Group 1 - The company will announce its first quarter results next Friday, and investors are advised to pay attention [1] - Major tech companies including Alibaba, Tencent, JD.com, and NetEase are set to report their earnings next week, along with other Chinese concept stocks like Tencent Music and Beike [2] - Economic data to watch includes the US April CPI and PPI figures, as well as various speeches from FOMC members [2][3] Group 2 - Upcoming economic events include the release of US retail sales, PPI, and GDP data, as well as initial jobless claims and consumer confidence indices [3][6] - The National Energy Administration will publish electricity consumption data around the 15th of each month [6] - Vice Premier He Lifeng will visit Switzerland from May 9 to 12 for talks with Swiss leaders and will also co-host the 10th China-France High-Level Economic and Financial Dialogue in France [6]

Group 1 - Hong Kong tech stocks experienced a midday surge, with Xiaomi Group (01810.HK) and Tencent Holdings (00700.HK) both rising over 2% [1] - Alibaba (09988.HK) and Sunny Optical Technology (02382.HK) increased by 1.6% [1] - SenseTime (00020.HK), AAC Technologies (02018.HK), and Kuaishou (01024.HK) all saw gains of over 1% [1]

Summary of Key Points Core Viewpoint - The Hong Kong stock market experienced significant net selling from Northbound trading, totaling 78.66 billion HKD, with notable net sell-offs in major tech stocks like Tencent and Xiaomi, while banks and food delivery services saw net buying [1]. Group 1: Northbound Trading Activity - Northbound trading recorded a net sell of 78.66 billion HKD, with 41.78 billion HKD from Shanghai Stock Connect and 36.88 billion HKD from Shenzhen Stock Connect [1]. - The most bought stocks included China Construction Bank (00939), Meituan-W (03690), and CNOOC (00883) [1]. - The most sold stocks were Tencent (00700), Xiaomi Group-W (01810), and Alibaba-W (09988) [1]. Group 2: Individual Stock Performance - Xiaomi Group-W saw a net sell of 20.26 billion HKD, influenced by changes in its vehicle promotion strategy related to smart driving features [5]. - Tencent experienced a net sell of 27.8 billion HKD, amid concerns over ongoing U.S.-China trade tensions and market volatility [6]. - Alibaba-W faced a net sell of 9.55 billion HKD, reflecting broader market trends and investor sentiment [6]. Group 3: Sector Insights - China Construction Bank received a net buy of 6.68 billion HKD, supported by recent monetary policy measures aimed at enhancing liquidity in the market [4]. - Meituan-W attracted a net buy of 3 billion HKD, with analysts predicting a stable competitive landscape in the food delivery sector [5]. - CNOOC had a net buy of 2.13 billion HKD, with first-quarter earnings showing a decline but better than expected due to favorable oil prices [5].

Group 1 - Southbound funds recorded a net sell of 78.66 billion HKD [1] - Tencent Holdings, Xiaomi Group-W, and Alibaba-W faced net sells of 27.83 billion HKD, 20.26 billion HKD, and 9.55 billion HKD respectively [1] - China Construction Bank saw a net buy of 6.68 billion HKD, bucking the trend [1]

Investment Rating - The report maintains a "Buy" rating for Tencent Holdings [4] Core Viewpoints - Tencent showcased 46 products at the SPARK 2025 game launch event, highlighting its comprehensive capabilities in development, publishing, and investment [1] - The report emphasizes Tencent's strong research and development capabilities, particularly in adapting popular PC game IPs into mobile formats, which is expected to yield ongoing benefits [1][2] - Tencent's investment in external companies is anticipated to enhance its IP portfolio and improve its capabilities in the overseas PC gaming market, contributing to long-term growth [1][2] Summary by Sections Product Development and Investment - Tencent's game publishing strategy has evolved to include a focus on investment, showcasing successful overseas ventures and a clear direction for future IP and PC game development [1] - The report notes that the Chinese mobile gaming market is expected to maintain a significant global share, with Tencent's investments likely to bolster its competitive edge in the PC gaming sector [1] User Engagement and Market Position - Tencent's gaming user engagement remains robust, with a reported 66.5% share of industry usage time, indicating a strong operational advantage [2] - The upcoming release of new titles is expected to mitigate growth pressures from high baseline comparisons in the latter half of 2025 [2] Financial Projections - The report projects Tencent's Non-IFRS net profit to reach 254.72 billion RMB in 2025, with a slight upward revision from previous estimates [2][10] - The target price for Tencent's stock has been adjusted to 605 HKD, reflecting confidence in its sustained performance across gaming and digital content sectors [2][4]

Core Insights - The multi-modal large model industry focuses on deep learning models capable of processing, understanding, and generating various types of data, including text, images, audio, and video, enabling complex and intelligent tasks [1] - The industry has a wide application potential across various sectors such as natural language processing, image recognition, speech recognition, intelligent driving, and medical imaging diagnosis [1] Industry Overview - The multi-modal large model industry chain is complex, encompassing hardware facilities, software development, and various model types, including CLIP, BLIP, and LLaMA, among others [1] - The industry is divided into three layers: the foundational layer (hardware and basic software), the model layer (various types of multi-modal large models), and the application layer (industry-specific applications) [1] Cost Structure - The training costs for mainstream domestic large models range from tens of millions to hundreds of millions of dollars, with major companies like Baidu, Alibaba, and Tencent investing over $200 million [3][5] - Startups like Kimi and DeepSeek have managed to reduce training costs to between $30 million and $60 million through technological optimizations [3] - Cloud hosting costs are significantly influenced by model scale, with major companies leveraging their own cloud platforms to reduce costs [3] Development History - The global large model industry has evolved through several phases: early exploration (1956-2005), rapid growth (2006-2019), the rise of large models (2020-2022), and the current phase of widespread application starting in 2023 [6] Computational Demand - The demand for computational power in AI is increasing, with larger models requiring exponentially more computational resources; for instance, the GPT-3 model requires 3640 PF-days of computation and at least 10,000 GPUs [9] - As model parameters increase, the computational investment needed grows significantly, influenced by model architecture, optimization efficiency, and hardware capabilities [9]

AI为核心的新一轮科技创新浪潮已至，AI将与互联网一样，带来新一轮范式革命，人类社会将进一步加速发展。我们精心准备，重磅推出【智能时代专 题】，目前已规划106篇深度原创研报，将全方位梳理AI产业、技术、代表性公司等发展历史、现状、趋势，展望智能时代未来图景，挖掘投资机会。 凭借在移动互联网时代打下国民级基础设施生态优势，不仅赋予腾讯强大网络效应与深厚护城河，更让腾讯成为AI在中国大规模应用与普及的超级载体 与核心驱动力，在AI时代将进一步放大，具备巨大成长潜力。 智能时代专题：腾讯（HK：00700） 免费版2,672字，预计阅读5分钟 完整版18,349字 智能时代第二季，重磅升级，更前瞻，更硬核，支持专题/单篇付费解锁 一、腾讯过去十多年，通过互联网特别是移动互联网时代创新、布局，凭借微信为主的生态系统，深刻融入到中国经济、社会、政务服务体系，从单纯的 互联网平台，发展成不可或缺的国民级基础设施。 截至2024年，微信与WeChat 合并月活账户数增至13.85 亿（+3%），覆盖中国几乎所有互联网用户，成为中国绝大多数网民数字生活入口。 微信支付与庞大小程序生态，构成重要商业基础设施，服务零售、餐饮、 ...

Group 1 - The Nasdaq Golden Dragon China Index experienced a preliminary decline of 0.8% [1] - Notable Chinese concept stocks included Tencent ADR rising by 4.8%, Xiaomi ADR increasing by 1.2%, and Li Auto rising by 1.2% [1] - Other stocks such as Netease and Alibaba saw modest gains of up to 0.5%, while Xpeng fell by 0.2%, JD.com dropped by 0.8%, Baidu decreased by 1.8%, NIO declined by 4.3%, Pony.ai fell by 5.2%, and Yum China dropped by 7.3% [1]

Investment Rating - The report maintains a "Buy" rating for Tencent Holdings with a target price of HK$681, indicating an expected share price increase of 47.9% from the current price of HK$460.40 [2][7]. Core Insights - The 2025 Spark Game Conference showcased a total of 46 game titles, including 26 new games, which is a 35% increase from the previous year. Key titles such as HoK: World, Roco Kingdom, Valorant, and The Finals received significant attention from viewers [1]. - Tencent's strength in self-development and publishing capabilities was highlighted, with a diverse range of game genres presented from both domestic and global studios [1]. - The report emphasizes the positive user feedback and engagement during the conference, suggesting a constructive outlook for Tencent's gaming segment [1]. Summary by Sections Self-Development Games - Tencent presented 18 games from its in-house studios, including notable titles like HoK and CODM, which achieved over 1 billion downloads in 2024 [4][5]. Publishing Games - A total of 13 new licensed titles were showcased, including Path of Exile 2 and The Finals, indicating a robust pipeline for Tencent's publishing segment [5]. Investment Games - Eight games from Tencent's investees were highlighted, including Phantom Blade Zero and Dune: Awakening, showcasing Tencent's strategic investments in the gaming industry [6]. Valuation - The target price of HK$681 is derived from a sum-of-the-parts (SOTP) valuation approach, applying various price-to-earnings (P/E) ratios to different segments, including online games, advertising, social networks, fintech, and cloud services [7].

Core Insights - The article discusses the multi-modal large model industry chain, highlighting its complexity and the various layers involved, including the foundational layer, model layer, and application layer [1][3]. Industry Overview - The multi-modal large model industry chain consists of three main layers: foundational (hardware and basic software), model (various types of multi-modal large models), and application (industry-specific large models) [1]. - Key players in the foundational layer include Intel and NVIDIA, while foundational software participants include Huawei, Tencent, and Unisoc [3]. Regional Distribution - The multi-modal large model industry is concentrated in major cities such as Beijing, Shanghai, Suzhou, Hangzhou, and Shenzhen, with a strong presence of upstream companies in Beijing and Shenzhen [4]. - Upstream companies in Beijing include Cambricon, Lenovo, and Inspur, while Shenzhen hosts Unisoc, HiSilicon, and ZTE [4]. Revenue Insights - The revenue distribution in China's multi-modal large model sector shows a concentration among leading companies, with Alibaba Cloud generating over 110 billion yuan, accounting for approximately 15% of its group's revenue [7]. - Huawei Cloud and Tencent Cloud follow, with revenues of 68.8 billion yuan and 63.6 billion yuan, respectively, each contributing around 5-7% to their parent companies [7]. Cost Structure - The training costs for mainstream large models in China range from tens of millions to hundreds of millions of dollars, with major players like Baidu, Alibaba, and Tencent investing over 200 million dollars [10]. - Startups like Kimi and DeepSeek manage to lower their training costs to between 30 million and 60 million dollars through technological optimization [10]. Product Cost Details - The training and cloud hosting costs for various multi-modal large models are detailed, with notable examples including: - Pangu Model (Huawei): ≥ 100 million USD - Wenxin Model (Baidu): ≥ 300 million USD - Mix Yuan Model (Tencent): ≈ 250 million USD - Xinghuo Model (iFlytek): = 80 million CNY [12].