Market Performance - The Hang Seng Index (HSI) rose by 106 points or 0.4%, closing at 25,540 points after a low of 25,257 points during the day[1] - The Shanghai Composite Index (SSE) fell by 9 points or 0.2%, closing at 3,900 points, with a total turnover of 7,304 billion yuan[1] - The Shenzhen Component Index increased by 39 points or 0.3%, closing at 13,316 points, with a turnover of 1.05 trillion yuan[1] Monetary Policy - The People's Bank of China conducted a reverse repurchase of 189.8 billion yuan, resulting in a net injection of 110.5 billion yuan for the day[1] - The Renminbi's central parity rate was adjusted up by 20 points to 7.0753 against the US dollar[1] - The Hong Kong Monetary Authority followed the US rate cut, lowering the discount window base rate by 0.25% to 4%[2] Corporate Updates - Hillstone Technology reported a total camera module sales of 41.755 million units in November, a month-on-month decrease of 11.9% but a year-on-year increase of 0.4%[3] - Sunny Optical Technology announced a smartphone lens shipment of 119 million units in November, a month-on-month decrease of 2.3% but a year-on-year increase of 7.5%[3] - The automotive lens shipment increased by 3.4% month-on-month and 69.4% year-on-year, totaling 12.634 million units[3]

Group 1 - Alibaba Cloud officially launched Function Compute AgentRun, a one-stop Agentic AI infrastructure platform that integrates Serverless features with AI-native application scenarios, helping enterprises optimize costs and efficiency with an average TCO reduction of 60% [1] Group 2 - Pop Mart announced the appointment of Wu Yue, President of LVMH Greater China, as a non-executive director, effective from December 10, 2023, following the resignation of He Yu due to other work commitments [2] Group 3 - Luckin Coffee established a new company in Yunnan with a registered capital of 10 million USD, focusing on food sales, production, catering services, and tea product manufacturing, fully owned by Luckin Coffee Trading (Hong Kong) Co., Ltd [3]

Core Insights - Alibaba Cloud officially launched AgentRun on December 10, which is a one-stop Agentic AI infrastructure platform based on its leading Function Computing (FC) technology [1] - The platform integrates the extreme elasticity, zero maintenance, and pay-as-you-go features of Serverless with AI-native application scenarios, helping enterprises achieve significant cost and efficiency optimization [1] - The average Total Cost of Ownership (TCO) is reduced by 60% with the implementation of this platform [1]

Core Insights - The core challenge of AI applications going global is the timeliness, scalability, and stability of model inference services [2][18]. Group 1: Event Overview - The 36Kr WISE2025 Business King Conference, recognized as an annual technology and business trendsetter, took place in Beijing on November 27-28 [3]. - This year's WISE is not a traditional industry summit but an immersive experience using "tech short dramas" to convey insights [4]. Group 2: AI Application Trends - GMI Cloud's VP of Engineering, Qian Yujing, presented on the efficiency upgrade of AI applications going global, focusing on breaking through computing power and evolving inference architecture [6][11]. - The company, a North American AI Native Cloud service provider and one of NVIDIA's first six Reference Cloud Partners, emphasizes the importance of globalizing computing power and demand for AI applications [7][8]. Group 3: Market Dynamics - The AI application market is experiencing exponential growth, with a significant increase in monthly active users for Chinese AI applications overseas [15]. - Over 90% of knowledge workers in the U.S. are now proficient in using AI tools, indicating a strong adoption of AI [15]. - The demand for AI services in regions like the Middle East and Latin America has reached a high level, suggesting that user education for overseas markets is largely complete [16]. Group 4: Challenges in AI Globalization - Key challenges in AI globalization include the timely delivery of services, scalability, and stability, particularly due to the rapid technological iterations in AI [18][20]. - The need for companies to keep pace with technological advancements poses a significant challenge for enterprises [21]. Group 5: GMI Cloud's Solutions - GMI Cloud is investing $500 million to build a 3 million card AI factory in Asia in collaboration with NVIDIA [14]. - The company has developed three product lines: computing hardware, cluster management, and inference services, catering to various AI enterprise needs [14]. - The Cluster Engine and Inference Engine are designed to address different customer segments, with the former focusing on complex applications and the latter on lightweight, end-user applications [25][29]. Group 6: Inference Engine Features - The Inference Engine supports global deployment and automatic scaling across clusters and regions, addressing the challenges faced by companies when their traffic peaks [30][31]. - It features a three-layer architecture for resource scheduling, with two main scheduling methods: queue-based and load balancing-based [31]. - The core features of the Inference Engine include global deployment, elastic scaling, high availability, and unified workload management [33][35][36]. Group 7: Future Outlook - By 2026, the paradigm of AI globalization is expected to shift from a one-way technology output to a global value resonance, emphasizing a two-way empowering ecosystem [43]. - The transformation will involve a new cycle of value creation, where computing power, technology, demand, and applications interact globally [43].

Core Insights - Huawei has successfully implemented a Serverless NPU elastic computing scheduling technology in collaboration with Industrial and Commercial Bank of China, significantly enhancing computational efficiency [1][3] Group 1: Technology Implementation - The Serverless NPU elastic computing scheduling technology can reduce the startup time for inference services of a trillion MoE large model to a level of seconds, achieving over a 10-fold increase in startup efficiency compared to traditional scheduling methods [1][3] - This technology represents a shift in computing supply models from "long-term binding" to "on-demand usage," indicating a more flexible approach to resource allocation [1][3]

Core Viewpoint - Akamai Technologies has acquired Fermyon, a serverless WebAssembly company, to enhance its edge computing capabilities and improve application performance while reducing costs compared to traditional cloud-native applications [1][2]. Group 1: Acquisition Details - The acquisition of Fermyon will allow Akamai to integrate Fermyon's cloud-native WebAssembly function-as-a-service with its globally distributed platform, enabling enterprises to build edge-native applications [1][2]. - Fermyon is recognized for its leadership in serverless functions and WebAssembly, and it actively participates in the open-source community, which Akamai plans to continue supporting [2]. Group 2: Strategic Implications - Akamai aims to deepen the integration between its edge functions platform and its performance and security products, facilitating faster and easier application development, deployment, and security at the edge [3]. - The acquisition is expected to provide developers with a broad range of cloud-native and serverless options, enhancing innovation and execution of lightweight code at the edge [2]. Group 3: Financial Impact - Akamai anticipates no material impact on its financial guidance for 2025 as a result of the Fermyon acquisition [4].

Core Insights - The article discusses the launch of OceanBase's first AI database, seekdb, which allows developers to create AI applications with minimal coding, indicating a significant increase in database demand and new requirements for data management [1][6][12] - OceanBase aims to integrate AI and data management, focusing on global expansion and open-source contributions to enhance its market presence [1][8][10] Group 1: AI and Database Evolution - AI is democratizing technology, enabling non-computer science professionals to develop applications and use databases, leading to a substantial increase in database users [3][5] - The concept of hybrid search is identified as a critical turning point for databases, requiring them to support both human and AI interactions [1][3] Group 2: Product Launch and Features - The seekdb database supports unified hybrid search across various data types, including vector, text, scalar, and spatial data, and is compatible with over 30 mainstream AI frameworks [5][12] - The product is open-sourced under the Apache 2.0 license, allowing global developers to freely use, modify, and extend it, which aligns with OceanBase's strategy to foster a more developer-friendly environment [6][12] Group 3: Globalization and Market Strategy - OceanBase has surpassed 4,000 global customers since its commercialization in 2020, with an annual growth rate exceeding 100% for five consecutive years [8] - The company is focusing on localizing its technology and services to better integrate into international markets, emphasizing the importance of adapting to local needs [9][10] Group 4: Future Outlook and Challenges - The development of AI technologies is seen as cyclical, with expectations of both peaks and troughs, but ultimately leading to mature applications that will transform the software ecosystem [2][6] - OceanBase recognizes the need for a robust infrastructure to support the scaling of AI applications, highlighting the importance of simultaneous advancements in various technological areas [7][10]

Group 1 - The article discusses the seemingly counterintuitive choice of Claude Code to use a grep-only approach instead of vector indexing, which has sparked debate among developers [3][5]. - Critics argue that this decision represents a technological regression, while supporters highlight its alignment with Unix philosophy and the redefinition of what constitutes a good tool [3][5]. - Claude Code's approach emphasizes real-time search without maintaining a persistent code index, which has been shown to outperform other methods in performance tests [5][49]. Group 2 - The essence of state is explored, distinguishing between stateful and stateless systems, with examples illustrating the impact of state on system design [9][10]. - Historical context is provided, tracing the origins of stateless design from mathematical functions to the Unix pipeline philosophy, which emphasizes simplicity and composability [11][14]. - The advantages of stateless design include composability, natural parallelism, simplicity, and testability, making it a preferred choice in modern computing [30][34][36]. Group 3 - The article discusses scenarios where state is necessary, such as in gaming, user interfaces, and resource management, emphasizing the importance of context in design choices [41][47]. - A mixed strategy is suggested, where stateless computation is combined with stateful storage, allowing for flexibility and efficiency in system architecture [43][46]. - The core insight is that the choice between stateless and stateful design is not a matter of technical belief but an engineering trade-off, focusing on managing necessary state wisely [47]. Group 4 - In the AI era, Claude Code's choice reflects a shift in understanding intelligence, prioritizing predictability and behavior over mere functionality [54]. - The article concludes that simple tools endure, and the design that embraces "forgetfulness" offers greater freedom and adaptability in a rapidly evolving technological landscape [55].

Investment Rating - The report does not explicitly state an investment rating for the gaming industry Core Insights - The gaming industry has evolved into a complex ecosystem that integrates technology, art, and culture, with increasing demands for immersive experiences and operational stability [4] - The report emphasizes the importance of efficient architecture and data-driven decision-making in addressing the challenges faced by game developers [4][5] - The book serves as a practical guide, offering solutions across various technical domains, from architecture design to AI-driven art creation [5][16] Summary by Sections Chapter 1: Game Architecture - Discusses different game types and their architectural needs, highlighting the importance of selecting optimal architecture based on gameplay requirements [7][54] - Explores the evolution of game architecture from local hardware to cloud-based solutions, enabling dynamic and scalable game environments [54] Chapter 2: Game Networking - Examines the impact of network quality on player experience and presents strategies for building low-latency, high-availability networks [8] - Introduces advanced technologies such as edge computing and serverless architecture for future network optimization [8] Chapter 3: Game Security - Analyzes various network security architectures and proposes a comprehensive defense system to protect against cyber threats [9] - Highlights the necessity of robust security measures to ensure long-term operational success in the gaming industry [9] Chapter 4: Game Downloading - Focuses on the importance of download speed and player experience, detailing methods for building high-availability architectures [10] - Discusses practical techniques for managing traffic surges during peak times [10] Chapter 5: Game Databases - Addresses the critical role of data management and recovery strategies in maintaining game integrity [11] - Provides insights into minimizing data loss and ensuring rapid recovery through cloud-native tools [11] Chapter 6: Game Big Data - Explores the significance of data-driven decision-making in game operations and marketing [12] - Discusses the construction of real-time data lakes and the use of tools like Flink for player behavior analysis [12] Chapter 7: Game Art - Investigates the transformative impact of AI-generated content on game art and design [13] - Analyzes the potential and limitations of AI in enhancing creative processes within the gaming industry [13] Chapter 8: Game Content Review - Discusses the importance of content moderation in maintaining a safe and compliant gaming environment [14] - Explores various methods and technologies for effective content review [14] Chapter 9: The Future of the Gaming Industry - Reflects on the technological evolution of the gaming industry and anticipates future trends such as cloud gaming and AI integration [15][52] - Envisions a future where gaming becomes a gateway to the metaverse, emphasizing the need for innovative governance and content moderation strategies [52]

Core Offering - Beam 提供了一种开源替代方案，用于部署无服务器 AI 工作负载，类似于 Modal [1] - 通过添加 Python 装饰器，即可轻松部署 AI 应用程序 [1] - 只需添加 `beam-client`，添加装饰器，即可将任何工作流程转换为无服务器端点 [1]