欢迎大家 点击【预约】 按钮 本文重点 观点来自： 6 月 3 日本周二直播 【 强 烈建议直接看】 本段视频精华，逻辑更完整 文字版速览 一、财务表现与业绩结构拆解 预约 我 下一场直播 "NIMs"（NVIDIA Inference Microservices）成为关键转折点，标志着从大模型阶段迈向Agent阶段。未 来AI不仅用于聊天交互，而是具备感知、推理与执行能力的智能体，将在实际业务中承担更多职责。 NIM提供标准化、容器化AI微服务，加快部署效率，支撑企业级AI应用落地。 "AI工厂"的概念代表数据中心角色的根本性转变——未来不再只是处理计算任务，而是批量生成AI智能 体，作为企业的"虚拟员工"运行于各类场景中。这一趋势背后，是英伟达软硬一体化生态能力的系统体 现。 三、中美市场影响与全球交付逻辑 英伟达未正面详细回应出口管制影响，但表示正以极快速度向全球客户交付新产品，包括中国市场在 内。这说明公司正在调整产品结构，以合规方式保障中国业务的持续性。 英伟达2025财年第一季度营收达260亿美元，同比增长262%，环比增长18%；毛利率78.9%，同比提升 12.6个百分点，环比提升1.8个百分 ...

Group 1 - The core theme of the conference was "Tide Rising in the East, New Quality Leading" with a significant turnout of over 5000 participants, including more than 200 speakers from various sectors [1][2] - The conference marked a record high in the number of sub-forums, attendees, and listed companies participating, indicating strong interest and engagement from the investment community [2] - The chairman of Guotai Junan Securities emphasized that investing in China is increasingly seen as a more certain opportunity due to government policies aimed at boosting confidence and economic growth [4] Group 2 - The macroeconomic outlook for the second half of 2025 suggests continued positive policy adjustments, with potential for further fiscal and monetary easing [7] - The chief strategist highlighted a clear trend towards a "transformation bull market" in Chinese stocks, driven by improved investor sentiment and reduced impact from valuation contractions [8] - Investment opportunities are identified in emerging technologies, cyclical finance, and themes such as AI, domestic brands, and regional economies, reflecting a shift in focus from real estate to broader economic strength [9][10]

Core Viewpoint - Guotai Haitong Securities is optimistic about China's "transformation bull market," with a strategic outlook favoring 2025 as the market evolves towards new opportunities in emerging technologies and cyclical finance [1][5]. Macro Perspective - Guotai Haitong Securities anticipates that macro policies may continue to marginally increase, especially after July, with a potential for comprehensive interest rate cuts in the second half of the year [2][3]. Market Strategy - The firm believes that the "transformation bull market" in the Chinese stock market is becoming increasingly clear, driven by improved investor understanding of economic conditions and a shift in the main concerns from economic cycles to declining discount rates [5]. - Investment opportunities are identified in emerging technologies as the main focus, with cyclical finance seen as a dark horse [5]. Investment Opportunities - Key areas of investment include: - Financials and high-dividend stocks, focusing on brokers, banks, and insurance companies, as well as companies with stable dividend yields and monopolistic advantages [5]. - Emerging technology sectors, particularly in internet, media, gaming, robotics, pharmaceuticals, and military industries [5][6]. - Revitalization of cyclical consumption, with attention to tight supply-demand dynamics in materials like rare earths and chemicals, as well as new consumer demands in retail and cosmetics [5]. Hong Kong Market Outlook - The Hong Kong stock market is expected to trend upwards in the second half of the year, driven by strong performance in the technology sector, particularly due to AI-driven growth [8]. - The Hang Seng Technology Index is highlighted as a key area of focus, with expectations of improved return on equity (ROE) and valuation uplift [8]. Convertible Bonds Market - The convertible bond market may experience a bull market due to supply-demand mismatches, with low interest rates driving demand for attractive investment options [10]. - Historical trends suggest that as risk appetite increases, the market may see a gradual upward trajectory in convertible bonds over the coming years [10].

Group 1 - The core viewpoint of the article emphasizes that the Chinese stock market is entering a "transformation bull" phase, driven by a clearer understanding of economic conditions and a reduction in the marginal impact of valuation contraction [1][2] - The key drivers of the market include a decline in the risk-free interest rate and a systematic reduction in risk perception, which have historically hindered investor willingness to enter the market [2][3] - The article highlights the importance of timely and reasonable economic policies, as well as reforms in the capital market that focus on investor returns, which are crucial for changing the conservative attitudes of investors towards risk [2][3] Group 2 - Emerging technology is identified as a main investment theme, while cyclical finance is seen as a dark horse opportunity. The article suggests that the real issues in the Chinese economy are not solely in real estate but in broader economic stability and innovation [3] - Recommendations for investment include sectors such as financial services (brokerages, banks, insurance), high-dividend companies (telecom operators, highways, public utilities), and emerging technologies (internet, media, robotics, pharmaceuticals, military) [3] - The article also points to new consumption trends and cyclical recovery in commodities, recommending investments in rare earths, chemicals, retail, and cosmetics, as well as themes like AI, regional economic development, and self-sufficiency in technology [3]

2024年，被称为是"大语言模型（LLM）元年"，然而，2025年，聚光灯正迅速转向。 "毫无疑问，2025年是大模型智能体元年。"近日，Gartner研究副总裁孙志勇在接受包括《IT时报》在内的媒体采访时表示，智能体已成为全球最热门的科 技话题，其热度甚至超越了炙手可热的大模型本身，一个由智能体驱动的商业模式创新时代正加速到来。 孙志勇认为，2025年标志着"大模型驱动的AI智能体元年"和"生成式AI的变现元年"。这一转变意味着从基础模型开发转向实际的、可创收的应用，而AI智 能体正是LLM能力的主要"出口"。 Gartner预测，到2028年，人类可读的数字界面交互中将有20%由AI智能体完成，约15%的日常业务决策可由AI智能体自主完成，预计三分之一的企业级 软件将集成AI智能体。 从"辅助"到"执行"的跃迁 到底什么是智能体？尽管智能体似乎已经成为2025年人人必谈的AI名词，但这个概念在市场中，显然仍有"泡沫"。 "智能体并非横空出世，严格意义上讲，世界上第一个智能体诞生于1960年。"孙志勇指出，"之所以现在智能体突然爆火，是因为大模型智能体，也就是 大模型驱动的智能体，让智能体变得更加聪明了"。 ...

Group 1: Market Overview - The Hong Kong IPO market has seen a significant rebound, with fundraising amounting to HKD 145 billion in the past year, led by major companies like CATL and Midea Group [1] - AI remains a focal point in the market, with high valuations and substantial financing, exemplified by Fourth Paradigm's market cap exceeding RMB 29.4 billion on its first trading day [1] - The AI sector is experiencing a trend of differentiation, shifting focus from large model capabilities to more practical AI agent capabilities, emphasizing deep integration with application scenarios [1] Group 2: Decision AI Market Growth - The decision AI application market in China is poised for explosive growth, with a projected market size of RMB 34.5 billion in 2024 and a remarkable compound annual growth rate (CAGR) of 36.2% over the next five years [2] - The marketing and sales decision AI platforms are particularly prominent, expected to grow from RMB 20.3 billion in 2024 to RMB 94.4 billion by 2029, reflecting a 36.5% CAGR [2] Group 3: Company Positioning - DeepYuan Intelligent leads the decision AI application market with a 2.6% market share, significantly ahead of other competitors, attributed to its robust business model and innovative product matrix [3] - The company has developed a leading product matrix, including AlphaDesk for public traffic acquisition, AlphaData for private customer management, and Deep Agent for AI agent solutions, creating a comprehensive intelligent decision-making engine [4] Group 4: Innovation and R&D - DeepYuan Intelligent prioritizes R&D as a core driver for sustained leadership, increasing its R&D investment from RMB 46.9 million in 2022 to RMB 56.3 million in 2024, with R&D expenditure as a percentage of revenue rising from 8.6% to 10.5% [7] - The company has established three key technological pillars: an innovative full-stack marketing algorithm architecture, breakthrough large-scale real-time data processing technology, and advanced collaborative multi-agent technology based on large language models [8] Group 5: Financial Performance - DeepYuan Intelligent is one of the few AI companies that has achieved profitability for three consecutive years, with a strong customer base including nearly 90 Fortune 500 companies [10] - Despite a revenue decline to RMB 537.9 million in 2024 due to macroeconomic pressures, the company maintains a high customer retention rate of over 85% for its core products [10][13] - The company is expanding its global service network, with overseas markets contributing 13.8% to total revenue by 2024, indicating a steady increase in international influence [10][11] Group 6: Future Outlook - The shift in the intelligent agent paradigm from process-oriented to results-driven problem-solving is expected to drive significant revenue contributions for DeepYuan Intelligent's AI agent business by 2025 [13] - The company is well-positioned for sustainable growth, leveraging its strong customer relationships, global service capabilities, and technological advancements to navigate macroeconomic challenges [13][14]

Core Viewpoint - The Chinese stock market is entering a "transformation bull" phase, driven by a clearer understanding of economic conditions among investors and a shift in the main contradictions from economic cycle fluctuations to a decline in discount rates, particularly the reduction in risk-free rates and systemic risk awareness [1][2]. Group 1: Market Dynamics - The key drivers of the market are the decline in risk-free interest rates and a systemic reduction in risk perception. The past three years saw high opportunity costs hindering investor willingness to enter the market, but recent monetary policies have improved liquidity [2]. - The long-term government bond yield has fallen below 2%, and deposit rates have dropped below 1%, indicating a substantial decrease in risk-free rates in the Chinese stock market [2]. Group 2: Investment Opportunities - Emerging technologies are a primary investment theme, while cyclical finance is seen as a dark horse. The focus is on sectors such as finance and high dividends, with recommendations for brokers, banks, and insurance companies, as well as companies with stable dividends and monopolistic advantages [3]. - New technology growth is emphasized, particularly in sectors like internet, media, robotics, pharmaceuticals, and military, driven by competition between China and the U.S. [3]. - The revival of cyclical consumption is anticipated, with a focus on sectors experiencing improved competitive dynamics and tight supply-demand conditions, such as non-ferrous metals and chemicals, along with new consumer demands in retail and cosmetics [3]. Group 3: Thematic Recommendations - Recommendations include investments in AI applications, which are expected to accelerate demand for computing power, benefiting internet giants and infrastructure spending [4]. - The diversification of products and applications in embodied intelligence is highlighted, with a focus on core components and lightweight materials that benefit from large-scale production [4]. - The rise of domestic brands in consumer goods is noted, with recommendations for sectors like beauty, IP toys, and pet services that resonate with new consumer demographics [4]. - Regional economic policies, such as the Western Development strategy and the upcoming Hainan free trade zone, present opportunities in infrastructure, specialty raw materials, and tourism [4]. - The restructuring of global technology supply chains suggests a positive outlook for advanced semiconductors, domestic computing, and foundational software [4].

Core Insights - The article discusses the rapid transformation of enterprise business through generative AI, emphasizing the need for deeper integration of AI into core business processes rather than just superficial assistant functions [2] - Microsoft has introduced the concept of an "open agent network" at its Build conference, highlighting the collaborative role of AI agents across various organizational levels to enhance productivity [3] - The future workforce will transition to "agent managers," overseeing AI teams to boost productivity rather than being replaced by AI [4][5] AI Transformation Stages - The evolution of enterprise AI will occur in three stages: 1. Human with Assistant: Each employee has an AI assistant to improve work efficiency [7] 2. Human with Agent Team: AI agents act as "digital colleagues" under human guidance for specific tasks [8] 3. Human-led Agent Operations: Humans set the direction while AI agents autonomously manage business processes [9] Market Dynamics - 84% of Chinese leaders view 2025 as a critical year for strategic restructuring, with the pace of implementation depending on leadership commitment [9] - Microsoft aims to create an "open agent world" rather than competing with other firms like Tencent, focusing on collaboration [10] Business Model and Adoption - Microsoft's strategy revolves around building an "open agent network" supported by Azure AI Foundry, which integrates over 1,900 models for creating AI systems [10] - Over 60,000 enterprises globally are utilizing Azure AI Foundry to develop AI applications, showcasing Microsoft's unique position in providing mature commercial solutions [10]

Core Insights - The era of AI experimentation has ended, and competitive advantage for enterprises now relies on tailored AI applications and quantifiable business outcomes [2] - AI technology is transitioning from experimental phases to core business applications, with significant investments expected in the next two years [3] Group 1: AI Implementation and Challenges - Over half of CEOs are actively deploying AI agents, but only 25% of AI projects achieve expected returns, indicating a fragmented technology landscape [3] - The complexity of IT environments poses a significant barrier, with medium-sized enterprises averaging over a thousand applications across various heterogeneous systems [3] - Key factors for successful enterprise AI deployment include data quality, proprietary vertical models, and security governance [4] Group 2: Evolution of AI Agents - AI agents are evolving from mere conversational tools to productivity engines capable of autonomous decision-making and complex task execution [4] - IBM's AI agents have demonstrated significant efficiency gains, such as saving over $5 million annually in HR queries and reducing procurement contract cycles by 70% [4] Group 3: Data and Automation - The activation of unstructured data is crucial, as 90% of enterprise data is unstructured, and organizations lacking AI-ready data practices risk abandoning over 60% of their AI projects by 2026 [6] - IBM's methodology enhances accuracy by 40% through entity-value extraction and integrates structured and unstructured data governance [6] Group 4: AI Model Strategy - IBM advocates for flexible, secure, and efficient smaller models rather than large, all-encompassing ones, emphasizing a "small but beautiful" approach for initial AI agent deployments [7]

Group 1 - The China Academy of Information and Communications Technology, in collaboration with major tech companies like ICBC, Baidu, Tencent, Alibaba, and Huawei, has unveiled the first part of the series of standards titled "Technical and Application Requirements for Software Engineering Agents" focusing on "Developing Intelligent Agents" [1] - The standard emphasizes two core areas: technical capabilities and service capabilities, establishing a comprehensive technical specification system for software development agents, marking a significant step in China's exploration of AI agent standardization [1] - AI agents are increasingly recognized as a new digital workforce capable of executing complex instructions and communicating fluently with humans, gathering data, reasoning logically, and autonomously solving multi-step problems [1] Group 2 - The global AI agent market is projected to grow from $5.1 billion in 2024 to $47.1 billion by 2030, with a remarkable compound annual growth rate of 44.8% [3] - 2025 is recognized as the pivotal year for AI agent development, transitioning from traditional auxiliary tools to enabling human autonomous decision-making, significantly reshaping productivity and human-machine interaction [3] - Notable examples of AI agents include Microsoft's ability to analyze business emails and OpenAI's o1/o3 model for processing complex orders, showcasing their strong capabilities in autonomous decision-making and task execution [3] Group 3 - Leading domestic companies are actively investing in the AI agent field, with Tencent upgrading its large model knowledge engine to the "Tencent Cloud Intelligent Agent Development Platform (TCADP)," facilitating easier task decomposition and action planning for users [4] - JD.com operates over 14,000 AI agents internally and is leveraging its JoyAgent platform to deliver AI productivity to various industries [4] - The launch of the widely discussed general AI agent Manus reflects the market's high interest and expectations for intelligent agents [4] Group 4 - The CEO of Toinvest, Zou Qing, highlighted that AI agents are driving profound changes in the industry landscape through technological standardization [4] - The release of the new standard not only sets a technical benchmark for the industry but also signals that leading companies like Baidu and Alibaba have successfully transitioned AI agents from the concept validation stage to a new phase of industrial application [4]