Core Insights - The 22nd Shenzhen International Electronics Expo and Embedded Expo (elexcon2025) opened on August 26, attracting over 10,000 professional visitors on its first day [1] - The expo focuses on "All for AI, All for Green," emphasizing embedded AI, edge computing, power semiconductors, and advanced packaging technologies to support industry development in the AI era [1] - The event featured the 7th China Embedded Technology Conference, with industry experts from Qualcomm, Renesas, ARM, and others discussing trends in embedded AI and its role in industry transformation [1] Industry Participation - Over 400 global embedded and electronic technology suppliers participated in the expo, including companies like Advantech, SEGGER, and National Technology [2] - Multiple forums during the expo addressed hot topics such as AI computing power, smart cockpit technology, and advanced packaging [2] - The "Kaifa Gala" event attracted over 10,000 engineers for learning and interaction, showcasing new technologies and products, which were well-received by attendees [2]

Core Insights - The 22nd Shenzhen International Electronics Expo and Embedded Expo (elexcon2025) opened on August 26, focusing on "All for AI, All for Green" and highlighting embedded AI, edge computing, and power semiconductors [1] - Over 400 global embedded and electronic technology suppliers participated, including notable companies like Guomin Technology, Lingdong Micro, and Zhiduo Crystal [1] - Qualcomm's product marketing director delivered a keynote on the role of AI, computing, and connectivity technologies in driving industry transformation [1] Group 1 - The expo aims to provide technical and supply chain support for industries related to AI, featuring over ten forums on topics such as AI computing power and advanced packaging [1] - Qualcomm introduced its "Qualcomm Leap" brand, which supports industrial IoT and cellular infrastructure with a product line achieving up to 100 TOPS of dense computing power [2] - The Qualcomm Leap brand encompasses a wide range of products and supports various operating systems, catering to the intelligent needs of multiple industries [2] Group 2 - Qualcomm has collaborated with over 70 partners to create an IoT application case collection, showcasing more than 150 cases across ten industries [2] - The focus on edge computing and connectivity is crucial for integrating AI technologies into industrial applications [2] - Qualcomm's solutions are designed to support long lifecycle product development and operations, enhancing the adaptability of IoT applications [2]

Group 1: Market Overview - The semiconductor value for data center servers is projected to reach $500 billion by 2030, indicating a rapidly expanding market [1] - IDC's VP Mario Morales predicts that data centers will become the fastest-growing sector in the semiconductor industry over the next five years [1] Group 2: Server CPU Landscape - The server CPU market is undergoing a silent architectural revolution with x86, ARM, and RISC-V architectures competing for dominance [2] - x86 architecture has historically dominated the server CPU market, primarily led by Intel, but this stronghold is beginning to weaken [3] Group 3: Market Share Dynamics - Intel's market share in server CPUs has been declining, from 91.1% in January 2021 to 72.7% in Q2 2025, while AMD's share has increased from 8.9% to 27.3% in the same period [5][8] - AMD's EPYC series has significantly contributed to its market penetration, with expectations to become the largest x86 CPU supplier in data centers by 2026 [8] Group 4: ARM Architecture Growth - ARM architecture has shown a growth rate of 70% since 2018, with cloud service providers increasingly adopting ARM-based CPUs for their efficiency and cost advantages [10][15] - Amazon AWS has been a pioneer in deploying ARM CPUs, with over 2 million units shipped since the launch of its Graviton series [12] Group 5: RISC-V Architecture Emergence - RISC-V architecture is gaining attention as a new path in server CPUs, although its current influence is less than that of ARM [17][18] - RISC-V's open-source nature allows for customized chip development, which could disrupt the traditional x86 and ARM markets [19][20] Group 6: New Entrants in the Market - Qualcomm is re-entering the server CPU market with a focus on ARM architecture, having previously exited due to ecosystem challenges [22] - Nvidia is making significant strides in the CPU space with its Grace CPU, designed to work closely with its GPUs for enhanced performance [25][26]

Group 1 - The 2025 Qualcomm Ventures - Sequoia China Startup Competition has announced its judging panel, featuring top investment institutions to support participating companies [2][3] - The judging panel includes representatives from ten leading investment firms, such as Qualcomm Ventures, Sequoia China, and Xiaomi Investment [3] - The competition aims to provide professional feedback and strategic guidance to finalists, enhancing their opportunities for future collaboration and funding [7] Group 2 - The competition focuses on cutting-edge technology sectors, including edge computing, robotics, IoT, and smart driving, among others [8] - It serves as an accelerator for companies from angel to Series B funding stages, helping innovative projects transition from technological breakthroughs to commercial success [10] - The final registration window is open for six more days, encouraging eligible companies to seize the opportunity for comprehensive support in growth and brand exposure [12]

2025中国算力大会日前在山西大同举行。"算力"一词频繁出现在各类新闻报道和讨论之中。它究竟是什么"力"？通算、智算、Token等一个个术语 又代表什么意思？ 图为2025中国算力大会开幕式现场。新华社记者 柴婷 摄 ■什么是算力？ 算力即计算能力，其实我们每个人都拥有算力，比如口算、心算和速算等，只是算力有限。当进行复杂计算时，就要借助计算工具来提升算力。 从最早的结绳计算、算盘计算，到发展出计算机计算等。 如今，算力是集信息计算力、网络运载力、数据存储力于一体的新型生产力，主要通过算力基础设施向社会提供服务。 通俗讲，算力就是对数据的处理能力。它看不见、摸不着，却时刻影响着你我的生活，比如智能音箱响应用户指令、电子支付实时清算、在线购 物个性化推荐、电影的特效处理等，背后都有算力支撑。 可以说，在数字经济时代，算力如同水、电、燃气等公共基础资源一样，走进千家万户、服务千行百业。 ■算力可以分为几类？ 在分类上，根据使用设备和提供算力强度的不同，中国信息通信研究院等机构将算力分为三类，即基础算力、智能算力和超算算力。 基础算力（通用算力），一般用于基础通用计算，比如办公、上网、看视频等，不需要太高的性能。 ...

截至2025年中报，金山云营业总收入43.19亿元、净利润-7.71亿元。 本文源自：金融界 8月26日，金山云(03896)盘中下跌2.05%，截至13:32，报7.66元/股，成交3.14亿元。 金山云控股有限公司以提供全球范围内的云服务为主营业务，构建了一整套的云计算基础架构和运营体 系，服务领域涉及互联网、公共服务、数字健康、金融等多个领域。公司运用大数据、人工智能和边缘 计算等高新技术，为500余家优质客户提供超过150种解决方案，并于2020年和2022年分别在美股和港股 完成上市。 作者：行情君 ...

Group 1 - The latest stock price of Kela Software is 20.47 yuan, up 1.09% from the previous trading day, with an intraday high of 20.79 yuan and a low of 20.29 yuan, and a trading volume of 470,000 hands, amounting to 966.7 million yuan [1] - Kela Software operates in the software development industry, focusing on software product development and technical services in the fintech sector, headquartered in Beijing, and involved in multiple technology areas including edge computing and Huawei concepts [1] - According to the latest announcement, a major shareholder, Ningbo Kela Shenghe Investment Management Partnership, has released a pledge of 2 million shares on August 25, 2025 [1] Group 2 - On the day of the report, the net outflow of main funds was 30.11 million yuan, with a cumulative net outflow of 108 million yuan over the past five trading days [1]

Core Concept - The 2025 China Computing Power Conference highlighted the significance of "computing power" as a new form of productivity that integrates information computing, network transmission, and data storage capabilities [1][3]. Definition of Computing Power - Computing power refers to the ability to process data, which is essential in various applications such as smart speakers, electronic payments, personalized online shopping, and special effects in movies [3]. - It is categorized into three types: basic computing power, intelligent computing power, and supercomputing power [3][4][5]. Types of Computing Power - Basic computing power is used for general tasks like office work and internet browsing, requiring standard performance [3]. - Intelligent computing power is designed for complex data analysis and AI tasks, efficiently handling large datasets and adapting strategies for different AI scenarios [4]. - Supercomputing power is provided by high-performance computing clusters, primarily used for scientific research and engineering calculations [5]. Growth of Computing Power in China - China's total computing power has been growing at an annual rate of approximately 30%, with a significant increase in demand for intelligent computing driven by rapid AI development [5]. - As of June 2025, China's intelligent computing power is projected to reach 788 EFLOPS, accounting for 32% of the global total, positioning the country as the second-largest in the world [5]. Measurement of Computing Power - The capability of computing devices is measured using FLOPS (floating-point operations per second), with various scales such as MFLOPS, GFLOPS, TFLOPS, PFLOPS, and EFLOPS indicating increasing computational strength [6]. - For example, the supercomputer "Frontier" has a peak performance exceeding 1 EFLOPS, illustrating the vast difference in computational capacity compared to human capabilities [6]. Terminology in Computing Power - Tokens are the smallest data units used in text processing, essential for language models to break down text for computational purposes [7][8]. - The daily consumption of tokens in China surged from 100 billion to over 30 trillion within a year and a half, reflecting the rapid growth of AI applications [8]. Edge Computing - Edge computing is a concept that has emerged recently, focusing on data collection and analysis occurring close to the data source rather than relying solely on centralized cloud resources [9].

Core Viewpoint - Donghua Software Co., Ltd. has partnered with Inspur Information to develop a smart hydrology "Four Prevention" comprehensive management platform, enhancing flood prevention capabilities through advanced technology [1] Group 1: Technology and Innovation - The platform is built on an edge computing solution, integrating Donghua Software's flood control system and specialized hydrological algorithm models [1] - The combination of computational power and algorithms allows for a forecast accuracy error margin of less than 5% [1] Group 2: Operational Impact - Flood warning times have been advanced to 2-4 days, providing critical lead time for disaster prevention and response [1] - The system enables detailed pre-simulation and scientific planning for flood control, optimizing resource allocation during emergencies [1]

Core Viewpoint - The company has achieved mass production in thermal management solutions for energy storage and edge computing servers, indicating a strong demand for computing power in the market [1] Group 1: Product Development - The company has successfully mass-produced liquid cooling modules, air cooling modules, and other thermal management products such as vapor chambers and heat pipes [1] - The development, accumulation, and certification of these technologies will require a certain period [1] Group 2: Market Trends - There is a confirmed growth trend in the demand for computing power, which the company is actively addressing by collaborating with solution providers to advance the research and certification of server thermal management products [1]