12月11日，小鹏汽车董事长何小鹏发文称，小鹏汽车的VLA 2.0（Vision-Language-Action，视觉语言动作模型）将在下个季度发布，"因为是第一个版本， 所以压力很大"。 此外，何小鹏还与公司自动驾驶团队立下特别"赌约"：若2026年8月30日前，小鹏VLA系统在国内达到特斯拉FSD V14.2版本在硅谷的整体效果，他将在硅 谷筹建特色中国风味食堂。反之，自动驾驶负责人需在金门大桥完成裸跑挑战。 就在前一天，理想汽车自动驾驶研发高级副总裁郎咸朋在社交平台发布长文，回应宇树科技创始人王兴兴此前对VLA模型的质疑。 "我跟王兴兴观点最不一样的地方在于，他认为模型架构更重要，但我认为模型的关键是要与整个具身智能系统适配。在此基础上，数据是起决定意义 的。"郎咸朋认为，"VLA就是自动驾驶最好的模型方案。" 据了解，VLA被业内视为端到端方案的"智能增强版"。其名称中的V代表视觉感知（Vision），A代表动作执行（Action），而中间的L则是大语言模型 （Language Model）。V负责实时感知环境，A负责输出具体控制指令，L则像"中台"一样，把感知信息转译为可供A执行的规划与决策。 清华 ...

每经记者｜孙磊 每经编辑｜段炼 裴健如 12月11日，小鹏汽车董事长何小鹏发文称，小鹏汽车的VLA 2.0（Vision-Language-Action，视觉语言动作模型）将在下个季度发布，"因为是第一个版本， 所以压力很大"。 图片来源：何小鹏微博 此外，何小鹏还与公司自动驾驶团队立下特别"赌约"：若2026年8月30日前，小鹏VLA系统在国内达到特斯拉FSD V14.2版本在硅谷的整体效果，他将在硅 谷筹建特色中国风味食堂。反之，自动驾驶负责人需在金门大桥完成裸跑挑战。 就在前一天，理想汽车自动驾驶研发高级副总裁郎咸朋在社交平台发布长文，回应宇树科技创始人王兴兴此前对VLA模型的质疑。 "我跟王兴兴观点最不一样的地方在于，他认为模型架构更重要，但我认为模型的关键是要与整个具身智能系统适配。在此基础上，数据是起决定意义 的。"郎咸朋认为，"VLA就是自动驾驶最好的模型方案。" 近几年，辅助驾驶行业经历了多次"技术底座"的范式迁移——从企业普遍把激光雷达+高精地图奉为"黄金组合"，到引入BEV（鸟瞰图）+Transformer摆脱 高精度地图，再到端到端将辅助驾驶带入AI时代，企业普遍按照这个路径来推进辅助驾 ...

Core Viewpoint - The company has shown significant growth in vehicle deliveries and revenue, with a strong focus on technological collaboration and product innovation, while also facing challenges in profitability and cost management [1][2][3]. Group 1: Financial Performance - In Q3, the company delivered 116,000 vehicles, a year-on-year increase of 149%, leading to a revenue growth of 102% to 20.38 billion yuan [1]. - The automotive sales revenue and service revenue were 18.1 billion yuan and 2.3 billion yuan, respectively, reflecting year-on-year growth of 105% and 78% [1]. - The overall gross margin for Q3 was 20.1%, an increase of 4.9 percentage points year-on-year and 2.8 percentage points quarter-on-quarter [1]. - The net loss attributable to the parent company in Q3 was 380 million yuan, significantly narrowing compared to previous periods [1]. Group 2: Future Outlook - For Q4, the company expects vehicle sales between 125,000 and 132,000 units, representing a year-on-year growth of 36.6% to 44.3%, with total revenue projected between 21.5 billion and 23 billion yuan [2]. - The company anticipates maintaining a gross margin of around 20% in Q4 [2]. Group 3: Product and Market Development - The company plans to launch the X9 super range extender version on November 20, with three additional super range extender models set for Q1 2026 [3]. - The company is focusing on enhancing brand strength and emotional value through improved design and technology, with key design personnel joining from prestigious automotive brands [3]. - In the overseas market, the company delivered 29,706 vehicles in the first three quarters, with September sales surpassing 5,000 units, a year-on-year increase of 79% [4]. Group 4: Technological Advancements - The company announced the latest generation of its intelligent driving model, VLA 2.0, which has been trained on nearly 100 million clips, equating to extreme driving scenarios [5]. - The upcoming "Xiaolu NGP" feature is expected to significantly enhance intelligent driving performance in complex environments [5]. - The company plans to introduce three Robotaxi models in 2026, aiming for trial operations in the domestic market [6].

Summary of XPeng's AI Workshop and Equity Research Company Overview - **Company**: XPeng Inc. - **Industry**: Electric Vehicles (EV) and Robotics - **Founded**: 2015 - **Market Cap**: $39.0 billion - **Ticker**: XPEV (US), 9868 HK (Hong Kong) Key Highlights from AI Workshop 1. Robotaxi Development - XPeng is planning the operational rollout of its robotaxi service, focusing on city selection, site layout, and license applications - Target for mass production and commercial deployment is set for the second half of 2026 - The company is addressing specific challenges to bridge the gap between Level 4 (L4) and Level 2 (L2) autonomy, including passenger pickup and emergency fallback [2][2][2] 2. VLA 2.0 Technology - VLA 2.0 aims to simplify traditional language model approaches by avoiding manual data labeling, enhancing scalability - The model can adapt to new regions using local driving data without needing region-specific labeled datasets, improving localization efficiency [4][4][4] 3. Humanoid Robots - XPeng is developing full-sized humanoid robots designed to integrate into human environments, emphasizing a shift from hardware-driven to software-driven innovation - The humanoid robots utilize a unique waist design mimicking the human spine, enhancing posture and movement [6][6][6] - The robots share foundational hardware with XPeng's EVs, allowing for deep cross-domain integration and scalability [6][6][6] Financial Insights 1. Equity Research Rating - **Rating**: Buy - **Current Price**: $22.42 (US), HK$90.90 (Hong Kong) - **Price Target**: $29.90 (US) +33%, HK$116.90 (Hong Kong) +29% [5][5][5] 2. Price Target Methodology - Price targets are based on a price-to-earnings (P/E) ratio of 25x FY27E earnings - Risks include intensifying competition in China's New Energy Vehicle (NEV) market and potential setbacks in autonomous driving development [9][10][10] Additional Considerations - XPeng's humanoid robots could create significant demand in corporate settings, with potential deployment in over 20,000 Starbucks locations in China, indicating scalability across industries [6][6][6] - The company is actively exploring low-cost data acquisition methods to support continuous learning and iteration for its robotics technology [6][6][6] Conclusion XPeng is positioning itself as a leader in the EV and robotics sectors with innovative technology and strategic planning for future developments. The company's focus on scalability, integration, and adaptability presents significant investment opportunities, albeit with inherent risks in a competitive market.

Investment Highlights - This week's sector performance ranking: SW Commercial Passenger Vehicles (+0.8%) > SW Auto Parts (-0.2%) > SW Autos (-1.2%) > SW Commercial Freight Vehicles (-3.2%) > SW Passenger Vehicles (-3.4%) [4][13] - Top five stocks covered this week include Weichai Power, China National Heavy Duty Truck Group H, Naisite, NIO-SW, and Jifeng Co., Ltd. [4][13] Research Outcomes - The team released Q3 performance summaries and reports on Top Group, Xusheng Group, Guangyang Co., and Bertley, as well as a report on Xingyuan Zhuomei [5][13] Industry Core Changes 1. Tesla's $1 trillion compensation plan was approved, with performance targets including the delivery of 20 million vehicles, 10 million active FSD users, 1 million Robotaxi commercial operations, delivery of 1 million Optimus robots, and an adjusted annual EBITDA target of $400 billion [6][13] 2. XPeng's Technology Day introduced VLA 2.0, the first fully self-developed Robotaxi, the new generation humanoid robot IRON, and two flight systems [6][8][13] 3. Seres, Pony.ai, and WeRide officially listed on the Hong Kong Stock Exchange [6][8][13] Current Automotive Sector Configuration - The automotive industry is at a crossroads: the electric vehicle (EV) boom is nearing its end, while automotive intelligence is in a "dark before dawn" phase, and robotics innovation is in the 0-1 stage. Three main investment opportunities are present during this transition [9][15] - **AI Smart Vehicle Main Line**: Focus on Robotaxi/vans first, followed by C-end applications. Key downstream targets include: - Robotaxi perspective: Integrated models (Tesla, XPeng, Qianli Technology); technology providers + operational sharing models (Horizon Robotics, Baidu, Pony.ai, WeRide); transformation of ride-hailing/taxi services (Didi, Cao Cao Mobility, Ruqi Mobility, Dazhong Transportation, Jinjiang Online) [9][15] - Robovan perspective: Desay SV, Zhongyou Technology, Kaile Co., etc. [9][15] - C-end vehicle sales perspective: Complete vehicles (XPeng, Li Auto, Huawei, Xiaomi, etc.) [9][15] - **Upstream Supply Chain Key Targets**: - B-end vehicle OEMs: BAIC Blue Valley, GAC Group, SAIC Group, etc. [9][15] - Core suppliers: Testing (China Automotive Research, China Automotive Industry Group, etc.), chips (Horizon Robotics, Black Sesame Intelligence), domain controllers (Desay SV, Jingwei Hirain, Joyson Electronics, Coboda, etc.), sensors (Sunny Optical Technology, Hesai Technology, Suteng Juchuang), steer-by-wire chassis (Bertley, Naisite, Zhejiang Shibao), lights (Xingyu Co., Ltd.), glass (Fuyao Glass) [9][15] - **AI Robotics Main Line**: Preferred components (Top Group, Joyson Electronics, Xinquan Technology, Precision Forging Technology, Fuda Co., Ltd., Xusheng Group, Aikedi, etc.) [9][15] - **Dividend & Good Pattern Main Line**: Buses (Yutong Bus), heavy trucks (China National Heavy Duty Truck A-H, Weichai Power), two-wheelers (Chunfeng Power, Longxin General) [10][15]

Investment Rating - The report maintains a positive outlook on the automotive sector, particularly emphasizing investment opportunities in AI and smart vehicles [1][3]. Core Insights - The automotive industry is at a crossroads, transitioning from electric vehicle (EV) benefits to a focus on smart technology and robotics. Investment opportunities are identified in three main areas: AI smart vehicles, AI robotics, and traditional vehicle segments [3][54]. - Key developments include Tesla's $1 trillion compensation plan approval, aiming for significant production and delivery targets, and Xiaopeng's launch of new AI-driven products [2][3][62]. Summary by Sections Market Performance - This week, the automotive sector showed mixed performance, with commercial passenger vehicles leading with a +0.8% increase, while passenger cars saw a decline of -3.4% [2][7]. - The report highlights the best-performing stocks, including Weichai Power and China National Heavy Duty Truck [2][25]. Investment Opportunities - The report suggests focusing on AI smart vehicles, with key players like Tesla, Xiaopeng, and various technology providers leading the charge. The report also identifies opportunities in the supply chain, including vehicle manufacturing and component suppliers [3][54]. - The expected growth in the domestic market is projected at 4.1% year-on-year for 2025, with total retail sales anticipated to reach 23.7 million units [50][57]. Future Outlook - The report forecasts a significant increase in the penetration of L3 and L2+ autonomous driving technologies by 2025, with expected market shares of 20% and 33% respectively for new energy vehicles [52][54]. - The report anticipates a continued recovery in the bus market and a 15% growth in domestic sales driven by policy support [57]. Key Companies and Developments - Notable companies mentioned include Tesla, Xiaopeng, and various suppliers like Top Group and Yanfeng Automotive. The report emphasizes the importance of these companies in the evolving landscape of smart vehicles and robotics [61][65]. - The report also highlights the recent IPOs of companies like Xiaopeng and WeRide, indicating a growing interest in the sector [62][65].

Core Viewpoint - Xiaopeng Motors has recently released VLA 2.0, which represents a significant advancement in autonomous driving technology, particularly in the context of competing with Tesla's innovations [2][10]. Summary by Sections VLA Development - Xiaopeng's VLA is being developed in two parallel paths: V/L→A and V→L→A, with the former aligning more closely with Tesla's recent ICCV sharing, where L is not a middleware but a parallel input to V [3][6]. - The V/L→A model eliminates language translation while maintaining a focus on visual inputs [6]. Technical Specifications - The first mass-produced physical world model boasts a maximum effective computing power of 2250 TOPS [6]. - Future plans include entering the robotaxi market, utilizing four Turing AI chips with a total computing power of 3000 TOPS [8]. Industry Context - The competition in L3 technology is intensifying, with various companies analyzing and following Xiaopeng's VLA developments [10]. - The ongoing debate between world models and VLA pathways remains unresolved, indicating a need for continued exploration in both academic and industrial sectors [10]. Community and Learning Resources - The "Autonomous Driving Heart Knowledge Planet" community has been established to provide a comprehensive platform for learning and sharing knowledge in the autonomous driving field, with over 4000 members and plans to expand to nearly 10,000 [14][31]. - The community offers a variety of resources, including video tutorials, technical discussions, and job placement mechanisms, aimed at both beginners and advanced practitioners in the field [17][29][95].