首届智元机器人合作伙伴大会：8 月21 日，智元机器人在上海以「与智同行 共启新元」为主题举办首 届智元机器人合作伙伴大会，展示公司在产品、技术、商业、生态、资本的全链条布局，加速具身智能 商业化进程，推动行业从「技术探索」迈向「规模化商业应用」。 2025 年合作伙伴政策将分层次向优质伙伴直接输送高端资源。 公司已与上市公司合资创立机器人企业，并投资覆盖研发、制造、组装、代工的上中下游企业。未来目 标三年部署10 万台+通用机器人，支持100+自主任务，构建开放生态。智元将持续全球合作，使具身智 能机器人成为新质生产力。 构建生态系统：在策略发布中，智元机器人董事长兼CEO 邓泰华强调2025 年是身智能的转折点，将成 为继手机、汽车后的「未来智能终端」。公司策略聚焦于量产通用人形机器人、构建强协同的全栈软硬 件平台、拓展工业/商业/家庭场景。关键的生态共建会通过三方面实现：开源中间件（AimRT）、数据 集及「灵渠OS」系统；推动产业协同的「一体化战略」；启动「智元计划A」三年孵化50+早期项目， 构建千亿规模生态。 「1+3」战略与产品发布：CTO 彭志辉提出「1+3」全栈技术战略： 一个机器人本体， ...

衡宇 发自 上海 量子位 | 公众号 QbitAI 智元机器人实现对A股科创板公司上纬新材63.62%的控股权后，终于来了次公开、正式的亮相。 今年3月宣布"智元机器人董事长兼CEO"身份的 邓泰华，也在这场智元机器人首届合作伙伴大会上首次公开亮相。 倒也没讲虚的，一股脑吐出许多外界一直想知道又不知道的东西： 慷慨而谈的不只有邓泰华。 会前交流时， 智元机器人通用业务部总裁王闯 、 智元机器人合伙人&具身业务部总裁姚卯青 也出面回答了许多疑问。 我们整理其中精华，分成"吃瓜""技术""合作"三个板块，与大家分享。 融资这事儿智元想融就能融，年底将启动C轮，引入更多国际产业方； 无营收前提下，现金流能撑三年； 今年计划出货几千台，明年出货数万台，未来每年出货几十万台； 商业落地节奏，先ToB再ToC； 落地场景选择路径，简单环境+简单任务→简单环境+复杂任务→复杂环境； 当前主要在打基础 （能力、产品、团队） ，今年起产品成熟度提高，市场与团队具备基础，开始"逐步踩油门"； 智元团队超1000人，平均年龄31岁； 团队成员75%是研发，其中2/3投入AI，1/3投入本体。 未来三年，投入数十亿资金，孵化50个早 ...

以全链条布局加速具身智能商用落地。 本文为IPO早知道原创 作者｜ Stone Jin 微信公众号｜ipozaozhidao 据 IPO早知道消息， 智元机器人首届合作伙伴大会 于 8月21日 在上海举办。 智元机器人董事长兼 CEO邓泰华 在开场演讲中 提出，全球正处于 "具身智能大爆发的前夜"，人工 智能正加速朝着AGI（Artificial General Intelligence，通用人工智能）迈进，2025年具身智能机 器人将迎来商用发展的拐点，最终成为继手机和汽车之后的"下一代海量智能终端"。 邓泰华 进一步指出， 智元机器人以通用量产人形机器人为牵引目标，着力打造强智能、易协作的软 硬件全栈平台，梯次推进工业、商业和家居商用场景的落地，最终构建通用智能机器平台的应用生 态。 目前， 围绕 "一体三智"持续构建核心产品竞争力，智元机器人实现了业界唯一的全系列全场景产品 布局，并构建了本体-小脑-大脑软硬全栈技术布局。 在商业运营方面，通过不断强化产品能力，梯 次推进适用于不同场景的产品落地，逐步实现跨领域规模商用，在保持强劲业务发展势头的同时，持 续突破高质量大客户群。 此外， 邓泰华着重强调 了 ...

Core Insights - The development of embodied intelligence is likened to a fast-moving train, emphasizing the rapid pace of technological advancement in robotics [1] - The company has identified seven key scenarios for commercialization, including exhibition guidance, entertainment performances, logistics sorting, and data collection training [1][2] - A recent outdoor live demonstration of the full-size robot Expedition A2 showcased its capabilities in extreme conditions, completing a 24-hour task in high temperatures [1][2] Technological Breakthroughs - The robot achieved four major technological breakthroughs: autonomous movement and navigation in complex lighting, adaptability to extreme environments, reliability in continuous operation, and ease of task deployment [2] - The robot successfully operated in outdoor conditions of 37°C and a ground temperature of 61°C, indicating its potential for applications in harsh environments [2] Collaboration and Market Strategy - The company aims to explore practical applications of embodied intelligence through partnerships with various listed companies, focusing on partners with strong industry backgrounds and technical service advantages [2] - The primary opportunities for deployment this year are in exhibition guidance and entertainment performances, targeting partners with existing industry foundations [2] - The company is also planning international expansion, particularly in the fields of scientific education and commercial markets, with initial progress noted in Malaysia [2]

王闯将具身智能技术比作一辆高速驶来的火车，一旦错过便再也追不上。他认为，继远征A2完成24小时 不间断高温行走直播后，智元机器人在自主移动、极端环境适应性、高可靠性和任务快速部署能力上已取 得关键性突破，标志着商业化的起点。 机器人全身触觉、关节性能、安全性等方面，硬件远未达到理想状态，必须与软件协同进化。 硬·AI 作者 | 申思琦 编辑 | 硬 AI 8月21日，在智元机器人首届合作伙伴大会上，具身智能的商业化落地路径被清晰地勾勒出来。 智元机器人通用业务部总裁王闯，与来自软通动力、灵西科技、大丰硅基方舟、富临精工的核心高管—— 软通动力集团副总裁李永辉、灵西科技CEO丁丁、大丰硅基方舟CTO张将红、富临精工工程中心主任邓 扬，共同接受了媒体采访，深入探讨了人形机器人的技术突破、场景落地、生态合作与未来愿景。 对于行业热议的软硬件瓶颈问题，王闯明确表示"不同意硬件不是瓶颈"，他认为在全身触觉、关节性能、 安全性等方面，硬件远未达到理想状态，必须与软件协同进化。 智元今年的策略十分明确：务实地聚焦于展厅讲解接待、文娱表演、料箱转运、科研教育等七大有信心落 地的场景，并与合作伙伴携手，共同将产品推向市场。在出 ...

Core Insights - The commercialization path for embodied intelligence has been clearly outlined at the inaugural partner conference of Zhiyuan Robotics, emphasizing the importance of collaboration with various industry leaders [1][2] - Zhiyuan Robotics has achieved significant breakthroughs in autonomous mobility, extreme environment adaptability, reliability, and rapid task deployment, marking the beginning of its commercialization journey [1][2] - The company aims to focus on seven key application scenarios, including exhibition reception, entertainment performances, material handling, and scientific education, while also exploring international markets [1][4] Group 1: Strategic Partnerships - Zhiyuan Robotics has a clear and diverse logic for selecting partners, including forming joint ventures with leading companies like Fulian Precision Engineering to refine products in real environments [2][6] - Collaborations with technology service giants like Softcom Power leverage their extensive customer networks to expand into vertical markets such as government services [2][9] - Partnerships with technical experts like Lingxi Technology enhance Zhiyuan's capabilities in industrial vision, which is crucial for precise robotic operations [2][6] Group 2: Technological Breakthroughs - Key technological advancements include fully autonomous movement, adaptability to extreme conditions, reliability demonstrated through extensive testing, and ease of task deployment [8][14] - The company predicts that humanoid robots will soon surpass human speed in running, with aspirations for robots to serve as personal assistants in home care scenarios [2][18] Group 3: Commercialization Metrics - A clear metric for successful commercialization is achieving over 1,000 units operating in a single scenario while maintaining high daily activity rates, indicating maturity in technology, product, and customer demand [3][16] - The company is currently experiencing rapid growth, with expectations of delivering thousands of units this year [16] Group 4: International Expansion - Zhiyuan Robotics is focusing on international expansion through partnerships in North America, Southeast Asia, Japan, and the Middle East, with initial progress noted in Malaysia [1][4] - The company is leveraging educational collaborations as a gateway to enter the North American market [4] Group 5: Industry Challenges - The company disagrees with the notion that hardware is not a bottleneck, emphasizing the need for hardware improvements in areas such as tactile feedback, joint performance, and safety [2][14] - The ROI for industrial applications is targeted to be within three years, contingent on optimizing operational efficiency and reducing costs [14][15]

Core Insights - The article discusses advancements in generative AI models, highlighting new releases and updates from various companies, including Nvidia, OpenAI, and Tencent, among others. Group 1: Nvidia's Nemotron Nano 2 Model - Nvidia released the Nemotron Nano 2 model with 9 billion parameters, utilizing a Mamba-Transformer hybrid architecture, achieving inference throughput up to 6 times that of traditional models [1] - The model competes with Qwen3-8B, showing comparable or superior performance in mathematics, coding, reasoning, and long-context tasks, fully open-source and supporting a context length of 128K [1] - It was trained on 20 trillion tokens, compressing a 12 billion parameter model to 9 billion, and can be run on a single A10G GPU [1] Group 2: OpenAI's GPT Model Comparison - OpenAI's president Greg Brockman shared a comparison of responses from GPT-1 to GPT-5 using the same prompts, showcasing significant improvements in knowledge retention, logical structure, and language coherence [2] - The results indicated that earlier models like GPT-1 and GPT-2 often produced nonsensical answers, while GPT-5 provided more logical, rich, and emotionally valuable responses [2] - Interestingly, some users expressed a preference for the earlier models, finding them more "wild" and "unconventional," with GPT-1 being likened to "true AGI" [2] Group 3: DeepSeek Model Update - DeepSeek's latest online model has been upgraded to version 3.1, extending context length to 128K, available through official web, app, and mini-programs [3] - This update is a routine version iteration and is not related to the anticipated DeepSeek-R2, which is not expected to be released in August [3] - The expanded context capacity will enhance user experience in long document analysis, codebase understanding, and maintaining consistency in long conversations [3] Group 4: Nano Banana Model - The mysterious AI drawing model Nano Banana demonstrated exceptional character consistency in LMArena evaluations, accurately preserving facial features and expressions, outperforming competitors like GPT-4o and Flux [4] - Although not officially claimed, the model is said to originate from Google DeepMind and is currently only available in LMArena's battle mode without a public interface [4] - Besides character consistency, it excels in background replacement, style transfer, and text modification, effectively executing various complex image editing tasks [4] Group 5: Alibaba's Qwen-Image-Edit Model - Alibaba launched the Qwen-Image-Edit model, based on its 20 billion parameter Qwen-Image model, which supports both semantic and appearance editing capabilities [5][6] - The model can perform precise text editing while retaining the original font, size, and style, achieving state-of-the-art performance in multiple public benchmark tests [6] - It has shown excellent performance in tasks like adding signage, replacing backgrounds, and modifying clothing, though it still faces limitations in multi-round modifications and complex font generation [6] Group 6: Tencent's AutoCodeBench Dataset - Tencent's Mixyuan released the AutoCodeBench dataset to evaluate large model coding capabilities, featuring 3,920 high-difficulty problems across 20 programming languages [7] - The dataset is notable for its high difficulty, practicality, and diversity, with existing evaluations showing that leading industry models scored below 55, indicating its challenge [7] - A complete set of open-source tools is also available, including the data generation workflow AutoCodeGen and the evaluation tools AutoCodeBench-Lite and AutoCodeBench-Complete [7] Group 7: Higgsfield's Draw-to-Video Feature - AI startup Higgsfield introduced the Draw-to-Video feature, allowing users to draw arrows and shapes on images and input action commands to generate cinematic dynamic visuals [8] - This feature is complemented by the Product-to-Video function, supporting various video generation models, making it easier to create advertisement videos compared to text prompts [8] - Founded in October 2023, Higgsfield has garnered attention for its advanced cinematic control technology and user-friendly design [8] Group 8: Zhiyuan's A2 Humanoid Robot - Zhiyuan Robotics completed a 24-hour live broadcast of its humanoid robot A2 walking outdoors, achieving this feat in high temperatures of 37°C and ground temperatures of 61°C [9] - The A2 showcased strong environmental adaptability, autonomously navigating obstacles, planning paths, and adjusting gait without remote control, utilizing "hot-swappable" battery technology for quick recharging [9] - During the event, three industry dialogues were held to discuss the development path of humanoid robots, marking a significant milestone in transitioning from technology development to commercial production [9] Group 9: Richard Sutton's OaK Architecture - Richard Sutton, the father of reinforcement learning and 2024 ACM Turing Award winner, introduced the OaK architecture (Options and Knowledge), outlining a path to superintelligence through operational experience [10][11] - The OaK architecture consists of eight steps, including learning policies and value functions, generating state features, and maintaining metadata [11] - It emphasizes open-ended abstraction capabilities, enabling the active discovery of features and patterns during operation, though key technological prerequisites like continuous deep learning must be addressed to realize the superintelligence vision [11] Group 10: OpenAI's GPT-5 Release Review - OpenAI's VP and ChatGPT head Nick Turley acknowledged the failure to continue offering GPT-4o, underestimating user emotional attachment to models, and plans to provide clearer timelines for model discontinuation [12] - Turley noted a polarized user base, with casual users preferring simplicity while heavy users require complete model switching options, aiming to balance both needs through menu settings [12] - Regarding the business model, Turley mentioned strong growth in subscription services, with enterprise users increasing from 3 million to 5 million, and future exploration of transaction commissions while ensuring commercial interests do not interfere with content recommendations [12]

Core Viewpoint - The robot concept stocks have experienced a significant surge, with over 40 related stocks hitting the daily limit or rising more than 10% on August 18, indicating a robust market interest and potential growth in the robotics industry [1][3]. Group 1: Market Performance - On August 18, stocks such as Xiangming Intelligent, Qiangrui Technology, and Zhongke Haixun saw a daily limit increase of 20%, while many others also performed strongly, contributing to a historical high for the Wind Robot Index and the Tongdaxin "Robot Concept" Index [1][3]. - The overall trend since late June shows a continuous rise in robot concept stocks, reflecting growing investor confidence in the sector [3]. Group 2: Product Launches - Zhiyuan Robotics announced the launch of its entire product line, including six major product categories, marking its full entry into the commercialization phase [1][6]. - The pricing for various humanoid robots has been set, with models like the Yuanzheng A2 priced at 168,000 yuan and the Lingxi X2 at 98,000 yuan, indicating a diverse product offering aimed at different market segments [6]. Group 3: Industry Events - The 2025 World Humanoid Robot Games concluded on August 17, showcasing the advancements in robot performance across various tasks, which has garnered significant attention and is expected to drive further industry development [1][6]. - The event featured 280 teams from 16 countries competing in 26 events, highlighting the global interest and competitive nature of the humanoid robotics field [6]. Group 4: Technological Advancements - Analysts noted significant improvements in humanoid robots' stability, joint performance, control algorithms, and overall capabilities compared to earlier in the year, suggesting accelerated industry development [9][10]. - The performance of robots in competitive scenarios, such as the 1500-meter race completed in 6 minutes and 34 seconds, demonstrates advancements in speed and control, indicating a leap in technology [9][10]. Group 5: Future Outlook - The humanoid robot industry is projected to see a cumulative global demand of approximately 2 million units by 2030, with current developments marking a critical breakthrough phase [10]. - The trend towards lightweight designs is expected to enhance humanoid robots' performance and safety, presenting investment opportunities in the related supply chains [10][11].

2025年8月15日上午，北京国家速滑馆内，九尊身披3D打印秦俑战甲的机器人随鼓声"苏醒"。 它们身高1.7米、体重70公斤，在人类舞者的引领下迈出凝滞而有力的步伐。当队形从方阵渐变为长 龙，金属关节折射的冷光与赳赳大秦的雄浑气势交织，现场评委打出92.54的高分——全球首个人形机 器人运动会的首枚群体金牌由此诞生。 从"机械木偶"到"文化载体" 这场大型人机共舞表演，是由光谷东智与北京舞蹈学院联袂打造、智元远征A2团队赋能。智元机器人 提供了高性能的人形机器人本体远征A2及核心算法，光谷东智在智元机器人平台基础上，进行二次开 发和深度定制。 《秦俑魂》原本是中国古典舞的经典独舞作品，此次改编为人机共舞的群体表演，难度呈几何倍数增 长。为契合秦俑高大威严的形象，团队选用了体型较大、更具视觉冲击力的1.7米全尺寸机器人，这对 机器人的稳定性、协调性及运动控制能力提出了极高要求。 同时，软件算法也历经无数次调试与迭代，通过先进的运动规划算法，精确计算每一个动作的轨迹、速 度与力度，让机器人的舞动既流畅自然，又能精准卡点音乐节奏。 舞蹈动作采集是由舞蹈演员穿上动作捕捉服进行的。由于机器人缺乏人类天生的平衡感与柔韧性 ...

Core Insights - The 2025 World Artificial Intelligence Conference (WAIC) showcased significant advancements in humanoid robots, indicating a pivotal shift from laboratory experiments to large-scale applications in the coming years [1][5] - The Chinese robotics industry is experiencing rapid iteration, with a notable increase in financing events for humanoid robots in the first half of 2025, surpassing the total for the previous year [1][5] Group 1: Technological Advancements - Humanoid robots have shown qualitative improvements in movement precision and task completion compared to previous years, with capabilities such as flexible navigation in complex terrains and more fluid grasping actions [1][5] - The integration of AI large models with robotic systems has enhanced their operational capabilities, allowing for real-time decision-making and improved adaptability to various environments [6][7] - Companies like Magic Atom and Beijing Humanoid Robot Innovation Center are demonstrating advanced collaborative capabilities among robots, enabling them to communicate and execute tasks autonomously [6][7] Group 2: Market Dynamics - The financing landscape for humanoid robots is expanding, with leading companies like Yushu Technology and Zhiyuan Robotics securing significant contracts, indicating a growing market demand [5][7] - The industry is witnessing a clear differentiation among companies, with those unable to achieve "intelligent closed-loop + mass production capability" facing potential risks of elimination [7][8] - The ongoing investment boom in the humanoid robot sector is expected to continue, driven by the increasing recognition of their practical applications and market potential [7][8] Group 3: Application Scenarios - Humanoid robots are being deployed in diverse fields, including industrial tasks, education, and rehabilitation, showcasing their versatility and potential for real-world applications [2][4] - The introduction of interactive robots designed for rehabilitation and emotional companionship reflects a growing focus on human-robot interaction and care [4][7] - The advancements in quadruped robots, which have been operational in practical scenarios longer than humanoid robots, highlight the broader applicability of robotic technology across various sectors [5][6]