每经记者｜刘旭强 每经编辑｜刘艳美 一个越来越明显的共识是，AI产业正从早期的技术模型探索，进入以"应用创新""产业深耕"为核心的新阶段，深度赋能政务、医疗、工业制造、交通、能源 等关键领域，持续催生新业态、新模式。 破解算力难题 "AI PC"加速走进现实 "现在智算算力需求非常旺盛，租GPU做AI推理，得提前两个星期排队预约。"郑纬民表示，曾经由"模型训练"主导的算力需求，如今已经转向"推理"，简单 说就是AI从"学习成长"阶段，进入了"给人干活"的实用阶段，而这也催生出爆发式的算力缺口。 要让AI高效"干活"，选对芯片是关键。相较于CPU，更擅长高效矩阵运算和高速数据读写的GPU脱颖而出。 不过高效的GPU也面临"烦恼"，其内存主要用于存储两类数据，一是训练所得的模型参数，二是推理过程中产生的中间结果。随着Kimi等支持数百万字上下 文的模型走红，AI大模型趋向"数据更多、模型更大、上下文更长"，但这也带来了极高的推理负载，导致用户量激增时频繁宕机。 郑纬民以热门论文的推理需求为例，"十万个用户同时查询，如果每个用户都单独存储文章原文和推理中间结果，AI必然不堪重负。" 郑纬民提到，业内的破解思路是"公 ...

各位朋友，大家好。 2025年已经接近尾声，在这个关键时刻，我想向大家公布一个消息，并分享一段我个人的分析。 在过去的一年中，无论是我个人还是我们团队的研究都表明，即将过去的2025年，只是一个转型之年、 过渡之年，即便人工智能等新生事物逐渐进入了我们的生活，但其实并没有动我们的"奶酪"。 而即将到来的2026，才是真正能带来巨大颠覆，或者至少带来动荡的一年。 需要指出一点，上面的结论并不是危言耸听，而是基于严谨的研究。下面，就来说说我的看法。 正如上文提到的，过去一年只是过渡之年。随着今年年初DeepSeek在中国横空出世，以及过去几年一 系列人工智能工具，如ChatGPT逐渐进入我们的生活，2025年其实是人工智能真正进入我们的生活、职 业以及各行各业的元年。 在此过程中，我们目前使用的多是大语言模型，而明年开始，大行动模型将登堂入室，这也是香港全球 化中心2026年度十大预测之一的内容。 大行动模型指的是什么？其实就是智能体。 虽说当前的大语言模型已经足够令我们惊艳，但它归根结底只是和我们对话，输出通过思考和样本收集 得到的结论，没有办法帮我们完成实际的工作。 与过去的机械化、电器化、自动化革命相比，人 ...

在全球经济不确定性加剧的背景下，新兴市场特别是亚太地区跨境企业面临多重金融服务挑战：本地支付壁垒高企、全球流动性割裂、多市场资金管理能 见度下降。特别是对于广大中小型企业，这些障碍不仅制约其市场拓展，更拉大了与大型跨国企业的竞争差距。 时势迫人，经过移动支付、超级应用这两个成长阶段，金融科技产业需要加速定义自己的3.0时代。 蚂蚁国际首席执行官杨鹏在2025年新加坡金融科技节指出，今日本地区金融科技的核心使命，不仅是连接交易，更是让所有市场和所有企业能克服不可避 免的长期动荡，获得"经济增长的真实驱动力"和"过去仅属于跨国企业的全球触达能力"。 破碎、不稳定、地区和体系壁垒林立的跨境支付长期以来是全球化的关键瓶颈。蚂蚁国际聚焦于推进公私营部门和跨领域合作，在不同地区和新旧支付体 系之间建立纽带，形成一个合规协同、低摩擦、高效率的全球支付生态，以此为起点逐步弥合不同商业生态间的鸿沟。 在促进数字账户的连接上，蚂蚁国际业务旗下 全球整合钱包网关服务Alipay+ 已与全球40家主流电子钱包和10多个国家二维码网络建立合作，覆盖100多 个市场。通过标准化接口，全球超过1.5亿商户无需逐一适配各国支付系统，即可无缝 ...

Core Insights - The core mission of fintech in the Asia-Pacific region is to connect transactions and enable all markets and enterprises to overcome long-term disruptions, achieving true economic growth and global reach previously reserved for multinational corporations [2][4]. Group 1: Financial Services Challenges - Emerging markets, particularly in the Asia-Pacific region, face multiple financial service challenges, including high local payment barriers, fragmented global liquidity, and decreased visibility in multi-market fund management [2]. - These challenges particularly hinder small and medium-sized enterprises (SMEs), widening the competitive gap with large multinational companies [2]. Group 2: Fintech 3.0 Vision - Ant Group is accelerating investment in a Fintech 3.0 vision characterized by an open ecosystem, multi-dimensional payment innovations, and trusted security technology [4]. - The company aims to establish a compliant, low-friction, and high-efficiency global payment ecosystem by promoting collaboration between public and private sectors across different regions and payment systems [5]. Group 3: Payment Innovations - Ant Group's Alipay+ has partnered with 40 major e-wallets and over 10 national QR code networks, covering more than 100 markets, allowing over 150 million merchants to seamlessly access major payment methods in the Asia-Pacific region [5]. - The introduction of real-time payment pathways between mobile payment systems and traditional banking systems is a significant breakthrough, enabling users to make instant payments at POS networks [6]. Group 4: AI and Payment Transformation - AI and blockchain technologies are driving profound changes in the payment sector, with Ant Group focusing on developing agentic payment solutions that assist in complex tasks beyond simple transactions [7]. - The company has launched Antom Copilot, an AI assistant for merchants, which simplifies onboarding and payment integration processes [8]. Group 5: Real-Time Payment and Forex Management - Real-time global payment settlement (RTP) is a key area for Fintech 3.0, allowing businesses to achieve 24/7 real-time settlements and improve liquidity management [9]. - Ant Group's Falcon AI model processed $15 trillion in funds in 2025, helping clients reduce forex hedging costs by 40% [9]. Group 6: Security Innovations - Ant Group introduced Alipay+ GlassPay, a smart glasses payment solution featuring advanced biometric identification methods, enhancing security in payment transactions [11][13]. - The company has developed the AI SHIELD TRANSFORMER risk control model, which processes over 720,000 transactions per minute with a high accuracy rate in identifying high-risk transactions [14]. Group 7: Privacy and Data Protection - Ant Group is investing in advanced privacy computing architectures to ensure data protection while sharing business insights securely [15]. - The establishment of the "Digital Wallet Guardians Alliance" aims to promote security technology sharing and best practices among partners [15]. Group 8: The Future of Fintech - The emergence of Fintech 3.0 represents a "super connector" era, integrating various payment tools and accounts, AI practices, transaction processes, and growth resources [16]. - Ant Group's new AI operational app, EPOS 360, aims to provide SMEs with insights and responsiveness comparable to large institutions, promoting inclusive growth [16].

Core Insights - Ant Group's chairman, Jing Xiandong, emphasized that generative AI will expand the boundaries of financial services, enabling personalized AI financial assistants for individuals and dedicated CFOs for SMEs [1][2] - The integration of generative AI in financial services will enhance efficiency and promote inclusive finance, with AI playing a significant role in risk management, fraud prevention, and customer experience optimization [1] - Ant Group is actively developing AI solutions to assist SMEs in navigating complex global payment environments, making it easier for them to access global payment capabilities [2] Group 1 - Generative AI will influence all areas of financial services, not just specific segments, due to the data and language-intensive nature of the industry [1] - The rise of payment agents is anticipated, with current capabilities already assisting in purchases and bookings, and future developments expected to involve more complex tasks [1] - Ant Group's AI applications for SMEs are designed to provide operational support, acting as both CFO and COO, and have been well-received in the market [2] Group 2 - Ant Group aims to create an open ecosystem with global partners to further support the development of SMEs through AI [2] - The company believes that AI will be a core engine driving financial inclusion and supporting the growth of SMEs in the future [2]

Core Viewpoint - The article discusses the upcoming conference focused on AI and its impact on various industries, highlighting the importance of AI in driving business transformation and innovation. Group 1: Conference Agenda - The conference will feature several key speakers discussing various topics related to AI, including its role in enterprise transformation, the empowerment of AI ecosystems through domestic computing power, and the development of general artificial intelligence and large model construction [2][3]. - Notable sessions include discussions on AI applications in the financial sector and the prospects of the intelligent economy, as well as exploring the digital foundation and potential of AI in enterprise-level applications [3]. Group 2: Participants and Contact Information - Attendees are encouraged to contact their respective sales representatives at Guotai Junan Securities for registration and further information regarding the conference [4].

Core Insights - The Chinese government emphasizes the importance of artificial intelligence (AI) and digital technology in promoting high-quality economic development, with a focus on standardization and international openness [1][2] - The development of AI is characterized by trends such as deep hardware-software collaboration, the emergence of intelligent agents, and the influence of open-source models [4][5][6] Group 1: AI Development Trends - Deep hardware-software collaboration is becoming a new paradigm for the development of large models, which is essential for rapid iteration and efficiency [4][5] - Intelligent agents are emerging as the primary form of large model applications, leading to the formation of an intelligent agent economy [5][6] - The acceleration of open-source models is driving the democratization of AI, lowering barriers for application deployment [5][6] Group 2: Industry Application and Penetration - AI's penetration into various industries follows a pattern where sectors with higher digitalization levels adopt AI first, gradually expanding to others [15][16] - Significant applications of AI are observed in industries like finance, healthcare, and transportation, which have established digital foundations [15][16] - The current disparity in AI application depth across industries is a reality, influenced by factors such as digital transformation readiness and resource availability [15][16] Group 3: Future Directions and Challenges - The future of AI development may see breakthroughs in reinforcement learning and the emergence of new model architectures that enhance efficiency and performance [12][13] - Intelligent agents are expected to evolve significantly, with the potential to independently complete complex tasks that currently require human intervention [14] - The coordination between industry levels and transformation routes, as well as between technological capabilities and actual needs, is crucial for advancing AI applications [15][16]

Core Insights - The core viewpoint emphasizes the transition of artificial intelligence from large models to intelligent agents, which are likened to interns or virtual employees, requiring a human-like approach in their management and integration into businesses [1][2]. Group 1: AI Transformation Framework - The framework for AI transformation in enterprises includes treating intelligent agents as "people" and clearly defining their roles [6]. - Initial pilot projects should focus on roles with clear processes and high human resource demands, such as contract review and market research [3]. - Establishing "virtual teams" where multiple intelligent agents collaborate to handle complex tasks is essential [4]. Group 2: Characteristics and Capabilities of Intelligent Agents - Intelligent agents possess four human-like characteristics: autonomous planning, continuous memory, tool usage, and collaborative division of labor, allowing them to exceed traditional Q&A interactions [2]. - The ability of intelligent agents to break down tasks into executable paths through reinforcement learning and task planning is highlighted as a core capability [2]. Group 3: Future Implications and Competitive Landscape - The emergence of intelligent agents is predicted to reshape competitive dynamics, enabling individuals to have multiple intelligent agents working as a "cyber assistant team" [2]. - Companies can achieve the efficiency of larger teams with fewer employees by leveraging intelligent agents, thus creating a new "intelligent agent economy" [2]. Group 4: Human Oversight and Education - Key decisions should retain human oversight to ensure control and management [6]. - There is a call for educational institutions to offer courses on "natural language programming" to cultivate management talent capable of collaborating with AI [6].

Core Insights - Google Cloud has launched Gemini Enterprise, an enterprise-level AI solution aimed at creating a comprehensive platform that integrates AI intelligence, enterprise data, business processes, and employees [1][2] - The annual revenue of Google Cloud has surpassed $50 billion, with over 65% of cloud customers utilizing its AI products, indicating a significant shift in the cloud market driven by AI [1][6] Gemini Enterprise Platform Ambitions - Gemini Enterprise is defined as an end-to-end platform unifying six core components rather than a standalone application [2] - The platform includes advanced Gemini models as its core, a no-code framework for building AI agents, pre-built specialized agents, secure connections to various data sources, centralized governance and security, and a marketplace for partner-built agents [2][3][4] Full-Stack AI Strategy - Google's full-stack AI strategy consists of four tightly integrated layers: infrastructure, research, model, and product/platform, which collectively form a competitive moat [4][5] - The infrastructure layer includes advanced TPU technology, while the research layer is supported by world-class teams driving innovation [4][5] Building an "Agent Economy" - Google emphasizes the importance of an open ecosystem, collaborating with partners to enhance product integration and accelerate the deployment of Gemini Enterprise [5] - The company is establishing an "agent economy" with standards for agent communication and secure transactions, enabling agents developed by different companies to collaborate and transact [5][41][42] Customer Case Studies - Early adopters of Gemini Enterprise, such as HCA Healthcare and Best Buy, have reported significant efficiency gains, with HCA saving millions of hours annually and Best Buy increasing customer service efficiency by 200% [6][32] - The platform's capabilities are already demonstrating value in real-world applications, showcasing its potential to transform workflows and enhance productivity [6][32] Global Enterprise Impact - Companies like Banco BV and Macquarie Bank have leveraged Gemini Enterprise for improved operational efficiency and customer service, highlighting the platform's versatility across industries [32][36] - The integration of Gemini models into various products is driving significant business outcomes, such as increased order volumes and enhanced customer experiences [43][44] Future Innovations and Collaborations - Google is collaborating with the LA28 Olympic Games to enhance event experiences through AI, showcasing the platform's scalability and potential for global impact [46][47] - The company is expanding its partner ecosystem to support AI stack development, ensuring a comprehensive approach to AI integration across various business functions [49]

Core Insights - The emergence of AI Agents is expected to significantly reshape industries, with 2025 being identified as the "Year of AI Agents" [1] - AI is transitioning from data-driven knowledge accumulation to a new phase characterized by strong reasoning capabilities, which will transform business processes and value chains [1][3] - By 2028, it is predicted that 33% of enterprise software applications will integrate AI Agents, leading to 15% of daily decision-making tasks being fully autonomous [1] Industry Trends - The AI Agent wave presents a crucial opportunity for China's industrial upgrade and digital transformation, fundamentally altering organizational work methods and structures [1] - The current evolution of Agent technology has progressed from workflow Agents to strong reasoning Agents and eventually to Multi-Agents, which collaborate to form networks [5] - The global market for AI Agents is projected to reach approximately $5.29 billion by 2024, with China's enterprise AI Agent application market expected to exceed $27 billion by 2028 and grow to $47.1 billion by 2030 [6] Challenges in Implementation - Despite the opportunities, there are significant challenges in the practical application of generative AI within enterprises, including internal resistance, communication barriers, and a lack of cross-departmental data sharing [7][8] - Many enterprises face skill gaps and difficulties in finding suitable application scenarios for large models, which complicates the deployment of AI technologies [7] - Successful integration of AI systems into business processes requires collaboration between traditional enterprises and AI technology companies, emphasizing the need for strategic alignment from top management [8] Company Strategies - Companies like Zero One Thousand focus on providing privatized deployment services for leading enterprises in specific industries, rather than pursuing a broad market coverage [9] - The "One Leader Project" strategy has been implemented, resulting in significant revenue growth, with 2025 revenue already surpassing the total for the previous year [9]