Core Viewpoint - Li Auto is undergoing a significant organizational restructuring to enhance its AI strategy, aiming to align with the operational models of leading AI companies and redefine personnel roles for collaborative development of silicon-based life forms [4][6]. Group 1: Organizational Changes - A new software core team has been established, led by Vice President of Intelligent Space, Guo Xiaofei [4]. - A humanoid robot team has been formed, headed by Senior Vice President of Autonomous Driving R&D, Lang Xianpeng [4]. - The base model team will be led by Zhan Kun, with most employees from the autonomous driving department transitioning to this team [4]. Group 2: Leadership Background - Guo Xiaofei previously served as Chief Architect at LeEco Auto and has been involved in key projects at Li Auto, including the intelligent cockpit system for Li ONE [6]. - Lang Xianpeng joined Li Auto in 2018 and has risen from R&D Director to Senior Vice President, previously working at Baidu on various autonomous driving projects [6][7]. - Zhan Kun, who graduated from Beihang University in 2016, has focused on autonomous driving architecture and technology stack deployment since joining Li Auto in 2021 [7]. Group 3: Future Plans and Innovations - Li Auto plans to launch its "end-to-end + VLM" solution by October 2024, aiming to position itself among the top tier in driving experience [7]. - The full rollout of the VLA solution is scheduled for September 2025, with an updated OTA 8.2 version and enhanced VLA driver model expected in January 2026 [7].

蓝山进入经销商渠道。 文 丨 曾兴 编辑 丨 龚方毅 长城汽车近五年来一直在转型，其中冲击高端化的魏牌尤受关注。去年 12 月初，魏牌 CEO 从 新势力背景的冯复之切换为 "老长城" 赵永坡后，相关的调整还在继续。 我们独家获悉，去年 12 月中旬，长城集团在内部宣布，将魏牌调整至品牌公司并更名为魏品 牌，组织内按车型设置作战单元——高山、蓝山、摩卡和 "DE" 作战单元，以及品牌策划、销 售管理、运营管理等职能部门。魏牌曾在 2024 年底从品牌公司拆出与长城智选整合，并在去 年 5 月对外官宣更名为 "魏牌新能源"，相当于独立运营。 渠道端配合着品牌和组织的调整。两位长城渠道人士告诉我们，从 2 月开始，魏牌蓝山车型将进入经 销商渠道，和直营门店同步销售，目前已在向经销商培训。"经销商将用代理模式，和直营店一样在 APP 订车，避免扰乱价格。" 这种模式下经销商不承担库存，用户通过官方渠道下单，厂家可以控制 终端价格稳定。目前小米、极氪均使用代理制。 一位长城渠道侧人士说，"2026 年会减少开商超店，还会开一些综合店或汽车商圈的零售店。" 我们 了解到，关于今年新开门店的数量长城内部还在商讨，目前设定 ...

汽车行业里，达成百万辆里程碑的用时长短，常被视为衡量品牌发展速度的指标之一。近日，问界品 牌迎来第 100 万台整车下线，站上全新的发展阶段。 从 0 到 100 万辆，问界用时 46 个月，创下多项行业纪录。作为对比，特斯拉用了约 12 年。而且问界 的产品矩阵清晰地锚定高端市场，综合各车型销量加权计算，问界单车均价超过 40 万元以上，这便 更加难得。 问界百万辆是一种新造车模式从探索到规模化落地的标志性节点。 文 丨 沈行 "选车过程从纠结到坚定，是因为问界始终懂得我们真正需要的是什么。" 此前有车主在交流会中提 到，对比其他豪华品牌的 "符号化豪华"，问界的 "新豪华" 会将生活需求融入每一处设计。 正如赛力斯集团董事长（创始人）张兴海在央视《微对话》采访中表示，问界的这 100 万，代表着高 质量、高端车，是高于传统豪华品牌在中国的售价，是符合新质生产力推进的高质量发展。 此外，问界百万辆也是一种全新造车模式从探索到规模化落地的标志性节点，为整个行业提供了向上 升维的路径——跨界融合与用户思维。 赛力斯和华为跨界融合始于 2021 年，赛力斯负责整车研发、制造和交付，华为提供智能座舱、辅助 驾驶系统 ...

以下文章来源于晚点AI ，作者晚点团队 晚点AI . 关注人工智能的一切，一切都关于人工智能。 不做 VLA；不做遥操，以可穿戴设备采真实数据。 文 丨 程曼祺 编辑 丨 宋玮 2024 年 11 月，《晚点 LatePost》曾独家报道，华为车 BU 前首席科学家陈亦伦投身具身智能创业。 不久后，它石智航浮出水面——由陈亦伦和百度前自动驾驶事业部负责人李震宇等联合创立；成立不 足半年，它石先后完成 1.2 亿美元和 1.22 亿美元的天使轮与天使 + 轮融资，创下中国具身智能领域天 使轮融资纪录，投资方包括蓝驰、美团、启明、线性、高瓴、联想创投等。 陈亦伦获得了很多支持，在技术思考上却 "离经叛道"。用一种平时的口吻，陈亦伦分享了很多反主流 的判断。 Google RT-2 开启的 VLA（视觉-语言-动作）模型，是当前具身模型的主流架构。而它石则开发了 AWE（AI World Engine）： 我们追求表达时间、空间、力和环境交互等物理量和 "世界信息"，而不是 VLA 那样的 "视网膜信息"。 VLA 的主流做法是从 LLM（大语言模型）得到 VLM（视觉-语言模型），再在 VLM 基础上训 VLA。 ...

福兰口中的旧模式，曾是跨国车企的黄金法则：开发一个平台，在全球铺开产能，用销量摊薄成本， 追求份额最大化。雷诺-日产-三菱联盟是这种模式的典型产物。卡洛斯·戈恩执掌时期，这个联盟一度 冲击全球销量第一，年销量超过 1000 万辆。 而他此行聊得最多的，是如何让欧洲人学会中国速度，使未来的全球项目都能具备同样的高效率。此 前雷诺集团中国工程副总裁 Jeremie Coiffier 也说过 "在中国，决策按日推进；在欧洲，我们习惯按周 来"。去年 11 月，雷诺 Twingo E-Tech 在法国上市，在中国研发中心 ACDC 的协同下，项目从立项 到量产准备只用了 21 个月，欧洲车企的常见周期是三年以上。 雷诺 Twingo E-Tech。指导价与比亚迪海鸥持平、约 2 万欧元。但在法国补贴后售价约 1.4 万欧元。电池容 量 27.5 kWh，WLTP 续航里程 263 公里。 旧模式远去，"最重要是学会中国速度"。 文 丨 龚方毅 "过去的汽车本质上是一个基于规模和销量的生意，通过规模效应来摊薄成本，提升产能利用率。那种 模式已经结束了。" 2026 年 1 月，雷诺集团 CEO 福兰（François ...

李想认为机器人主要有两种分类形式：一种像人，一种像车。 文 丨 赵宇 李梓楠 编辑 丨 龚方毅 理想汽车在官网社招页面释放若干人形机器人研发岗位，重启具身机器人研发。 从招聘需求看，理想的机器人研发计划与特斯拉、小鹏等追求极致仿生人形机器人的公司类似。这种 路线基于一种基本假设——现实世界是为人类打造的，机器人只有尽可能像人，才能使用人的工具， 掌握人的技能，进而将人从繁重的劳动中解放出来。 一位行业人士称，灵巧手是人形机器人硬件上最难的部分，也是体现研发具身机器人意愿强弱的点。 它既是执行动作的机械部件，也是收集触觉数据的重要传感器。当前主流的 VLA（视觉-语言-动作） 模型训练中，如果缺乏灵巧手的触觉反馈，机器人算法模型就缺失了 "动作" 这个关键模态的数据，影 响其智能化水平。一些机器人厂商曾在早期采用供应商方案。 2024 年 1 月，廖平平加入理想任机器人算法专家，主要负责带领团队开发轮式双臂人形机器人，用于 理想汽车制造工厂内 "螺钉拧紧" 等细分工业场景。他硕士和博士均毕业于北京航空航天大学，有约 10 年机器人研发和管理经验。 受技术发展限制、人形机器人供应链不成熟等因素影响，理想一度暂停推 ...

Core Viewpoint - The founders of Hesai Technology have established a new robotics company, Sharpa, which aims to develop general-purpose robots and core components, indicating a significant shift in focus from automotive lidar technology to robotics, which they believe has a much larger market potential [2][3][5]. Company Overview - Sharpa was founded by Hesai's co-founders, who will not hold operational roles but will guide the strategic direction. The company operates independently and has over 100 employees, primarily recruited externally [3][5]. - The first product, SharpaWave, is set to launch in October 2025 and features 22 degrees of freedom, allowing it to perform tasks like playing table tennis and origami, with a price tag of several tens of thousands of dollars [6][7]. Market Potential - The general robotics market is projected to have a trillion-dollar potential, with the founders aiming to create a world-class AI company. They believe that the opportunities in robotics are ten times greater than those in the electric vehicle sector, but the challenges are also significantly higher [5][8]. - The founders anticipate that the robotics lidar market could surpass the automotive lidar market, and they plan to develop additional components needed for robotics beyond just lidar [6]. Industry Landscape - There are over 100 startups in China focused on humanoid robots, with at least 300 entrepreneurs exploring various key components and applications in robotics [7]. - The technology roadmap for general-purpose robots is still evolving, with many companies exploring commercial applications [7]. Future Outlook - The first generation of Sharpa's products will primarily target research and experimental applications, with plans to enter service-oriented scenarios and eventually the household market, aiming for a cost below $5,000 for consumer models [7]. - The founders believe that the current moment in the robotics industry lacks a transformative breakthrough akin to the emergence of ChatGPT in the tech sector [8].

Group 1 - The core strategy of the company has become clearer, focusing on the VLA (Vision-Language-Action) model as the central AI direction [2][3] - The VLA model development has transitioned from being managed by two separate departments to a unified approach, emphasizing the importance of embodied intelligence in automotive products [5][6] - The new head of the VLA model, Zhan Kun, has a background in autonomous driving algorithms and has been promoted to oversee the integration of technology for autonomous driving and smart cockpit applications [3][6] Group 2 - The VLA model consists of three components: "V" for perception, "L" for planning and decision-making, and "A" for action, applicable to both physical and digital environments [5][6] - The company plans to fully implement the VLA driver model in the AD Max vehicle by September 2025, aiming for smoother longitudinal control and improved accuracy at complex intersections [7] - The company is also set to enhance language and behavioral interactions and adapt its self-developed M100 chip by 2026 [7]

Core Viewpoint - The article discusses the competitive landscape of intelligent driving in China, highlighting the potential for latecomers to succeed by leveraging advanced technology and strategic partnerships, as exemplified by Geely and Qianli's collaboration on the G-ASD system [3][6][11]. Group 1: G-ASD System Overview - The G-ASD system, developed by Geely and Qianli, utilizes a state-of-the-art end-to-end model architecture, enabling full-range intelligent driving capabilities without relying on high-precision maps [4][10]. - The system is powered by 23.5 EFLOPS of computing power and vast amounts of real driving data, allowing for rapid iteration and self-evolution through a "world behavior model" [4][10]. Group 2: Strategic Insights - Geely's CEO emphasizes the integration of new energy and intelligence as a unified goal, aiming to establish Geely's intelligent driving as a benchmark in China and compete globally [6][11]. - Qianli's competitive edge lies in its lack of historical technical burdens and a composite team with expertise from various sectors, enhancing its ability to innovate and execute [7][8]. Group 3: Market Positioning and Future Outlook - The article predicts that only three to four leading intelligent driving suppliers will emerge globally, with Qianli and Geely's collaboration positioned to capitalize on economies of scale and technological advancements [11][24]. - Qianli aims to focus on a "big customer strategy," seeking deep partnerships with a select few major automotive clients to build a robust competitive barrier [24][25].

Core Viewpoint - The launch of the L90 model by NIO marks a significant turning point for the company, showcasing a competitive pricing strategy that has revitalized its market position and sales performance [2][3]. Group 1: Product Strategy and Market Positioning - The L90 was priced at 279,900 yuan during its pre-sale, breaking previous pricing norms for NIO and leading to a 6.5% increase in its stock price on the day of the announcement [2]. - The official launch price was further reduced to 265,800 yuan, making it only 10,000 yuan more expensive than the top configuration of the L60, while offering a larger size and more features [2]. - The L90 successfully filled a market gap for electric large six-seat SUVs priced under 300,000 yuan, achieving over 10,000 deliveries in its first full month, accounting for nearly one-third of NIO's total sales during that period [2][4]. Group 2: Product Development Insights - The product definition of the L90 focused on comfort and safety for six adult passengers, addressing common complaints about third-row seating in other models [7]. - The L90 features a length of 5,145 mm and a wheelbase of 3,110 mm, making it larger than mainstream extended-range SUVs, while its design allows for a spacious interior without sacrificing aerodynamics excessively [7][8]. - The introduction of a front trunk with a capacity of 240 liters was a key feature, enabling the vehicle to accommodate six passengers and their luggage comfortably [9]. Group 3: Organizational Changes and Team Dynamics - The management changes in April 2022 led to improved team morale and efficiency, allowing for quicker decision-making and product development [3]. - The L90 project benefited from a collaborative approach among various teams, overcoming significant engineering and design challenges to meet user needs [12]. - The shift in focus from a global market strategy to a more localized approach allowed the L90 team to optimize the vehicle for the Chinese family market, enhancing its appeal [17][18]. Group 4: Market Trends and Competitive Landscape - The L90's launch coincided with a growing trend in the Chinese market for large SUVs, with several competitors also entering this segment, although it remains a niche market [24]. - Despite initial success, the L90's sales began to decline, with retail sales dropping to approximately 6,000 units in November, indicating potential market saturation or increased competition [24]. - The overall strategy of NIO has shifted towards ensuring sustainability and competitiveness in a challenging market environment, emphasizing the importance of adapting to consumer preferences [30].