Core Insights - The report emphasizes the importance of a new generation underground network for coal mines, designed based on "SDN (Software Defined Network) + Slicing" architecture, which aims to enhance communication reliability, bandwidth, and operational efficiency in the context of increasing automation and intelligence in coal mining operations [5][27][36]. Group 1: Requirements for Communication Networks - The development of intelligent coal mining necessitates a robust communication network that can support high reliability, low latency, and simple maintenance, as well as high bandwidth to accommodate various applications such as remote control and safety monitoring [8][12]. - Intelligent coal mining applications include high-definition video monitoring, safety monitoring, wireless communication, and remote control, all of which require a network capable of handling large data volumes and ensuring real-time communication [9][10][11]. Group 2: Current Status and Challenges of Underground Networks - The existing underground networks consist of various systems, including wired communication networks, safety monitoring networks, and industrial Ethernet, which often leads to a complex and fragmented network environment [18][19][20]. - Challenges faced by current underground networks include insufficient bandwidth, network storms affecting communication reliability, and difficulties in network maintenance due to the complexity of multiple independent systems [24][25]. Group 3: New Generation Underground Network Solutions - The new generation underground network is defined as a comprehensive carrier network that integrates various intelligent applications, utilizing advanced technologies such as IPv6 and SDN to achieve unified data transmission and operational efficiency [26][27]. - The architecture of the new generation network is structured into three layers: the surface core, underground backbone, and underground access, ensuring effective data flow and management [27][28][29]. Group 4: Key Features and Technologies - The new generation underground network features include multi-service integration, ultra-wide coverage, intelligent operation and maintenance, dedicated network slices, and high reliability, which collectively enhance the operational capabilities of coal mining [31][32][33]. - Key technologies supporting the new network include network slicing, SRv6, clock synchronization, and Wi-Fi 6, which address the limitations of traditional networks and facilitate the deployment of intelligent applications [33][34].

Investment Rating - The report does not explicitly state an investment rating for the industry or company Core Insights - The emergence of large models has triggered an exponential growth in global computing power demand, with AI computing requirements increasing by 300,000 times from 2012 to 2023, and expected to grow another 500 times in the next decade [11][12] - Huawei Cloud's Ascend AI Cloud Service is positioned as the best cloud-based full-stack computing service for the era of large models, providing comprehensive support for model training, inference, and application development [17][18] - The AI industry is experiencing a paradigm shift, moving from specialized models to general-purpose models capable of handling diverse applications through extensive pre-training on massive datasets [8][12] Summary by Sections Section: Global Computing Demand - The demand for computing power has surged due to the rise of large models, with video generation models like Sora requiring significantly more computing resources compared to traditional models [11][12] - The report highlights that the scale of datasets will increase from a few terabytes to 100 terabytes, and the token length for models will expand from thousands to hundreds of thousands [11] Section: Full-Stack Computing Services - Huawei Cloud's Ascend AI Cloud Service integrates cluster computing, computing engines, and AI development frameworks to provide stable and reliable full-stack computing support for large model training and inference [12][18] - The service offers various computing usage, management, and deployment modes to cater to diverse business needs [19][21] Section: Business Innovation Focus - The report emphasizes that enterprises require full-stack computing services to focus on business innovation, leveraging accumulated training and operational experience to avoid redundant problem-solving [14][19] - The Ascend AI Cloud Service supports a wide range of AI frameworks and models, facilitating rapid application development and deployment [68][70] Section: Performance and Recovery - The service boasts rapid fault recovery capabilities, with industry practices showing that cluster faults can be detected in one minute, diagnosed in five minutes, and recovered in ten minutes [36][40] - The report indicates that the average interruption for large model training in the industry occurs every 2.8 days, with recovery times traditionally taking much longer [36][41] Section: Ecosystem and Community - The AI Gallery serves as a one-stop community service platform, fostering an open community for AI development and collaboration [69] - The D-Plan ecosystem partnership program aims to build a collaborative AI ecosystem, providing partners with comprehensive support for training, technology, marketing, and sales [70]

Investment Rating - The report does not explicitly state an investment rating for the industry Core Insights - The modernization of financial core systems is essential for banks to enhance operational efficiency and meet the evolving demands of digital finance, regulatory compliance, and customer service [5][10][15] - Financial institutions are increasingly focusing on integrating financial technology to drive innovation and improve service delivery, aligning with national strategic goals [5][10] - The shift from centralized to distributed core systems is becoming a consensus among financial institutions, driven by the need for resilience, reliability, and agility [28][30] Summary by Sections Chapter 1: Challenges and Trends in Modern Financial Core Systems - The banking sector is transitioning from an inward-focused improvement model to an outward-focused ecosystem integration model, emphasizing the release of data value to enhance financial services [10][11] - Key challenges include differentiated customer service, product innovation, regulatory compliance, stable operations, and cost structure optimization [5][10] - The concept of "new quality productivity" highlights the central role of financial technology in driving business innovation and service model transformation [5][10] Chapter 2: Goals and Key Designs for Modern Financial Core Systems - The target architecture for modern financial core systems integrates infrastructure modernization, technology architecture, data architecture, security, compliance, and innovation [15][16] - The business architecture includes defining core business goals, rules, processes, and stakeholder needs to enhance operational efficiency [16][18] - The technical architecture aims for resilience, trustworthiness, agility, and intelligence, utilizing cloud computing, distributed systems, and AI technologies [25][28] Chapter 3: Methodologies and Pathways for Core System Construction - The report outlines a six-dimensional new quality system for building modern financial core systems, including resilient architecture, full-stack trust, open integration, engineering practices, smooth migration, and agile intelligence [6][10] - A "four stages and ten steps" implementation methodology is proposed to guide the construction of modern financial core systems [6][10] Chapter 4: Solutions for Distributed New Core Systems - The report discusses collaborative solutions developed by Huawei and leading domestic software vendors to support the transition to distributed core systems [6][10] - Emphasis is placed on the need for a robust digital infrastructure to facilitate the integration of core business with modern distributed capabilities [6][10] Chapter 5: Practical Cases of Modern Financial Core System Construction - The report includes case studies from various financial institutions that illustrate the practical challenges and solutions encountered during the core system transformation process [10][30]

Industry Investment Rating - The report highlights the transformative potential of AI, particularly in the context of AI-driven data centers (AI DC), which are becoming the core infrastructure for enterprises aiming to achieve intelligent transformation [5][6][9] Core Viewpoints - AI is an irreversible trend, not a passing trend, and it will reshape every industry and organization [4][22][23] - The future of data centers is defined by AI, with AI DC being a comprehensive reconstruction of traditional data centers, shifting from cost centers to innovation centers [5][6][46] - AI DC will play a critical role in supporting AI model training, inference, and application, becoming the cornerstone of enterprise intelligent transformation [5][6][9] Chapter Summaries Chapter 1: AI World Vision and Macro Drivers - AI is a major direction that cannot be stopped, with generative AI and large models driving a new industrial revolution [16][18] - The global AI market is rapidly expanding, with significant investments in generative AI, despite a slight decline in overall AI investment [16] - AI is expected to trigger a once-in-a-century transformation, reshaping industries and driving economic growth [18][20] Chapter 2: All in AI - Generative Business Systems - Enterprises face both uncertainties and certainties in adopting AI, with over 70% of leaders expecting AI to significantly impact their business within five years [29] - The key to successful AI adoption lies in building an enterprise-level AI architecture that can handle the instability of large models and the "impossible triangle" of generalization, specialization, and economy [31][33] - Enterprises should focus on application scenarios, data, models, and computing power to achieve value creation and business transformation [34][36] Chapter 3: Development and Changes in Data Centers - Data centers are evolving into AI DC, which are designed to support AI model training, inference, and application, with a focus on high-performance computing and energy efficiency [49][51] - AI DC differ from traditional data centers in terms of business load, computing power type, and cooling methods, with a shift towards xPU-centric architectures and liquid cooling [51][52] - AI DC are categorized into three types: ultra-large, large, and small, each serving different needs and facing unique challenges [56][57][58] Chapter 4: Typical AI DC Planning and Construction - Ultra-large AI DC are primarily used for foundational model pre-training, facing challenges such as power supply, reliability, and effective computing power [59][60] - Large AI DC are used by industry leaders for secondary training and central inference, with a focus on optimizing inference performance and resource utilization [60] - Small AI DC are designed for lightweight inference and AI applications, requiring flexible deployment, easy maintenance, and security [61][62] Chapter 5: AI DC Construction and Development Initiatives - The report proposes four key initiatives for AI DC development: moderate超前 construction, intensive and green development, open collaboration, and building three foundational bases to accelerate AI adoption [15] - AI DC will be redefined by AI, providing diverse computing power and enabling the innovation of AI-native applications [47][65]

Investment Rating - The report does not explicitly state an investment rating for the ICT services and software industry Core Insights - Digitalization, intelligence, and low-carbon development are identified as certain trends, with artificial intelligence (AI) being the biggest opportunity in the next decade, projected to grow AI computing power by 500 times by 2030 compared to 2020 [2][10] - The integration of AI into various sectors such as digital government, intelligent finance, and smart manufacturing is expected to accelerate the digital transformation of industries [2] - The report emphasizes the need for the ICT industry to prepare for challenges and demands in six key areas, including the integration of computing and networking, the development of new professional service capabilities based on large models, and ensuring a reliable service level agreement (SLA) experience [2][10] Summary by Sections Section 1: Accelerating Green Network Development - The report highlights the necessity for a green transformation of ICT infrastructure to support sustainable business development, addressing issues like outdated equipment and high energy consumption [11][12] - A three-tier management framework for green governance, planning, and execution is proposed to guide the green transformation strategy [12][15] Section 2: Data Center Green Low-Carbon Leadership - The demand for AI model training computing power is expected to grow exponentially, leading to increased energy consumption in data centers [19][21] - New energy efficiency metrics are being introduced, shifting from PUE to more comprehensive sustainability indicators [19][21] Section 3: Intelligent Data Management - Data is recognized as a critical asset for AI model development, necessitating robust data storage and management systems to handle the increasing complexity and volume of data [27][28] - The report discusses the importance of ensuring data security and efficient data flow to support AI applications [27][30] Section 4: Transitioning to Business-Centric Operations - The report outlines a shift from network-centric to business-centric operational models, driven by the need for automation and intelligent operations [34][40] - New operational models are expected to enhance customer experience and operational efficiency, with a focus on agile business recovery capabilities [42][46] Section 5: 5G-A and Business Scenario Integration - The integration of 5G-A with business scenarios is anticipated to create new value and enhance customer experience through tailored service offerings [46][47] - The report notes that operators are moving towards experience-driven business models, emphasizing the need for network optimization to meet diverse customer demands [47]

Investment Rating - The report does not explicitly state an investment rating for the industry or company Core Insights - The HiSec Endpoint intelligent terminal security system addresses the increasing network security risks faced by enterprises due to digital transformation and the evolving nature of cyber threats [9][14] - The system integrates big data security analysis capabilities to form a multi-layered adaptive defense loop, enhancing threat perception, detection, and response [14][29] - The HiSec Endpoint Agent is designed for lightweight deployment and efficient data collection, ensuring minimal resource consumption while providing comprehensive threat awareness [18][19] Summary by Sections Introduction - The report discusses the challenges in network security construction, emphasizing the vulnerabilities of terminal security systems in the face of advanced threats [9][11][12] - It introduces the HiSec Endpoint intelligent terminal security system as a solution to these challenges, highlighting its innovative architecture and capabilities [14] Advantages and Value - The HiSec Endpoint system offers comprehensive threat perception, precise threat detection, and superior threat response capabilities [17] - The system's lightweight agent collects data efficiently, reducing noise and ensuring data integrity for effective threat analysis [18][19] - It features advanced detection engines, including a third-generation antivirus engine and a threat tracing graph engine, significantly improving the detection rate of unknown threats [26][27] Working Principles - The system employs full-stack data collection to monitor terminal activities in real-time, facilitating effective threat detection and response [30][31] - It utilizes a dynamic behavior analysis approach to identify malicious activities, enhancing the overall security posture against various attack vectors [32][33] - The HiSec Endpoint system incorporates advanced mechanisms for ransomware detection and response, ensuring comprehensive protection against such threats [45][56] Threat Detection and Response - The system's proactive measures include automated threat analysis and response, enabling quick containment of security incidents [29][45] - It features a lightweight backup and recovery mechanism to restore encrypted files to their pre-attack state, ensuring data integrity [56] - The HiSec Endpoint system also addresses cryptocurrency mining threats through specialized detection engines that monitor suspicious behaviors [57][62]

2024 版 通信网络 2030 构建万物互联的智能世界 目 录 06 型超势 04 未来网络场景 | --- | --- | |-----------------------------------------------------------------------------------------------------------------------------------------------------|-------| | | | | 2.1 下一代人机交互网络：以人为中心的超现实体验 | | | 2.1.1 XR：虚实的完美结合，自然的交互体验…………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………… 08 | | | 2.1.2 裸眼 3D：逼真的影像再现，全新的视觉体验 ……………………………………………………………………………………………………………………………………… 09 | | | 2.1.3 数字触觉：多维的体感交互，可触摸的互联网 ……………… ...

2024 版 云计算 2030 构建万物互联的智能世界 目 录 01 宏观趋势 04 05 02未来行业场景 | --- | --- | |----------------------------------------------------------------------------------|----------------------------------------------------------------------------| | | | | 2.1 制药：AI 颠覆式创新，创新药设计成功率提升 10 倍， 10 年研发周期减半 | ................................06 | | 2.2 气象：数据驱动地球解码器，预报计算时间缩短 1 万倍 | ...............................................................07 | | 2.3 金融：高频模拟 Now-Casting，30% 经济指标即时精准预测 | ............................................ ...

2024 版 数据中心 2030 构建万物互联的智能世界 数据中心 2030 探索未来数据中心 引领智能时代 an I Wall I Wall I want of the may be and the may be and the ... 01 序言 汪涛 创新涌现，拥抱智能时代 在 AI 大模型训练过程中，当模型大到一定规模之后，性能会发生突变，开始呈现指数级快速增长，科 学界称这个现象为"涌现"。正是这个性能的突变，让人工智能的发展阶段从感知理解世界到生成创 造世界，这也造就了 ChatGPT 的火热，催生了面向行业的数百个 AI 大模型的出现。今天，"百模千态" 正走向每一个行业、每一个场景、解决客户每一个问题，加速千行万业的智能化转型。人工智能的"涌 现"时刻即将出现，人类社会也将迎来一个波澜壮阔的智能时代。 迈入智能时代，最大的需求是算力，最关键的基础设施是数据中心。根据华为《智能世界 2030》报告 预测，2030 年，人类将迎来 YB 数据时代，全球通用计算算力将达到 3.3ZFLOPS(FP32)，AI 算力需求 激增，2030 年将达到 864 ZFLOPS（FP16）。算力需求十年百倍的增 ...

2024 版 | --- | --- | --- | --- | --- | --- | |-------|-------|-------|-------|-------|-------| | | | | | | | | | | | | | | | ICT | 服务 | | | | | 构建万物互联的智能世界 前言 通讯行业从 2G 逐步走向 5G，ICT 服务和软件行业也经历了标准化，工具化，数字化的代际升级； 随着 Gen AI，数字孪生等新技术的兴起，从数字化到智能化已成为行业共识，2030 年，人工智 能将无所不在，基础设施智能泛在感知成为刚需，大模型逐步走向 AGI 像人类一样思考，服务模 式将从围绕"人"逐步走向"机器"，企业营销和赋能的方式也将变得更加实时敏捷 ...... 未来十年，千行百业的智能化转型类似 20 世纪的工业革命，Gen AI 是和蒸汽机，电灯一样伟大 的发明，第一次让机器像人脑一样学习和思考，这是跨时代的改变生产力的历史进程，将开启每 个企业 / 家庭 / 个人工作和生活的新时代…. 目 ICT 服务与软件 2030 愿景及核心技术 20 宏观趋势与展望 04 ICT 服务与 ...