但协议中一个出人意料的转折是： 英伟达也将分阶段向OpenAI注入1000亿美元，购买其未上市股票。 这一金额已远超OpenAI过去十年所筹集的资金 总和。据Crunchbase数据显示，OpenAI成立十年来的融资总额为72亿美元。 分析认为， 这次两家公司不仅是商业合作，更是一场资本战略表演 。这项交易通过股权投资而非传统供应商融资，规避了英伟达的财务风险。对 OpenAI而言，这笔巨资支持了其通用人工智能的宏大叙事；对英伟达而言，此举加剧了AI领域的竞争，促使其他公司寻求类似合作，巩固其市场主导 地位。 与电信炒作潮相比，这次投资有什么不同？ 这项协议乍看之下类似于2000年代初电信热潮期间常见的供应商融资模式。当时Nortel、朗讯和摩托罗拉等公司向客户提供资金以维持收入快速增长， 但在泡沫破裂后承担了大量坏账。 双方都能从"AI竞赛正在进入更高层级"的印象中获益。对OpenAI而言，这笔巨资支持了其通用人工智能的宏大叙事；对英伟达而言，此举加剧了AI领域 的竞争，促使其他公司寻求类似合作，巩固其市场主导地位。 虽然人工智能热潮已经有些过火，但这次情况有点不同 。英伟达是以股权投资的方式参与OpenA ...

英伟达与OpenAI宣布的一项价值千亿美元的重磅合作，与其说是一笔常规的商业交易，不如说是一场 精心策划的资本表演。 根据当地时间周一公布的协议，英伟达将为OpenAI提供其构建、训练和托管AI模型（如新近发布的 GPT-5）所需的大规模数据中心急需的先进芯片。 如果这笔交易看起来像一场"表演"，那或许正是其意义所在。双方都能从"AI竞赛正在进入更高层 级"的印象中获益。 对OpenAI而言，获得千亿级别的资金承诺，强力支撑了其迈向超越人类智慧的通用人工智能的宏大叙 事。宣布建设更多超大规模数据中心的计划，将使其高达5000亿美元的估值对投资者更具说服力，并让 这个"愿景飞轮"持续转动。 而对于英伟达，人工智能领域将上演一场OpenAI及其同行之间的"生死对决"，这种观念对其商业利益 是极大的利好。谷歌、Meta以及私人控股的Anthropic等公司正全力冲向"超级智能"，它们深信，投资不 足者将被历史淘汰。如今，这些竞争者有更充分的理由寻求与英伟达达成类似的巨额交易。 这项协议乍看之下类似于2000年代初电信热潮期间常见的供应商融资模式。当时Nortel、朗讯和摩托罗 拉等公司向客户提供资金以维持收入快速增 ...

来源丨 机器之心 （ID： almosthuman2014 ） 编辑丨 杜伟 图源丨英伟达 一觉醒来，芯片巨头英伟达与 AI 巨头 OpenAI 官宣「联手」。 双方将建立战略合作伙伴关系，OpenAI 将部署高达 10 吉瓦（gigawatts）的英伟达系统。吉瓦是 一个功率单位，1 吉瓦等于 100 万千瓦。举例来说，一个普通家庭的峰值用电功率可能在 10 千瓦 左右。这意味着，1 吉瓦的电力大约可以同时为 10 万个家庭供电。 OpenAI 与英伟达预计将在未来几周内最终敲定这一战略合作新阶段的细节安排。 OpenAI 联合创始人兼总裁 Greg Brockman 强调：「OpenAI 自成立初期就与英伟达紧密合作。我 们基于英伟达平台创建的 AI 系统已服务数亿用户。此次部署 10 吉瓦算力将推动智能边界拓展，让 技术红利惠及全球。」 OpenAI 将把英伟达作为其 AI 工厂扩展计划的首选战略算力与网络合作伙伴。双方将共同优化 OpenAI 模型及基础设施软件、英伟达软硬件的技术路线图。此次合作将深化双方与微软、甲骨文、 软银及星际之门（Stargate）合作伙伴等生态网络正在推进的先进 AI 基础 ...

截至2025年9月23日 09:44，中证数字经济主题指数(931582)上涨0.24%，成分股均胜电子(600699)上涨6.90%，盛美上海(688082)上涨5.75%，德赛西威 (002920)上涨5.62%，拓荆科技(688072)上涨3.78%，科博达(603786)上涨3.50%。数字经济ETF(560800)上涨0.10%，最新价报1.05元。 据Wind数据显示，截至2025年8月29日，中证数字经济主题指数(931582)前十大权重股分别为东方财富(300059)、寒武纪(688256)、中芯国际(688981)、海光 信息(688041)、北方华创(002371)、汇川技术(300124)、澜起科技(688008)、中科曙光(603019)、科大讯飞(002230)、豪威集团(603501)，前十大权重股合计占 比53.36%。 | 股票代码 | 股票简称 | 涨跌幅 | 权重 | | --- | --- | --- | --- | | 300059 | 东方财富 | -1.38% | 10.51% | | 688981 | 中芯国际 | -0.57% | 6.34% | | 00237 ...

Core Viewpoint - The launch of the "Ge Zhi Visual Large Model Training and Deployment Integrated Intelligent Platform" aims to address current pain points in AI visual applications and facilitate the digital transformation of governments and enterprises into a more efficient and intelligent phase [1][3]. Industry Background - The traditional AI visual development model faces multiple challenges, including long iteration cycles, low utilization rates, and poor usability [3]. - The new platform leverages "large model" technology to cover multiple business scenarios with a single foundational multimodal large model, marking a leap from "handicraft" to "industrialization" [3]. Product Highlights - The platform is a soft and hard integrated product based on multimodal large model technology, featuring full-stack domestic production and significant advantages in low power consumption, high computing density, and security [5]. - It is designed for immediate use, with a 60-minute setup time, significantly reducing deployment cycles [6]. - The platform ensures data privacy and security through dual encryption and supports local private deployment [7]. Core Advantages - The platform provides a complete toolchain for automated closed-loop processes from data preparation to model training and inference deployment, supporting cold starts and few-shot learning [9]. - It offers unified management capabilities, reducing operational complexity through one-click tools and rapid fault identification [9]. - Users can flexibly deploy AI models as needed, minimizing compatibility issues and supporting various applications in government, finance, and the internet [9]. Technical Core - The Chinese visual large model is trained on a high-quality dataset of hundreds of millions of "image-text pairs," demonstrating strong adaptability to Chinese semantic understanding [11]. - The platform supports industrial-grade model production, enabling rapid model generation and high-precision iterations within minutes [11]. Application Cases - In smart cities, the platform has facilitated the unification of video image resource standards, enhancing inter-departmental collaboration and governance efficiency [13]. - In smart campuses, it has established a comprehensive security monitoring system, processing over 10 million data points and generating more than 10,000 alerts [15]. - In industrial production, it has implemented an AI fire detection system for early warning and rapid response, ensuring safety and resource optimization [16]. - In smart grids, the platform has improved the deployment cycle of new equipment and achieved intelligent closed-loop management [17]. Recognition - The platform was selected in September 2025 by the China Academy of Information and Communications Technology as a high-quality digital transformation product, highlighting its technological innovation and practical application capabilities in the AI field [17].

Core Insights - The main focus of the article is on the hardware requirements for large language models (LLMs) as discussed by Noam Shazeer at the Hot Chips 2025 conference, emphasizing the need for increased computational power, memory capacity, and network bandwidth to enhance AI performance [1][5][9]. Group 1: Hardware Requirements for LLMs - LLMs require more computational power, specifically measured in FLOPS, to improve performance and handle larger models [23]. - Increased memory capacity and bandwidth are crucial, as insufficient bandwidth can limit model flexibility and performance [24][26]. - Network bandwidth is often overlooked but is essential for efficient data transfer between chips during training and inference [27][28]. Group 2: Design Considerations - Low precision computing is beneficial for LLMs, allowing for more FLOPS without significantly impacting model performance [30][32]. - Determinism is vital for reproducibility in machine learning experiments, as inconsistent results can hinder debugging and development [35][39]. - Addressing issues of overflow and precision loss in low precision calculations is necessary to maintain stability in model training [40]. Group 3: Future of AI and Hardware - The evolution of AI will continue to progress even if hardware advancements stall, driven by software innovations [42]. - The potential for achieving Artificial General Intelligence (AGI) remains, contingent on the ability to leverage existing hardware effectively [42][44]. - The article highlights the importance of creating a supportive environment for individuals as AI transforms job landscapes, emphasizing the need for societal adaptation to technological changes [56].

Core Viewpoint - The A-share market showed mixed performance on September 22, with the Shanghai Composite Index rising by 0.12%, driven by gains in sectors such as computer hardware, semiconductors, and electronic components, while the restaurant, tourism, and shipping sectors faced declines [1] Group 1: Market Performance - The Chip ETF (159995) increased by 2.06% as of 10:30 AM, with notable gains from component stocks such as: - Rockchip up by 6.72% - GigaDevice up by 6.37% - Haiguang Information up by 4.48% - Wingtech Technology up by 4.06% - Lattice Semiconductor up by 4.01% [1] Group 2: Company Developments - Huawei officially launched its computing power super nodes and clusters, having introduced the Ascend 910 CAI chip in Q1 2023, with plans to release the Ascend 950PR in Q1 2026, the Ascend 950DT in Q4 2026, the Ascend 960 in Q4 2027, and the Ascend 970 in Q4 2028 [1] - Huajin Securities highlighted Huawei's release of two reports, "Intelligent World 2035" and "Global Digital Intelligence Index 2025," predicting that general artificial intelligence will be the most transformative technological driver in the next decade, with total computing power expected to grow by up to 100,000 times by 2035 [1] Group 3: Industry Insights - There is a positive outlook on the entire domestic computing power chip industry chain, encompassing design, manufacturing, packaging, testing, and upstream equipment materials, with a recommendation to focus on the full industry chain of domestic chips [1]

Group 1: Technological Breakthroughs - The emergence of native multimodal large models will replace piecemeal multimodal systems, achieving a 300% improvement in inference efficiency through deep integration of text, images, audio, and 3D data [1] - The acceleration of world models will establish a core technology foundation for embodied intelligence by 2025 [1] - The training paradigm will shift towards post-training scaling laws, optimizing reinforcement learning to reduce computational power consumption by 50% [4] Group 2: Industry Restructuring Trends - AI agents will provide hyper-personalized product customization, increasing customer satisfaction by 40% [6] - Real-time decision systems will enhance the speed of market response by three times in logistics and marketing [6] - The penetration of humanoid robots in industrial scenarios will achieve millimeter-level control precision, with smart factory coverage exceeding 80%, reducing manufacturing R&D cycles by 28.4% [6] Group 3: Social Integration Challenges - "Responsible AI" will become a mandatory standard, with non-compliant companies facing regulatory penalties and user attrition risks [8] - The automation rate of repetitive jobs will exceed 30%, while demand for creative and emotionally interactive roles will grow by 200% [8] - New mechanisms for privacy and copyright will emerge, with blockchain-enabled AI data rights technology addressing content ownership disputes [8] Group 4: Future Milestones - By 2027, general artificial intelligence (AGI) is expected to pass the Turing test in closed environments, and by 2030, neuromorphic chips will achieve a 1000-fold increase in energy efficiency [12] - By 2035, AI is projected to contribute over 40% to global GDP growth [12]

关于征集"中国机电一体化技术应用协会具身智能分会"发起单位的通知 中机电协〔2025〕48号 各有关单位： 为贯彻落实党的二十大提出的前瞻谋划未来产业，抢占科技革命制高点重要精神，本着充分发挥行业协会 的组织作用，整合资源协同解决具身智能在发展中遇到的挑战和问题，推动我国具身智能实现跨越式发展，为 经济高质量发展锻造新质生产力引擎。经研究决定，拟成立"中国机电一体化技术应用协会具身智能分会"。现 将有关事项通知如下： 一、具身智能分会成立背景 （一）具身智能战略价值与产业情况 2025年3月，中央政府工作报告明确提出培育具身智能未来产业。作为人工智能的前沿领域，具身智能融 合了人工智能、机器人技术、认知科学等多学科成果，使智能体能够在环境中完成感知、决策与行动的闭环， 实现高效人机协同，已成为新一轮科技革命和产业变革的关键驱动力。 当前，具身智能产业迅猛发展，仅2025上半年国内融资额突破230亿元。国际科技巨头也纷纷布局，具 身智能作为实现通用人工智能重要路径，在工业、物流、医疗、养老等领域潜力巨大。我国已涌现一批具身智 能领域创新企业，关键技术取得进展，场景应用持续拓展，多地出台政策持续加持。 （二）机电 ...

国泰海通证券 | 研究所 推 荐 阅 读 上线了！国泰海通2025研究框架培训视频版｜洞察价值，共创未来 杨 林 计算机首席分析师 行业核心洞察 AI开花、科技自立, 计算机大时代 价值主张 靠谱前瞻把握科技浪潮, 寻找计算机投资最锋利 的矛 量度 《EAI (具身智能):驱动通用人工智能与机器 点击下方图片 查看电话会回放详细议程 报告来源 观点来自国泰海通证券已发布的研究报告。 报告名称：AI开花、科技自立，行业景气度已开始修复 ——计算机行业2025年中期策略；报告日期：20250531；报告作者：杨林S0880525040027；风险 提示：技术发展不及预期；政策落地不及预期；业绩持续性不及预期；市场波动较大；经济恢复不及预 期。 重要提醒 本订阅号所载内容仅面向国泰海通证券研究服务签约客户。因本资料暂时无法设置访问限制，根据《证 券期货投资者适当性管理办法》的要求，若您并非国泰海通证券研究服务签约客户，为保证服务质量、 控制投资风险，还请取消关注，请勿订阅、接收或使用本订阅号中的任何信息。我们对由此给您造成的 扫码关注 星标不迷路 国泰海通证券研究所官方公众号 海量研报 | 热门活动 | 视听内容 ...