Alibaba's cloud revenue grew by 34% year-on-year, with AI-related products seeing triple-digit growth for nine consecutive quarters [5][18]. - Google’s TPU has become a strong competitor to NVIDIA, with the latest generation achieving significant performance metrics [17][18]. Investment Recommendations - The report suggests focusing on several key areas: domestic chips (e.g., Haiguang Information, Cambrian), domestic servers (e.g., Zhongke Shuguang, Digital China), AI applications (e.g., Kingsoft Office, iFlytek), and space computing (e.g., Aerospace Intelligent Equipment, Zhongke Xingtou) [5][19].

近期， 算力之争已从地面扩展至太空，算力真的要上天了。 11月27日， 北京拟在700-800公里晨昏轨道建设运营超过千兆瓦（GW）功率的集 中式大型数据中心系统，以实现 将大规模 AI算力搬上太空 。在此刺激下，周一通宇通讯、航天发展等一字涨停，卫星互联网、商业航天、时空 大数据等相继迎来爆发。 （ 点此领取产业链名单 ） 下一场 AI算力的竞赛，或在太空苍穹，海外巨头以及国内科技巨头均已加速布局： 11月2日，美国初创公司Starcloud成功发射了Starcloud-1卫星，首次将英伟达H100GPU送入地球轨道。 本文首发于公众号"私募排排网"。 （点击↑↑ 上图查看详情 ） 谷歌在11月5日宣布启动"太阳捕手计划" (ProjectSuncatcher)，计划在2027年初发射两颗搭载TPU芯片的原型卫星，将AI算力直接部署到太空。 马斯克此前表示："只需扩大StarlinkV3卫星规模即可实现在太空建造大型数据中心， 并预言五年内，太空中的 AI计算会比地面便宜。" 0 1 太空算力是什么？ 太空算力是指将计算资源(服务器、AI芯片、存储设备)部署在近地轨道、地球同步轨道等太空环境，构建"太空数据 ...

Core Insights - Beijing plans to construct and operate large-scale centralized data centers with over 1 GW power on a 700-800 km orbit to facilitate the deployment of AI computing in space [1] - Guangfa Securities indicates that overseas tech companies are accelerating their space computing layouts, suggesting that space may become the optimal location for large-scale AI computation [1] Industry Developments - On November 2, NVIDIA, in collaboration with its invested startup Starcloud, successfully launched the first space AI server equipped with H100 via SpaceX's Falcon 9 rocket, marking the official entry of "space computing" into the AI era [1] - Starcloud claims that space data centers could reduce energy costs by 90% and lessen Earth's energy consumption demands [1] - On November 4, Elon Musk announced plans to expand the Starlink V3 satellite scale to build space data centers, aiming to deploy 100 GW of data centers annually through Starship within 4-5 years [1] - On November 5, Google announced the launch of the "Sun Catcher Project," planning to send two prototype satellites equipped with Trillium-based TPUs into orbit by early 2027 to directly deploy AI computing in space [1] - Google executives stated that space may become the best venue for achieving large-scale artificial intelligence computation in the future [1] Market Performance - The Guotai AI ETF (159388) tracking the ChiNext AI Index (970070) experienced a daily fluctuation of 20%, reflecting the performance of listed companies involved in AI technology and applications across various sectors, showcasing significant technological innovation and growth potential [1]

Investment Rating - Investment advice: The report suggests a focus on the accelerating US-China space computing competition, highlighting key companies to watch [3][9]. Core Insights - The US-China space computing race is intensifying, with Musk proposing a space AI satellite plan and Beijing initiating the construction of a gigawatt-level space data center [3][9]. - The US "Genesis Plan" aims to integrate resources from 17 national laboratories to accelerate foundational research in key areas such as nuclear fusion and biotechnology [3][9]. - Chinese companies are demonstrating full-stack AI capabilities, with Alibaba's Qwen app achieving over 10 million downloads and DeepSeek's math model reaching international gold standards [3][9]. Summary by Sections Space Computing Developments - Musk envisions deploying AI computing centers in space within 4-5 years, with plans for 100 gigawatts of solar AI satellites annually, which is equivalent to a quarter of US power [3][9]. - Beijing's plan includes a three-phase approach for the gigawatt-level space data center, focusing on key technology breakthroughs, cost-reducing assembly, and mass production from 2025 to 2035 [3][9]. US Research Initiatives - The "Genesis Plan" was initiated by an executive order in November 2025, aiming to identify 20 national challenges within 60 days and showcase initial results within 270 days [3][9]. - Collaboration with private sector companies like NVIDIA and AMD is part of the plan to enhance computational resources for scientific research [3][9]. Chinese AI Innovations - Alibaba's Qwen app, based on the Qwen3 model, has become one of the fastest-growing AI applications globally, indicating a significant breakthrough in the AI to consumer market [3][9]. - DeepSeek's math model, DeepSeekMath-V2, has introduced a self-verification mechanism, achieving gold-level performance in international competitions [3][9].

Investment Rating - The report rates the computer industry as "Outperform" [1] Core Insights - The report highlights the acceleration of AI applications in China, with the launch of Ant Group's AI assistant "Lingguang" achieving over 2 million downloads in just six days, indicating strong market demand for high-quality AI products [21][22] - The Chinese government is enhancing cybersecurity governance, with new regulations marking a shift towards more detailed and effective management of cybersecurity incidents [13][14] - China is leading the space computing race, with a three-phase plan for space data centers aiming for significant advancements in computational power and efficiency [16][20] - The human-shaped robot industry in Beijing is rapidly evolving, with government support aimed at commercializing and scaling production [24][26] - Singapore's shift to using Alibaba's Qwen model for its national AI program signifies a structural shift in the global AI landscape, showcasing the growing influence of Chinese AI technologies [29][31] Summary by Sections Cybersecurity - The report discusses the recent emphasis on cybersecurity governance by the Chinese government, including the implementation of the "National Cybersecurity Incident Reporting Management Measures" which aims to enhance accountability and efficiency in reporting incidents [13][14] - The new cybersecurity law set to take effect in January 2026 complements these measures, establishing a robust legal framework for cybersecurity management [15] Space Computing - A three-phase construction plan for space data centers has been outlined, with the first phase targeting a total power of 200KW and a computational capacity of 1000POPS by 2027 [16][17] - The report notes that advancements in AI and satellite technology are expected to drive down costs and enhance the scalability of space computing [20] AI Applications - The rapid adoption of AI applications in various sectors is highlighted, with "Lingguang" setting records for downloads, reflecting a shift towards practical and user-friendly AI solutions [21][22] - The report emphasizes the importance of a supportive ecosystem for AI development in China, including a robust industrial framework and a large user base [22][23] Human-Shaped Robots - The report outlines Beijing's strategic focus on developing the human-shaped robot industry, with significant government backing to foster innovation and commercialization [24][26] - Companies in this sector are already achieving substantial production milestones, indicating a promising market outlook [25][28] Global AI Landscape - The collaboration between Alibaba and Singapore's AI initiative marks a significant recognition of Chinese AI capabilities on the global stage, suggesting a shift in the competitive dynamics of AI technology [29][31] - The report notes that this partnership reflects a broader trend of countries seeking to establish sovereign AI capabilities, leveraging local technologies [30][31]

Core Insights - By 2030, space computing power is expected to reach 400,000 P, with the first space data center being established, playing a significant role in fields such as 6G, autonomous driving, and weather forecasting [1][4] Group 1: Space Data Center Development - The first space computing satellite, "Chenguang-1," has completed product development and is undergoing assembly tests, with plans for launch around late 2025 or early 2026 [3][4] - Space data centers are seen as a crucial solution to the computing power bottleneck in the AI era, overcoming limitations faced by ground data centers such as space, power supply, and heat dissipation [3] Group 2: Technological Innovations - "Chenguang-1" is designed to validate key technologies such as in-orbit energy production and heat dissipation, with its computing power equivalent to that of a ground server [3][4] - The construction of the space data center will occur in three phases: "sky computing," "ground computing," and "space-based main computing," utilizing modular and standardized designs [4] Group 3: Future Projections - Within three years, the space computing constellation is projected to achieve a computing power level of 1,000 P, with a long-term goal of 400,000 P by 2030, which would equal the total computing power of all ground data centers in China [4]

Group 1: Market Performance - The cloud computing sector saw a significant rise, with Hong Kong's Alibaba and A-shares "Yizhongtian" both surging [1] - The Huatai Securities report indicates that the demand for domestic computing power is rapidly expanding, driving further investment growth [5] - The cloud computing ETF Huatai (159273) experienced a strong increase of over 2%, with a trading volume exceeding 25 million yuan [1][3] Group 2: Company Developments - Alibaba's cloud revenue for FY2Q26 reached 39.824 billion yuan, a year-on-year increase of 34%, indicating strong customer demand [5] - The introduction of the Doubao mobile assistant, developed in collaboration with mobile manufacturers, showcases the integration of AI capabilities in mobile technology [2] - Alibaba has raised its AI investment guidance, suggesting additional investments beyond the previously announced 380 billion yuan [5] Group 3: Industry Trends - The U.S. market is anticipating a potential interest rate cut by the Federal Reserve, with a probability of 85%-86% for a 25 basis point reduction [1] - The shift in Singapore's AI project to utilize Alibaba's open-source model reflects the growing influence of Chinese AI models globally [2] - The capital expenditure (CAPEX) of major cloud service providers (CSPs) is expected to continue increasing, driven by revenue and profit growth [6]

Investment Rating - The report maintains a "Buy" rating for the communication industry and related companies [9][54]. Core Insights - The communication sector saw an 8.70% increase in the index last week, significantly outperforming the Shanghai Composite Index and Shenzhen Component Index, which rose by 1.40% and 3.56% respectively [2][13]. - Domestic computing power is expanding rapidly, with Alibaba's cloud revenue for FY2Q26 reaching 39.824 billion yuan, a year-on-year increase of 34%, exceeding Bloomberg's consensus forecast [3][14]. - The DeepSeek open-source model, DeepSeek-Math-V2, has shown significant improvements in mathematical reasoning capabilities, outperforming other models in various benchmarks [19][20]. - Beijing plans to establish a large-scale data center system in the 700-800 km dawn-dusk orbit, aiming to transfer substantial AI computing power to space [19][20]. Summary by Sections Domestic Computing Power - Alibaba's cloud revenue growth is driven by strong customer demand, with orders increasing faster than server deployment [3][14]. - The company may increase its previously announced investment of 380 billion yuan due to high demand [3][14]. - The DeepSeek-Math-V2 model has achieved near 99% accuracy in basic benchmarks, indicating a significant advancement in AI capabilities [19][20]. Space Computing Power - Beijing's initiative to build a data center in space aims to support large-scale AI computing [19][20]. - The project is structured in three phases, focusing on technological breakthroughs and cost reduction from 2025 to 2035 [19][20]. - The initiative is expected to create a new industry chain integrating reusable rockets, computing constellations, and data applications [19][20]. Recommended Companies - The report recommends several companies within the communication sector, including ZTE Corporation, Ruijie Networks, and Alibaba, highlighting their potential in the AI computing chain and core asset value reassessment [4][54][55]. - Specific target prices and investment ratings for these companies are provided, indicating a positive outlook for their performance [54][55].