这次，多年来国产AI与英伟达深度绑定的故事，可能真的要被改写了。 这两天外媒一条阿里与百度已经在人工智能模型训练中引入自研芯片，以部分替代英伟达的产品的消息，平地起惊雷，让资本市场对国产算力抱团加速的 预期再度升温。 据了解，阿里自年初以来已将自研芯片应用于轻量级模型训练，性能已可与英伟达H20相媲美；百度则在尝试用昆仑芯P800训练新版文心大模型。消息刺 激下，百度股价大涨逾10%创下2024年10月以来新高，阿里涨超6%，腾讯、网易、京东等中概股亦跟涨。 中国市场或将迎来算力的集体狂欢。 而另一边，英伟达交出的Q2财报中，营收467亿美元，同比增长56%，但因为对中国市场给出零销售的口径及数据中心略低于预期，之后几个交易日市值 单日蒸发约1300亿美元，舆论口径转向"高处不胜寒"。 从寒武纪一度超越贵州茅台，到阿里、百度加快自研芯片落地，叠加政策层面"人工智能+"行动意见的持续释放，国产算力的"朋友圈"正在加速壮大。英 伟达的"失落"，不仅是单一企业的挑战，更是整个产业格局重塑的开端。 01 营收高增长，股价大跳水 在最新发布的2026财年Q2财报中，英伟达整体营收同比增长55%，继续保持全球AI浪潮中的 ...

前十大沽空偏离值排行 | 股票名称 | 沽空金额 | 沽空比率 | 偏离值↓ | | --- | --- | --- | --- | | 李宁-R(82331) | 16.66 万元 | 100.00% | 35.57% | | 友邦保险-R(81299) | 36.23 万元 | 92.90% | 28.24% | | 中国海洋石油-R(80883) | 732.27 万元 | 71.40% | 27.47% | | 比亚迪股份-R(81211) | 207.24 万元 | 91.45% | 25.22% | | 新鸿基地产-R(80016) | 8.80 万元 | 100.00% | 24.06% | | 康哲药业(00867) | 6664.22 万元 | 39.46% | 21.96% | | 耀才证券金融(01428) | 9008.46 万元 | 28.46% | 21.53% | | 中国太保(02601) | 4.38 亿元 | 38.85% | 20.52% | | 永达汽车(03669) | 204.05 万元 | 33.90% | 18.94% | | 福寿园(01448) | 932. ...

9月13日，百度旗下、国内知名商业信息查询平台爱企查在广州举办首届"红运之夜 2025品牌红榜颁奖 盛典"。据悉，自8月25日正式官宣以来，这场活动带动了52万观众参与互动投票，300多个品牌联动冲 榜，榜单票数累计突破117万，2025品牌红榜由此诞生，覆盖经典传承榜、掌上智能榜、自在生活榜、 舌尖心动榜、宝藏零食榜五大细分榜单，五大核心消费场景的44家品牌登榜。 当下，企业的信用与品牌口碑正日益成为链接商业价值与消费信任的纽带，爱企查通过专业数据与全民 票选的双重护航，帮助广大消费者锁定"心水品牌红榜"。活动现场，百度To B垂类业务总经理甘露 以"荣耀揭榜官"身份登台正式揭榜。 对消费者而言，这一"榜单指引+工具查询"的组合拳，不仅打破了企业与消费者之间的信息不对称，更 显著降低了消费决策成本，真正将"维护消费权益"落到实处。于品牌方而言，"心水品牌"的荣誉既代表 了品牌的长期努力被大众看见，更能激励品牌推动整个消费市场逐步形成"以诚信立品牌、以品质赢市 场"的正向循环。 作为中国知名的企业大数据平台，爱企查依托百度先进的AI和大数据技术，为网民提供真实快速的企 业信息免费查询服务。此次"红运之夜"的举 ...

henry 发自 凹非寺 量子位 | 公众号 QbitAI 让人熬到头秃的毕业论文有救了！ 刚刚，在第12届AI Day开放日上， 百度学术 官宣全面"AI重构"—— 它将从我们熟悉的查文献、找引用格式的资料库，升级为覆盖论文全生命周期的"搜、读、创、编"的Research平台。 据悉，百度学术还将上线AI学术搜索、AI文献总结、AI阅读、论文图谱等数十项新能力，打造行业 首个一站式AI学术平台 。 怎么个一站式？一起瞅瞅。 搜、读、创、编全链路覆盖 一般来说，无论是科研小白还是上顶刊如喝水的大佬，科研流程都不外乎以下几个环节： 而最难的莫过于前期准备阶段，你得广撒网去搜文献，还得不停地捯饬研究问题、找research gap，一边翻一边琢磨。 更离谱的是——读着读着，好多人就干脆换了个研究问题……之前的搜索全白费！ | | | 不仅如此，为了确保文献检索密不透风，滴水不漏。 前期准备阶段：论文开题、资料搜集、阅读文献。 具体写作阶段：撰写大纲、厘定结构、写摘要、导论（文献综述、研究方法、问题）、正文（论证与实验） 修改与完善阶段：对论文的格式、语言、表达进行润色和优化。 所以啊，搜索效率，真的能决定你是顺利 ...

Core Insights - The WAVE Summit 2025 marks a significant advancement in China's science popularization efforts, focusing on deep learning and innovative science communication [1] - The launch of the "Science Museum Intelligent Matrix" and the "Future Wings Science Popularization Co-Creation Plan" signifies a shift towards intelligent and collaborative science services in China [1] Group 1: Science Museum Intelligent Matrix - The "Science Museum Intelligent Matrix" is a collaborative initiative between the China Science and Technology Museum and Baidu, integrating resources from 27 provincial science museums to create an AI-driven science service platform [4] - The matrix features multi-modal interactions, including image recognition, voice Q&A, and AI-generated travel logs, enhancing visitor engagement and transforming science communication from one-way to interactive [4] - The platform aims to make science resources more accessible and collaborative, involving 28 science museums, four enterprises, and 340 independent developers [4] Group 2: Future Wings Science Popularization Co-Creation Plan - The "Future Wings Science Popularization Co-Creation Plan" focuses on open and inclusive science ecosystems, leveraging Baidu's Wenxin large model to provide youth with easier access to knowledge [5] - The plan encourages science venues to open data resources and application scenarios, supporting developers with AI tools and technical assistance to foster innovative science communication [5] - Recent collaborations have resulted in various AI-driven projects, including the "AI Travel Log" and "Science Open Door," with 283 science intelligent entities developed across multiple fields [5] Group 3: Technological Innovation and Accessibility - The initiatives demonstrate the feasibility of technology-driven inclusivity and the potential for innovative science communication, moving from knowledge transmission to cognitive inspiration [6] - The integration of AI is driving the evolution of science services towards more intelligent, open, and creative approaches [6]

Core Viewpoint - The article emphasizes a transformative shift in academic research facilitated by AI, moving from traditional methods to an integrated, AI-driven approach that significantly enhances efficiency and user experience [2][26]. Group 1: AI-Driven Research Transformation - The introduction of AI in academic research allows for rapid processing of large volumes of literature, reducing the time needed to synthesize information from weeks to mere seconds [3][4]. - The new AI capabilities enable users to interactively engage with research papers, asking specific questions about content, which transforms the reading experience into a dynamic dialogue [6][14]. - The integration of various research functions into a single platform eliminates the need for multiple software tools, creating a seamless workflow for researchers [9][19]. Group 2: Features of the New Platform - The upgraded platform offers a "knowledge map" feature that visually represents the connections between research topics, classic literature, and emerging trends [7][10]. - AI assists in generating research outlines and provides suggestions for innovative research directions, streamlining the initial stages of academic writing [15][16]. - The platform includes built-in editing tools that automate formatting and citation processes, allowing researchers to focus on content creation rather than technical details [16][17]. Group 3: Strategic Vision and Market Position - The article outlines a broader strategic vision for the platform, positioning it as a "super app" for knowledge services, with a foundation built on extensive and high-quality data resources [20][21]. - The platform aims to create an academic identity system, enhancing the visibility and credibility of researchers, akin to an academic credit system [21][22]. - By leveraging its vast user base and data resources, the platform seeks to become a central hub for academic content distribution and collaboration [22][24]. Group 4: Future Implications - The AI upgrade signifies the beginning of a new era in academic research, where scholars can focus on innovative thinking and experimentation rather than data management [26][24]. - The platform's evolution reflects a shift towards a more integrated and efficient academic ecosystem, potentially reshaping how research is conducted and disseminated globally [26][24].

Core Insights - The collaboration between Baidu Smart Cloud and Dou Shen Education has led to the launch of an innovative educational hardware product, the Learning Companion Robot, which aims to provide personalized learning solutions for families through advanced AI capabilities [2][3] - The product signifies a shift in educational models from a "teacher-student" dynamic to a "human-machine-student" intelligent ecosystem [2] AI Capabilities - The Learning Companion Robot integrates three core AI technologies: Baidu's Wenxin 4.5 model, the Image Generation model, and the Object Recognition model, enabling human-like interaction and personalized tutoring [2] - The robot can analyze homework problems in a layered manner and engage students through gamified tasks, enhancing their learning experience [2] Visual Interaction - The breakthrough in visual interaction allows the robot to convert abstract concepts into concrete images, significantly improving knowledge absorption efficiency by over 40%, particularly benefiting younger children [2][3] Knowledge Expansion - The Object Recognition feature acts as a mobile "encyclopedia," achieving a recognition accuracy of 98.7% across 12 subject areas, creating a complete learning loop from observation to inquiry and explanation [3] Technical Innovations - The Baidu Baichuan AI computing platform enhances model training efficiency by 65% while reducing energy costs by 32%, ensuring smooth operation of complex AI functions on consumer-grade hardware [3] - The platform's distributed inference architecture supports high concurrency, maintaining millisecond-level response times for multiple users [3] Educational Ecosystem - Baidu Smart Cloud's "three-tiered progressive" educational solution is reshaping the industry ecosystem, providing optimized computing services, low-code development support, and comprehensive applications across teaching, assessment, and marketing [3] Emotional and Lifestyle Integration - The Learning Companion Robot transitions from a "tool" to a "growth partner," incorporating emotional computing to recognize user emotions and provide encouragement during learning challenges [4] - It also features a lifestyle management function, integrating services like schedule management and health reminders, thus redefining the emotional value of smart devices [5] Future Potential - Continuous upgrades to the Qianfan Model Builder platform will expand the robot's capabilities, including automatic generation of subject mind maps and video understanding for experimental operations [5] - The collaboration marks a deep integration phase in AI education, creating a scalable, customizable, and perceptible intelligent education ecosystem [5]

Core Viewpoint - The "AI Elderly Assistance Public Welfare Program" has been officially launched to help the elderly bridge the digital divide through AI applications and volunteer activities [1] Group 1: Program Overview - The program includes the "AI Elderly Assistance Intelligent Agent," which integrates knowledge dissemination, health maintenance, and emotional companionship, addressing over 1,000 common questions relevant to individuals aged 60 and above [1] - The intelligent agent serves as an authoritative information retrieval assistant, with health content sourced from the "100 Questions on Dietary Nutrition Guidance for the Elderly" manual by the China Aging Development Foundation [1] - It functions as an effective fraud prevention assistant, utilizing multimodal capabilities to analyze images and identify scam posters and spam messages [1] Group 2: Implementation and Outreach - On September 12, volunteers from Baidu's "Xiao Hao Shi" initiated the first AI public welfare lecture at the retirement office of government agencies [1] - The program aims to reach multiple provinces, elderly universities, communities, and nursing homes, promoting the "AI Elderly Assistance Intelligent Agent" and conducting AI volunteer activities to cover hundreds of thousands of elderly individuals [1] - The initiative seeks to provide replicable practical samples for achieving a positive aging perspective and healthy aging [1]

Core Insights - The article emphasizes the importance of SEM bidding operation services, particularly for Baidu's information flow, as a key driver for businesses to overcome growth bottlenecks [1][2] - It highlights the advantages of information flow advertising over traditional search ads, including proactive demand stimulation, algorithm-driven precision targeting, and seamless integration with content [2][3] - The article outlines four core capabilities of professional operation companies that enhance advertising effectiveness [5][6][7] Group 1: Advantages of Information Flow Advertising - Information flow advertising has three unique advantages: proactive push mechanism, algorithm-driven intelligent distribution, and native ad format that reduces user resistance [2] - Data from Baidu Marketing Research Institute shows that the average conversion rate of information flow ads is 40% higher than traditional display ads, particularly in e-commerce, education, and local life sectors [2] Group 2: Core Capabilities of Professional Operation Companies - The first capability is audience profiling, where top teams use Baidu's tools to create a three-dimensional model of user attributes, behaviors, and preferences [5] - The second capability involves dynamic bidding strategies, where skilled optimizers implement real-time pricing adjustments and keyword strategies to enhance conversion rates [5] - The third capability is a standardized creative production process, which includes attention-grabbing titles and optimized landing pages, leading to significantly higher click-through rates [6] - The fourth capability is a data monitoring system that tracks various performance metrics, enabling businesses to optimize their advertising strategies effectively [7] Group 3: Common Pitfalls and Future Trends - Common pitfalls for businesses include over-reliance on automation, focusing solely on low costs at the expense of quality, and misapplying search ad strategies to information flow ads [9] - The future trend indicates a dual evolution towards increased intelligence and vertical specialization in advertising operations, requiring a blend of psychological insight, mathematical modeling, creative strategy, and algorithm optimization [10]

CLNKraneSharesWeek in Review Asian equities had a strong week, while Mainland China underperformed. Oracle’s Open AI announcement and the stock’s very strong performance positively affected Mainland China’s tech sector, STAR Board, growth stocks, AI, cloud supply chain, and related ecosystem stocks. NBC News reported that U.S. lawmakers will visit China this month, marking the first visit by a U.S. House of Representatives delegation since 2019, according to Reuters. The State Administration for Market Regu ...