【导读】吴泳铭首次系统阐述了通往超级人工智能（ASI）的三阶段演进路线 中国基金报记者 赵心怡 "AGI并非AI发展的终点，而是全新的起点。AI不会止步于AGI，它将迈向超越人类智能、能够 自我迭代进化的超级人工智能（ASI）。"阿里巴巴集团CEO、阿里云智能集团董事长兼CEO 吴泳铭表示。 在2025年云栖大会上，吴泳铭发表主旨演讲。他认为，实现通用人工智能AGI已是确定性事 件，但这只是起点，终极目标是发展出能自我迭代、全面超越人类的超级人工智能ASI。 吴泳铭首次系统阐述了通往ASI的演进路线，具体分为三个阶段："智能涌现"，AI通过学习海 量人类知识具备泛化智能；"自主行动"，AI掌握工具使用和编程能力以"辅助人"，是行业当 前所处的阶段；"自我迭代"，AI通过连接物理世界并实现自学习，最终实现"超越人"。 通往ASI将经历三个阶段 "ASI将以指数级的速度推动科技的飞跃，引领我们进入一个前所未有的智能时代。"吴泳铭表 示。 吴泳铭称，通往ASI之路的第一阶段是"智能涌现"，特征是"学习人"。 过去几十年，互联网 的发展为智能技术的崛起奠定了基础，实现了知识的数字化。大模型通过理解这些数字化的 知识， ...

Core Insights - The journey towards Artificial Super Intelligence (ASI) is divided into three stages: "Emergence of Intelligence," "Autonomous Action," and "Self-Iteration" [2][3][4] - Achieving Artificial General Intelligence (AGI) is seen as a certainty, but it is merely the starting point for the development of ASI, which aims to surpass human intelligence and enable self-evolution [2][3] Stage Summaries - **Stage 1: Emergence of Intelligence** This stage is characterized by AI's ability to learn from vast amounts of human knowledge, leading to generalized intelligence and conversational capabilities [3][4] - **Stage 2: Autonomous Action** AI moves beyond language interaction to perform real-world actions, capable of breaking down complex tasks and using tools to interact with both digital and physical environments. This stage is currently being experienced [3][4][6] - **Stage 3: Self-Iteration** In this final stage, AI connects with comprehensive raw data from the physical world and achieves self-learning capabilities. This is crucial for AI to surpass human capabilities [4][6] Future Implications - AI is expected to become the next generation operating system, with large models facilitating user interactions and application creation through natural language [6][7] - The computing paradigm is shifting from CPU-centric to GPU-centric AI computing, requiring significant energy and infrastructure to support this transition [6][7] - AI is predicted to replace energy as the most critical commodity, driving productivity across various industries [6][7] Human-AI Collaboration - The future will see a new collaborative model between humans and AI, enhancing productivity significantly. AI is expected to amplify human capabilities exponentially [7] - Historical technological revolutions have consistently unlocked new productivity, suggesting that the rise of ASI will similarly create new demands and opportunities for human advancement [7]

Group 1 - Alibaba's stock surged over 7% on September 24, reaching its highest level since October 2021, driven by CEO Wu Yongming's announcement regarding significant investments in AI and cloud computing [1] - Alibaba Cloud's energy consumption is projected to increase tenfold by 2032, indicating a substantial rise in computational power investment [1] - The launch of Qwen3-Max, Alibaba's largest and most capable AI model, has positioned it ahead of competitors like GPT-5-Chat in various benchmarks [1] Group 2 - The Hong Kong internet sector is expected to rebound after a period of negative sentiment, with reasonable valuations and high growth potential in AI technology [2] - The Hong Kong Stock Connect Technology ETF (159262) rose by 2.84%, with significant gains in key stocks such as ASMPT and Hua Hong Semiconductor [2] - Major AI companies like Alibaba, Tencent, and Xiaomi account for nearly 45% of the ETF's weight, highlighting the concentration of technology leaders in the market [2] Group 3 - The Hong Kong Stock Connect Technology ETF saw a significant increase in scale, growing by 518 million yuan over the past week, marking the highest growth among comparable funds [3] - The ETF's latest share count reached 3.885 billion, a record high since its inception, indicating strong investor interest [3] - Continuous net inflows into the ETF over the past 11 days totaled 693 million yuan, with a peak single-day inflow of 325 million yuan [3] Group 4 - The Hong Kong Stock Connect Technology ETF provides exposure to leading technology companies, capitalizing on the opportunities presented by the AI revolution [4]

Core Viewpoint - The ultimate goal of AI is to evolve into Super Artificial Intelligence (ASI), which will surpass human intelligence and capabilities, progressing through three stages: "intelligent emergence," "autonomous action," and "self-iteration" [1][4][9] Group 1: AI Development Stages - The first stage, "intelligent emergence," involves AI learning from human knowledge and achieving capabilities comparable to a top-level human in various fields [6][9] - The second stage, "autonomous action," allows AI to perform complex tasks in the real world, utilizing tools and coding to assist humans [7][9] - The third stage, "self-iteration," will enable AI to connect with raw data from the physical world, allowing it to learn and evolve independently, ultimately leading to ASI [10][12] Group 2: Strategic Pathways for ASI - Alibaba's AI strategy focuses on two main lines: the open-source initiative "Tongyi Qianwen" to build a vast developer ecosystem and the establishment of a comprehensive AI cloud service [2][20] - The company is investing 380 billion yuan over three years to enhance AI infrastructure, anticipating a tenfold increase in energy consumption by 2032 compared to 2022 [2][22] Group 3: AI's Impact on Society - ASI is expected to revolutionize various fields, solving complex problems in medicine, materials science, and sustainable energy, thereby significantly enhancing human productivity [4][24] - The future will see a collaborative relationship between humans and AI, with numerous agents and robots working alongside people, amplifying human capabilities exponentially [24][25] Group 4: Technological Innovations - Large models are predicted to become the next generation of operating systems, facilitating interactions between users and AI resources through natural language [14][16] - The shift from traditional CPU-based computing to GPU-centric AI computing is essential for meeting the demands of future AI applications [18][20]

Core Viewpoint - The Hong Kong stock market showed a collective rebound on September 24, with the technology sector performing notably well, driven by significant movements in key stocks like Alibaba and Huahong Semiconductor [3][5]. Market Performance - The Hong Kong Technology ETF (520860) rose over 2%, with constituent stocks such as Huahong Semiconductor increasing by over 7%, and both SMIC and Alibaba-W rising by over 6% [3][5]. - The ETF has seen a net inflow of funds for four consecutive days, indicating strong investor interest [5]. Company Developments - Alibaba's CEO, Wu Yongming, emphasized the importance of large models as the next generation operating system and AI cloud as the next computing paradigm, with plans for a significant investment of 380 billion yuan in AI infrastructure [4][5]. - The company anticipates a tenfold increase in energy consumption for its global data centers by 2032, reflecting a substantial escalation in computing power investment [4]. Industry Trends - Nvidia's plan to invest up to $100 billion in collaboration with OpenAI to build AI data centers is expected to significantly boost global computing power demand, further stimulating the technology sector [5]. - Analysts from Haitong International predict that leading Hong Kong tech companies will benefit from new technological breakthroughs in the second half of the year, particularly in the AI sector [7]. - Long-term growth potential in AI technology and new consumption sectors is highlighted as a key driver for the sector's performance, alongside improved pricing power from southbound capital [7]. Investment Insights - The Hong Kong Technology ETF tracks the CSI Hong Kong Technology Index, which includes major tech companies like Tencent, Alibaba, and Xiaomi, collectively accounting for nearly 30% of the index [8]. - The ETF is positioned to help investors capitalize on opportunities within the technology industry, particularly as AI investments continue to grow [8].

Core Insights - Alibaba Group's CEO, Wu Yongming, announced at the Yunqi Conference that by 2032, the energy consumption of Alibaba Cloud's global data centers will increase tenfold compared to 2022, indicating an exponential increase in computing power investment [1] - Following this announcement, Alibaba's stock price rose by 6.4% in the Hong Kong market, reaching its highest level since October 2021 [2] - Wu emphasized that large models represent the next generation of operating systems, and AI Cloud is the next generation of computing, with a shift from CPU-centric to GPU-centric computing paradigms requiring massive infrastructure [4] Investment Activity - Cathie Wood's Ark Investment Management has resumed its position in Alibaba, purchasing approximately $16.3 million worth of Alibaba ADRs across two ETFs on September 22, marking the first re-entry since September 2021 [5] - Ark Investment's initial investment in Alibaba dates back to 2014, shortly after the company's IPO, but there have been no recorded investments or proxy votes since September 2021 [5]

Core Viewpoint - The realization of Artificial General Intelligence (AGI) is seen as a certainty, marking the beginning of a journey towards Artificial Superintelligence (ASI), which aims to surpass human intelligence and achieve self-iteration [1][7][24]. Group 1: AI Development Stages - The evolution towards ASI is outlined in three stages: 1) Emergence of intelligence through learning human knowledge; 2) Autonomous action where AI assists humans; 3) Self-iteration, where AI learns from raw data in the physical world to surpass human capabilities [1][9][12]. - The current stage is characterized by AI's ability to assist humans, utilizing tools and programming capabilities [1][10]. Group 2: AI Infrastructure and Investment - Alibaba is investing 380 billion yuan over three years in AI infrastructure, with plans for further investments to support the transition to the ASI era [2][20]. - In the past four quarters, Alibaba has invested over 100 billion yuan in AI infrastructure and product development [2]. Group 3: Financial Performance - Alibaba Cloud's revenue grew by 26%, reaching a three-year high, with AI-related revenue accounting for over 20% of external commercial income [4]. - The capital market is increasingly betting on Alibaba's AI business, with significant investments from notable funds [5]. Group 4: AI as a New Operating System - Large models are expected to become the next generation operating system, with AI Cloud serving as the new computing platform [2][16]. - The future will see a shift from traditional software to AI-generated agents, allowing users to create applications using natural language [16][18]. Group 5: Future of AI and Human Collaboration - The emergence of ASI will redefine human-AI collaboration, with AI significantly enhancing human productivity and capabilities [23][24]. - The AI revolution is anticipated to unlock new levels of productivity, similar to past technological revolutions [23][24].

AI播客：换个方式听新闻 下载mp3 音频由扣子空间生成 周三，阿里巴巴(09988.HK)盘中一度涨超7%，股价创下4年新高，截至发稿报170港元/股，其美股ADR 夜盘也一度录得超5%的涨幅。 此前，该公司发布迄今为止规模最大、能力最强的模型Qwen3-Max。另外，阿里巴巴还宣布，计划加 大对人工智能基础设施的支出，以更好地与美国竞争对手抗衡。 23日晚间，阿里通义千问团队负责人透露即将发布6个新品，包括1个产品、2个开源模型、3个API接 口。24日凌晨开始，新品接连发布：第一个是开源的安全模型Guard，包含generative和streaming两个版 本，后者能够接近实时判断安全性；第二个是"千问旅行规划师"，可生成详细旅行计划并导出PDF；第 三个是支持多种语言的同声传译API qwen3-livetranslate；第四个是编程模型Coder的API升级，其核心优 势在于强大的Agent特性；第五个是视觉-语言模型Qwen3-VL，其旗舰模型已开源；第六个就是全新旗 舰产品模型Qwen3-Max，大幅提升编码和智能体能力。 据"通义千问Qwen"公众号，Qwen3-Max是一款规模和能力都极 ...

Core Viewpoint - The realization of Artificial General Intelligence (AGI) is seen as a certainty, marking the beginning of a journey towards Artificial Superintelligence (ASI), which aims to self-evolve and surpass human intelligence [1][7]. Group 1: Stages of AI Evolution - The evolution towards ASI consists of three stages: 1. "Intelligent Emergence," where AI learns from vast human knowledge to develop generalized intelligence [3][10]. 2. "Autonomous Action," where AI gains the ability to use tools and programming to assist humans, representing the current stage of the industry [4][11]. 3. "Self-Iteration," where AI connects with the physical world to achieve self-learning and ultimately surpass human capabilities [5][13]. Group 2: Strategic Pathways - Alibaba Cloud aims to become a "full-stack AI service provider" through two core strategies: 1. Committing to an open-source approach with Tongyi Qianwen, aspiring to create the "Android of the AI era" [5][19]. 2. Building a super AI cloud as the "next-generation computer" to provide global intelligent computing power [21][22]. Group 3: Investment and Infrastructure - Alibaba is actively pursuing a three-year plan with an investment of 380 billion yuan in AI infrastructure, with expectations of a tenfold increase in global data center energy consumption by 2032 compared to 2022 [5][25]. Group 4: AI's Impact on Society - The advent of ASI is expected to revolutionize human-AI collaboration, enhancing productivity significantly and creating new demands across various sectors [27][28].

Core Viewpoint - Alibaba has launched its largest and most powerful model, Qwen3-Max, at the 2025 Yunqi Conference, with plans to invest significantly in AI infrastructure amounting to 380 billion [2][4]. Group 1: AI Infrastructure and Investment - Alibaba is actively advancing its AI infrastructure construction with an investment of 380 billion, with intentions to increase this investment further [3][4]. - The energy consumption of Alibaba Cloud's global data centers is projected to increase tenfold by 2032 compared to 2022, indicating an exponential rise in computing power investment [4][6]. Group 2: Super Artificial Intelligence (ASI) Concept - The realization of Artificial General Intelligence (AGI) is seen as a certainty, but it is only the starting point; the ultimate goal is to develop Super Artificial Intelligence (ASI) that can self-iterate and surpass human capabilities [6][9]. - The path to ASI is divided into three stages: emergence of intelligence, autonomous action, and self-iteration [7]. Group 3: Open Source Strategy - Alibaba's initiative, Tongyi Qianwen, aims to create an open-source model, aspiring to build the "Android system of the AI era," emphasizing the value of open-source models over closed ones [9].