来源：东方网 7月15日，为梦想添翼，让AI发光——"天翼云息壤杯"高校AI大赛总决赛开幕式在北京举行。自2024年9 月启动以来，大赛吸引全国3800多支队伍参赛，经过10个月的激烈角逐，来自中国科学技术大学、武汉 大学、华中科技大学等全国高校的60支精英团队脱颖而出来到决赛现场。中国电信、华为、中国自动化 学会、中国通信学会、中国图象图形学学会、全球计算联盟、天翼云等联合主办单位的领导，以及评 委、导师和参赛选手，共同见证了这场聚焦人工智能创新与青年人才培养的高水平赛事总决赛启幕。 中国电信集团有限公司副总经理黄智勇在致辞中表示，真正实现科技自立自强必须要加快推动关键核心 技术的自主掌控，必须要深深植根于我们自己的土壤。中国电信坚定履行建设网络强国、科技强国、数 字中国和维护网信安全的使命责任，实施云改数转战略，牵头打造云计算原创技术策源地和创新联合 体，天翼云作为国家云的框架全面成型，迈向了智能云发展的新阶段，研发了基于全国产化万卡集群训 练的万亿参数"星辰"大模型，构建了领先的一体化智算服务平台息壤，积极赋能千行百业数字化智能化 转型。以本次高校AI大赛为契机，一是希望各位参赛选手尽情释放才华，享受巅 ...

Investment Rating - The report maintains an "Overweight" rating for the telecommunications sector [6] Core Insights - The global telecommunications industry is experiencing steady demand growth, driven by emerging markets and increasing ARPU in regions like North America. AI is expected to bring transformative opportunities to the industry by enhancing operational efficiency and customer insights [1][4][12] Summary by Sections Global Telecommunications Development - The global telecommunications industry achieved a total revenue of 8.8 trillion RMB in 2024, with a year-on-year growth of 3.0%. The net profit attributable to shareholders was 834.2 billion RMB, reflecting a decline of 8.4% [12] - Revenue by region includes: Europe (2.2 trillion RMB, +4.9%), North America (2.4 trillion RMB, +2.7%), China (1.95 trillion RMB, +3.4%), Asia-Pacific (1.6 trillion RMB, -0.2%), Africa and Middle East (0.7 trillion RMB, +4.9%) [12] Market Observations - In North America and Europe, leading telecom operators are seeing steady user growth, while mobile penetration rates are high, resulting in low single-digit growth rates. In contrast, regions like India and Africa are experiencing rapid growth [2][17] - Fiber business is a significant growth driver, with AT&T's fiber revenue growing by 19% year-on-year in Q1 2025 [2] Profitability Trends - Profitability varies across regions, with Middle Eastern and North American operators showing strong profitability. The average ROE for major global telecom operators in 2024 was 15.66% [3][17] - Domestic operators in China have slightly lower ROE compared to the global average, but as 5G investments enter the recovery phase, profitability is expected to improve [3][17] AI Impact on the Industry - AI is transforming the telecommunications sector by reducing operational costs and enhancing customer service capabilities. For instance, Deutsche Telekom uses AI to monitor network performance and improve fiber deployment efficiency [4][22] - Companies like T-Mobile are leveraging AI to predict customer needs and enhance service offerings, improving customer satisfaction [4][22] Domestic Telecommunications Dynamics - In China, the telecommunications business revenue for the first five months of 2025 reached 748.8 billion RMB, a year-on-year increase of 1.4%. The average mobile internet access traffic per user reached a record high of 21.3 GB/month, up 14.9% year-on-year [5][57] - The three major operators in China are expanding their intelligent computing capabilities, with China Mobile achieving a computing power scale of 43 EFLOPS [5] Key Recommendations - The report recommends focusing on major Chinese telecom operators: China Mobile, China Telecom, and China Unicom, all rated as "Buy" or "Overweight" [9]

q 证券研究报告 | 公司首次覆盖 中国电信(601728.SH) 2025 年 06 月 12 日 买入（首次） 相关研究 中国电信（601728.SH）：高股 息带来稳健收益，数据入表+前 沿布局打开成长空间 投资要点 所属行业：通信/通信服务 当前价格(元)：7.76 证券分析师 李宏涛 资格编号：S0120524070003 邮箱：liht＠tebon.com.cn 研究助理 市场表现 -23% -11% 0% 11% 23% 34% 46% 57% 69% 2024-06 2024-10 2025-02 中国电信 沪深300 | 沪深300对比 | 1M | 2M | 3M | | --- | --- | --- | --- | | 绝对涨幅(%) | 0.51 | -1.01 | -2.37 | | 相对涨幅(%) | -0.75 | -4.86 | -1.18 | | 资料来源：德邦研究所，聚源数据 | | | | 请务必阅读正文之后的信息披露和法律声明 传统业务维持稳健，产业数字化成效卓著。2020-2024 年公司营收主体仍然是传 统通信业务（移动通信服务、固网及智慧家庭服务）。2020 年 ...

Core Insights - The three major telecom operators in China reported distinct financial results for Q1 2025, with China Mobile leading in revenue and profit, while China Telecom excelled in 5G user penetration [1][2][3] - The overall revenue growth for the telecom industry is slowing, with the operators entering a "stable" phase, focusing on high-quality development and profitability [2][3] Financial Performance - China Mobile reported revenue of 263.8 billion yuan and net profit of 30.6 billion yuan, nearly 1.3 times the combined revenue of China Telecom and China Unicom [1] - China Telecom's revenue reached 134.5 billion yuan with a net profit of 8.9 billion yuan, while China Unicom surpassed 100 billion yuan in revenue for the first time, achieving 103.35 billion yuan and a net profit of 2.61 billion yuan [1] - Revenue growth rates for Q1 2025 were 0.02% for China Mobile, 0.01% for China Telecom, and 3.88% for China Unicom, significantly down from 10.3%, 9.4%, and 9.2% in the same period of 2023 [3] User Metrics - China Mobile's total mobile users exceeded 1 billion, with 5G users reaching 578 million, accounting for 54% of the industry total [2] - China Telecom's 5G user penetration rate is 62%, while China Unicom's is below 50% [2] - In the fixed broadband market, China Mobile leads with 320 million users, followed by China Telecom with 173 million and China Unicom with 124 million [2] Profitability and Cost Management - Despite slowing revenue growth, net profits for the three operators showed slight increases: China Mobile up 3.45%, China Telecom up 3.1%, and China Unicom up 6.5% [4] - Operators are maintaining profitability through reduced capital expenditures and optimized cost structures, with total capital expenditures for 2025 projected at 289.8 billion yuan, down 9.1% year-on-year [4] Emerging Business Focus - The operators are shifting resources towards emerging fields such as computing power networks and AI to create a "second growth curve" [7][8] - China Telecom's computing power investment grew by 22%, with IDC revenue at 9.5 billion yuan and smart income surging by 151.6% [7] - China Unicom's cloud revenue reached 19.72 billion yuan, a 114% increase, while China Mobile plans to invest 37.3 billion yuan in computing power, representing 25% of its total capital expenditure for 2025 [7][8] Market Challenges - The telecom industry faces structural challenges in computing power demand, with a surplus in general computing power and a shortage in intelligent computing power [10] - The operators are also under pressure from competition with internet giants like Tencent and Alibaba, which are encroaching on the enterprise market [10][11] - The operators are attempting to build differentiated competitive advantages, but achieving significant revenue growth remains a challenge [11]

Group 1 - The core viewpoint emphasizes that computing power is the foundation of artificial intelligence (AI) development, with ongoing technological advancements and practical applications driving the deep integration of AI and computing power [1] - China Telecom's chairman highlighted that cloud computing is crucial for AI development, and the company aims to establish Tianyi Cloud as the national cloud, aligning with the national strategy for digital economic development [1] - The "National Cloud" project, promoted by the State-owned Assets Supervision and Administration Commission, aims to build a self-controlled cloud computing infrastructure by integrating resources from state-owned enterprises [1] Group 2 - China Telecom has established major intelligent computing clusters in Shanghai and Beijing, and launched the world's first commercial intelligent computing platform in the Greater Bay Area, achieving significant advancements in both quality and quantity of intelligent computing power [1] - The company has made breakthroughs in key technologies such as distributed cloud operating systems and server operating systems, and has developed a full-stack cloud platform [1] - China Telecom has introduced the first domestically developed trillion-parameter "Xingchen" foundational model system, leading the way in open-sourcing within state-owned enterprises [2] - The demand for computing power is shifting from general computing to intelligent computing, driven by rapid advancements in AI technology and the accelerated deployment of large AI models [2]

Group 1 - Shenzhou 19 manned spacecraft successfully completed its mission with the return capsule landing at Dongfeng landing site, confirming the good health of astronauts [2] - China Telecom's "Xirang" intelligent computing platform has reached a total computing power of 62 EFLOPS, with 35 EFLOPS from its own resources and 27 EFLOPS from partnerships, across 18 cities and industry computing centers [2] - The world's first four-seat electric aircraft is set to begin production in Yanqing, marking significant progress in the field of new energy general aviation aircraft [2] Group 2 - Tangem announced plans to launch Tangem Pay in 2025, a Visa card embedded with a cold wallet chip for self-custody of cryptocurrency transactions [2]

Core Viewpoint - China Telecom aims to build a comprehensive intelligent cloud capability system consisting of "computing power, platform, data, model, and application" to drive innovation in the artificial intelligence industry [1][2]. Group 1: Intelligent Cloud Development - China Telecom has established major intelligent computing clusters in Shanghai and Beijing, and launched the world's first commercial intelligent computing super node in the Greater Bay Area [1]. - The company has developed the first full-size, full-modal, domestically produced trillion-parameter "Star" foundational model system and has taken the lead in open-sourcing among state-owned enterprises [1]. - The "Star Sea" data intelligence platform has formed a high-quality dataset of 10 trillion tokens and has launched over 50 industry applications in sectors such as government, industry, and education, benefiting over 87 million users [1]. Group 2: Financial Performance - In the fiscal year 2024, China Telecom's intelligent revenue reached 8.9 billion yuan, representing a year-on-year growth of 195.7% [1]. - The revenue from Tianyi Cloud in 2024 was 113.9 billion yuan, with a year-on-year increase of 17.1% [1]. Group 3: Future Investment Plans - For 2025, China Telecom plans capital expenditures of 83.6 billion yuan, with a preliminary allocation for computing power investment expected to grow by over 20% year-on-year, with flexibility for adjustments [2]. - The company will enhance its intelligent computing capabilities by integrating "central clusters + edge data centers" and promoting the deep collaboration of computing networks [2]. Group 4: Platform and Service Upgrades - In the IaaS layer, China Telecom will advance intelligent computing construction and strengthen cloud-intelligent integration capabilities [2]. - The company introduced the Triless platform architecture to upgrade computing power scheduling technology and provide a one-stop service for computing power dispatch and AI development [2]. - In the DaaS layer, the "Star Sea" data intelligence platform will be upgraded to integrate proprietary, open-source, and third-party datasets [2]. Group 5: Focus on SaaS Layer - In the SaaS layer, China Telecom will focus on the digital intelligence needs of production, life, and governance, accelerating the large-scale promotion of AI phones and AI cloud computers [3].

运营商财经网讯 4月30日上午，以"数智国云、共谱新篇"为主题的智能云生态大会主论坛隆重召开。会上，中国电信天 翼云公司董事长、总经理胡志强详细介绍了天翼云"息壤"的最新技术成果Triless架构。 首先，在算力资源无关方面，"息壤"升级了算力调度技术，接入多方、跨域、异构算力，使用户以任务 提交方式使用算力资源，无需感知底层算力差异。 其次，在训推框架无关方面，"息壤"自研了异构训推框架。胡志强表示，这个框架可以兼容国产算力和 英伟达算力，提升了国产算力的性能，目前已完成120多个模型的适配，实现一个模型多框架运行，让 使用者无需纠结AI框架选择。 在AI工具无关方面，"息壤"推出标准化AI开发流水线，将AI工具链抽象和封装成流水线，打造开箱即 用的AI开发环境，令开发者无需关注工具差异。 运营商财经（官方微信公众号yyscjrd）—— 主流财经网站，一家全面覆盖科技、金融、证券、汽车、 房产、食品、医药、日化、酒业及其他各种消费品网站。 胡志强表示，2025年被称为人工智能的应用元年，未来所有应用必将生于AI、长于AI，成于AI。但 是，AI大规模落地还面临三大难题：一是如何高效获取匹配自身业务需求的算力 ...

Group 1 - The core point of the article is the announcement of the "Xirang" integrated intelligent computing platform, which currently has a total computing power of 62 EFLOPS [1] - The platform consists of 35 EFLOPS of self-owned computing power and 27 EFLOPS of cooperative computing power [1] - There are 18 intelligent computing centers established in various cities and industries [1]

Group 1 - Beijing is accelerating the integration of digital economy and artificial intelligence, focusing on advanced technologies and optimizing the digital economy ecosystem [1] - The "One District, Three Centers" strategy is being advanced to enhance data governance and promote the construction of a data trading system [1] - China Telecom is leveraging its integrated advantages of "cloud + network + data + AI + applications" to support the digital transformation across various industries [1] Group 2 - China Telecom is rapidly building computing power infrastructure, including a computing scheduling platform and a large model training platform [2] - The industry is shifting from a "hundred model war" to optimizing cloud inference service efficiency, emphasizing the need for a collaborative approach across hardware, software, models, applications, and professional services [2] - The China Academy of Information and Communications Technology will focus on building a technical service system for the engineering delivery of large models to support healthy industry development [2] Group 3 - A talent cultivation mechanism is needed in the digital economy era, emphasizing the importance of university-enterprise cooperation for key technological breakthroughs [3] - Beijing Telecom and Beijing University of Posts and Telecommunications have established a joint laboratory for industry data intelligent labeling to support AI model training [3] - Beijing Telecom is showcasing innovations in AI, computing power, and data elements, having built a comprehensive AI service system to provide digital transformation tools for various industries [3]