千问的崛起和广泛应用，证明了在软件和算法层面，中国已经具备了与硅谷分庭抗礼，甚至在开源生态 上略胜一筹的实力。 撰文丨沸雪 谁也没想到，美股科技七巨头之首的Meta创始人扎克伯格，居然有一天也成为了中国AI模型支持者。 12月10日，彭博社报道称，曾经的全球开源霸主Meta新模型"牛油果"（Avocado）项目，选择蒸馏中国 阿里千问的开源模型。 根据报道，扎克伯格密切关注新组建的TBD实验室团队，他们的"牛油果"模型训练，蒸馏了多方开源模 型，除了谷歌的Gemma、OpenAI的gpt-oss之外，这一次还出乎大家预料地选择了中国科技巨头阿里巴 巴旗下的通义千问。 这也意味着，扎克伯格面对日益强大的中国开源模型，出现了180度的态度转变，此前，扎克伯格多次 呼吁要支持美国模型，然而随着Meta今年Llama4的失败和中国模型的强势崛起，扎克伯格也转投阿里 千问。 那么问题来了，为什么一度被视为"美国优先"的硅谷开源强硬派的小扎，如今也开始选择中国AI大厂作 为自己的模型底座？ 开源开放、全栈AI 开源，应该可以算是这场自2023年打响的AI战争中的最大变量。 曾几何时，Meta凭借Llama系列模型，几乎以 ...

Core Viewpoint - The partnership between Yanfeng International and Alibaba Cloud aims to leverage AI capabilities and automotive industry expertise to enhance product innovation and manufacturing efficiency, establishing Yanfeng as a benchmark in intelligent automotive component manufacturing [1][6]. Group 1: Partnership Overview - Yanfeng International has signed a comprehensive AI cooperation agreement with Alibaba Cloud to deepen global strategic collaboration [1]. - The agreement will focus on optimizing global manufacturing systems and accelerating innovation in smart cockpit products [1][6]. Group 2: Product Innovation - Utilizing Alibaba's Tongyi Qianwen model, the partnership will explore new AI solutions for next-generation smart cockpit products [3]. Group 3: Manufacturing Enhancements - Yanfeng International will implement Alibaba Cloud's full-stack cloud technology to create an efficient and agile digital production system, enhancing quality and efficiency across global factories [4]. Group 4: Management and Operations - The collaboration will also aim to improve digital service capabilities, optimize organizational efficiency, and establish a scalable digital organizational system [5]. - Since the start of their cooperation in 2024, Yanfeng has migrated its overseas information systems to Alibaba Cloud, creating a robust digital infrastructure for global operations [5].

Core Insights - The article highlights a significant shift in the global open-source AI landscape, with Meta's latest AI model "Avocado" utilizing Alibaba's Qwen open-source model for distillation training, marking a pragmatic turn for Meta amid the rise of Eastern models [2][5][10]. Group 1: Market Dynamics - Alibaba's stock price rose over 4% following the news of Meta's use of its Qwen model, indicating market confidence in Alibaba's technology [2]. - The number of Qwen derivative models surpassed 180,000, and its global download count exceeded 700 million by October 2025, both metrics surpassing Meta's Llama model [5][12]. - Alibaba has open-sourced over 300 models since 2023, covering a wide range of parameters and supporting 119 languages and dialects, enhancing its global developer appeal [5][12]. Group 2: Industry Trends - The trend of Silicon Valley integrating Chinese open-source AI technologies is evident, with major companies like Nvidia and Microsoft adopting Qwen-based solutions for complex multi-language and multi-modal data scenarios [6][12]. - Airbnb's CEO noted a 30-40% cost reduction when using Qwen models compared to other solutions, highlighting the competitive advantage of Chinese models [6][12]. - Academic institutions and innovative labs in Silicon Valley are increasingly adopting Qwen-based technologies, further validating its effectiveness [7][12]. Group 3: Technological Advancements - Alibaba's "full-stack AI" strategy is recognized for its comprehensive integration of technology, from chips to applications, supporting rapid model iteration and commercialization [10][14]. - The Qwen model's optimization for multi-language support provides a significant advantage in non-English markets, facilitating its global expansion [12][14]. - The Qwen application achieved over 30 million monthly active users within 23 days of its public launch, showcasing its rapid adoption and alignment with market needs [13][12]. Group 4: Future Outlook - The article suggests a structural reshaping of the global AI open-source ecosystem, with Meta's shift reflecting deeper changes in technological capabilities and competitive dynamics [12][14]. - The collaboration across cultural and geographical boundaries in AI innovation is expected to accelerate, with the best technological solutions emerging from the most open ecosystems [14].

软件ETF（515230）跟踪的是软件指数（H30202），该指数从市场中选取涉及软件开发、销售和服务等 业务的上市公司证券作为指数样本，覆盖操作系统、应用软件、网络安全等领域的代表性企业，以反映 软件行业相关上市公司证券的整体表现。该指数具有显著的成长性和技术导向性，能够较好地体现软件 行业的市场趋势和发展动态。 长江证券指出，豆包手机助手作为首个豆包与手机厂商在操作系统层面合作的手机AI助手，实现了通 过系统语音、侧边AI键及智能耳机唤醒等功能，直接读取屏幕内容并跨应用调用服务。此举推动手机 操作系统交互逻辑从"图标点击"向"意图识别"转变。终端设备可能是AI落地的重要方向，而全栈AI或将 是未来模型厂商的发展趋势。可关注国产大模型厂商、AI基础设施提供商、AIAgent相关公司以及中国 推理算力产业链。 （文章来源：每日经济新闻） ...

丨证券研究报告丨 报告要点 [Table_Summary] 12 月 1 日，豆包手机助手以技术预览版正式亮相，搭载该功能的努比亚（nubia）M153 工程 样机同步开售。当前时点，以字节为代表的国产大模型企业，正持续布局手机、耳机等端侧流 量入口，希望通过将大模型能力、超级 App 与硬件终端结合，推动 AI 时代的交互模式变革。 在此背景下，我们认为端侧 AI 或将成为 AI 在日常使用中重要的落地场景，推动 AI Agents 能 力持续提升，进而加速国产 AI 的商业化。建议关注：1）国产大模型厂商；2）AI Infra：AI Infra 有望核心受益；3）AI Agent 相关厂商；4）中国推理算力产业链。 分析师及联系人 [Table_Author] 宗建树 陈耀文 SAC：S0490520030004 SAC：S0490525070002 SFC：BUX668 请阅读最后评级说明和重要声明 %% %% %% %% research.95579.com 1 行业研究丨点评报告丨软件与服务 [Table_Title] 豆包手机助手发布，关注端侧 AI 产业进展 软件与服务 cjzqdt11111 ...

Core Viewpoint - The Greater Bay Area Science Forum highlights the recognition of Lenovo's Vice President Dai Wei as a leading figure in technology transfer, emphasizing the company's significant achievements in the IT service market and its strategic shift towards AI-driven services [1][4][12]. Group 1: Recognition and Achievements - The Greater Bay Area Science Forum awarded Dai Wei the title of "Outstanding Business Leader in Technology Empowerment Innovation 2025" for his contributions to transforming cutting-edge research into industrial productivity [1]. - Lenovo's solution services have topped the Chinese IT service market, with annual revenue and compound growth rate leading the industry, as reported by IDC [1][4]. - Under Dai Wei's leadership, Lenovo's solution services have achieved significant milestones, including winning the IDC Asia-Pacific Smart City Award twice, making Lenovo the only company in mainland China to achieve this [4]. Group 2: Strategic Transformation - Dai Wei has transitioned from a pioneer to a "definer" of AI services, leading Lenovo's strategic shift towards intelligent service solutions since 2017 [4]. - The company has adopted a "hybrid AI" strategy, with a focus on deeper technological innovation and ecosystem building to thrive in the AI era [2][5]. - In November 2025, Dai Wei introduced a new strategic framework called "One Engine, Four Fleets," which aims to redefine AI-driven IT service paradigms in China [6]. Group 3: Market Performance - Lenovo's solution services reported a 15% year-on-year revenue growth to 146.4 billion RMB in the second quarter of the 2025/26 fiscal year, exceeding market expectations [7]. - The SSG solutions and services business group has maintained double-digit growth for 18 consecutive quarters, with an operating profit margin exceeding 22% [7]. - The proportion of high-value project and solution services in SSG's overall revenue has increased to approximately 60%, reflecting a positive trend in business structure [7]. Group 4: AI and Industry Impact - Dai Wei's vision includes making AI services accessible to a wide range of industries, emphasizing the importance of AI in the fourth industrial revolution [5][11]. - Lenovo has become a long-term official IT service partner for the Olympics, showcasing its technological capabilities in major global events [10][13]. - The concept of "City Super Intelligent Agents" has been implemented across multiple cities, marking a significant advancement in smart city initiatives and enhancing urban economic productivity [10][14].

Core Viewpoint - The article highlights the significant achievements of Lenovo's Vice President Dai Wei in transforming the company's service sector, particularly in the context of AI-driven innovation and market leadership in the IT service industry in China [1][4][10]. Group 1: Achievements and Recognition - The Greater Bay Area Science Forum recognized Dai Wei as the "Outstanding Business Leader" for his contributions to technology transfer and innovation in the industry [1]. - Under Dai Wei's leadership, Lenovo's solution services have topped the Chinese IT service market, achieving the highest annual revenue and compound growth rate according to IDC's latest report [1][6]. - Lenovo's service transformation has been marked by significant projects like the Beijing Smart Summer Palace and Hainan Wenchang Smart City, which won the IDC Asia-Pacific Smart City Award twice [4]. Group 2: Strategic Shifts and Innovations - Dai Wei has transitioned from a pioneer to a "definer" of AI services, leading Lenovo's strategic shift towards a "hybrid AI" model and the "All-Stack AI" strategy in China [2][4]. - The introduction of the "One Engine, Four Fleets" strategy aims to redefine AI-driven IT service paradigms, integrating various AI solutions and services [6]. - Innovative delivery models such as "Super Factory Fast Customization" and "AI as a Service" (AaaS) have been developed to lower the barriers for enterprises to adopt AI technologies [6][10]. Group 3: Market Performance and Growth - Lenovo's solution services group (SSG) has seen a 15% year-on-year revenue growth, reaching 146.4 billion RMB, with the SSG solutions and services business growing by 18% [6][7]. - The high-value project and solution services now account for approximately 60% of SSG's overall revenue, indicating a continuous upward trend [7]. - The successful implementation of AI solutions across various sectors, including smart manufacturing and healthcare, demonstrates the market's validation of Lenovo's strategic direction [11].

Group 1: Company Overview - Alibaba Group is a leading global e-commerce internet technology group that restructured its organization in June, merging Ele.me and Fliggy into its e-commerce business unit, and in August, it adjusted its business structure into four main segments: China E-commerce Group, International Digital Commerce Group, Cloud Intelligence Group, and Others, focusing on "E-commerce, Cloud + AI" as its core business, enhancing organizational concentration and cohesion [1] Group 2: Investment Logic - Instant retail is crucial for maintaining e-commerce traffic advantages, with Alibaba entering the market in May through food delivery services. User activity metrics show a year-on-year increase in Daily Active Users (DAU) from May to October, with growth rates of +3%, +7%, +17%, +19%, +12%, and +9% respectively [1] - Order volume peaked at 120 million in July, with market share for Taobao Flash Purchase and Ele.me rising significantly from 29.4% in Q4 2024 to 42.8% in Q3 2025 [1] - Delivery capability has seen over 240% growth in the number of delivery personnel from April to August 2025 [1] - User Experience (UE) investments are at their highest in FY26Q2, with losses halved from peak levels, and improvements expected as scale and efficiency increase [1] Group 3: AI Strategy - Alibaba Cloud is recognized as a "full-stack AI company," pursuing top-tier self-research capabilities in AI chips, cloud computing platforms, and foundational large model capabilities. It operates in 29 public cloud regions and 92 availability zones globally, with over 3,200 edge nodes [2] - Annual capital expenditures exceed 100 billion yuan, with the Qwen series large models significantly upgraded to Qwen3-Max, enhancing performance and creating a closed-loop with applications like Qwen APP, Quark, Taobao, and DingTalk [2] - The latest PPU chip technology from Alibaba's T-head is at the forefront domestically [2] Group 4: Profit Forecast and Valuation - The traditional shelf e-commerce advantage is expected to stabilize market share, while cloud business driven by AI shows significant future growth potential. Adjusted net profits for FY2026-2028 are projected at 107.9 billion, 149.4 billion, and 175.8 billion yuan, with corresponding adjusted P/E ratios of 24.8, 17.9, and 15.2 times [2] - The e-commerce business (excluding instant retail) is assigned an 11 times earnings valuation (EBITDA basis), while the cloud business is given a 7 times price-to-sales valuation, with other businesses currently not contributing to valuation. The target market capitalization is set at 3.36 trillion yuan, with a target price of 192.48 HKD, initiating coverage with a "Buy" rating [2]

Group 1 - The core viewpoint of the report is that Alibaba's traditional e-commerce has a strong advantage and is expected to maintain market share stability, while its cloud business has significant growth potential driven by AI [1] - The adjusted net profit estimates for Alibaba for FY2026-2028 are projected to be 107.9 billion, 149.4 billion, and 175.8 billion RMB, with corresponding adjusted PE ratios of 24.8, 17.9, and 15.2 times [1] - The target market capitalization is set at 3.36 trillion RMB, with a target price of 192.48 HKD, and the report initiates coverage with a "Buy" rating [1] Group 2 - Alibaba's entry into instant retail through food delivery has led to a recovery in user activity, with DAU increasing by 3% to 19% from May to October [2] - The order volume peaked at 120 million in July, with market share in the mainstream food delivery sector rising from 29.4% in Q4 2024 to 42.8% in Q3 2025 [2] - The number of delivery personnel increased by over 240% from April to August 2025, indicating improved fulfillment capabilities [2] Group 3 - Alibaba Cloud is recognized as a "full-stack AI company," pursuing top-tier self-research capabilities in AI chips, cloud computing platforms, and foundational large model capabilities [3] - The cloud service operates in 29 public cloud regions and 92 availability zones globally, with over 3,200 edge nodes [3] - Alibaba's annual capital expenditure exceeds 100 billion RMB, with significant advancements in its Qwen series large models and leading technology in its PPU chip [3]

Core Insights - Amazon Web Services (AWS) is focusing on making computing power truly usable and enabling Agents to operate effectively, rather than chasing short-term profits from selling computing power [2][38] - AWS maintains a leading position in the global cloud market with a market share of 37.5%, significantly ahead of its closest competitor [2][39] - The annual recurring revenue (ARR) for AWS is projected to reach $132 billion by December 2025, reflecting a 20% year-over-year growth [2][39] Competitive Landscape - AWS faces intense competition from Microsoft Azure, Google Cloud GCP, Oracle OCI, and CoreWeave, which are securing long-term contracts with major clients through investments and computing power collaborations [3][39] - The concept of "computing power financialization" is creating short-term pressure on AWS's stock and public perception [3][39] Technological Trends - The integration of full-stack AI, including chips and models, is becoming increasingly important for attracting enterprise clients [3][40] - The rise of Agentic AI is identified as a new battleground, with billions of Agents expected to emerge in the future [3][40] AWS's Strategic Response - At the re:Invent 2025 conference, AWS announced new products aimed at helping enterprise clients quickly implement Agents [4][40] - CEO Matt Garman emphasized that valuable Agents require four core components: AI infrastructure, AI inference platforms, data, and Agent development tools [4][40] Cost Efficiency Initiatives - AWS is developing its own AI chips to reduce the total cost of ownership (TCO) for computing infrastructure [8][44] - The newly launched Trainium 3 chip, built on a 3nm process, can produce five times more Tokens per megawatt compared to its predecessor and reduce training costs by up to 50% [9][45] Product Development - AWS has deployed over 1 million Trainium chips, which are expected to generate billions in revenue annually [11][47] - The Amazon Nova 2 series of self-developed models aims to provide cost-effective solutions for enterprises, with a focus on low-cost processing of simpler tasks [12][51] Market Positioning - Amazon Bedrock, AWS's model platform, integrates models from various vendors, allowing enterprises to utilize multiple models efficiently [16][52] - The company is positioning Amazon Bedrock as a significant growth driver, with expectations of it matching the revenue contribution of EC2 in the long term [19][55] Agent Development Tools - AWS launched Amazon Bedrock AgentCore, a standardized toolset for developing and deploying Agents, which has seen over 200,000 SDK downloads shortly after its release [20][56] - The company is also introducing official Agent tools, such as Security Agent and DevOps Agent, to enhance internal operations and customer offerings [23][59] Long-term Vision - AWS is focused on solving current customer pain points rather than pursuing speculative short-term gains, reflecting a pragmatic approach to technology development [32][34] - The company aims to build a comprehensive Agent infrastructure that can drive exponential growth in computing power consumption through user interactions with Agents [26][29]