Domestic News - The National Bureau of Statistics projects that profits in the high-tech manufacturing sector will grow by 13.3% in 2025 compared to the previous year, surpassing the overall industrial growth rate of 12.7% [7] - The Ministry of Transport emphasizes the implementation of a policy to waive tolls for small passenger vehicles during the Spring Festival, while also addressing the charging needs for new energy vehicles [8] - Dongguan plans to allocate 80 million yuan for new car purchase subsidies, with a maximum subsidy of 5,000 yuan per vehicle, as part of a broader 1.5 billion yuan consumer promotion initiative [9] - BYD and ExxonMobil have signed a strategic cooperation memorandum to enhance collaboration in the field of new energy hybrid technology [10] - Changan Automobile aims to achieve mass production of humanoid robots by 2028, as part of its transformation into a smart low-carbon mobility technology company [11] - NIO has surpassed 99 million battery swap operations, with a total of 8,595 energy swap stations, including 3,708 dedicated swap stations [13] - CATL has signed a comprehensive strategic cooperation agreement with the Yunnan provincial government to promote green low-carbon development and enhance cooperation in the fields of new energy batteries and green transportation [14] - Pony.ai and Aitb have reached a business cooperation agreement to build a fully autonomous driving service fleet [15] International News - India has agreed to open a car quota to the EU, allowing up to 250,000 European-made cars to enter the Indian market at reduced tariffs, significantly higher than the 37,000 quota for the UK [16] - South Africa's electric vehicle sales are projected to decline by 17% in 2025, with only 1,018 units sold, while overall new car sales are expected to reach 596,818 units, a 15.7% increase [17] - Tesla is partnering with Pilot Travel Centers to build high-power charging stations for electric trucks along major highways, with the first stations expected to be operational by late summer [18] - VinFast has partnered with Autobrains to develop autonomous driving technology, focusing on enhancing the driving assistance systems of its upcoming electric vehicles [19] Commercial Vehicles - Dongfeng Liuzhou Automobile and Yadea Technology Group have signed a strategic cooperation agreement to advance the development and application of new energy smart logistics equipment [20] - DeepWay has completed a Pre-IPO financing round, raising 1.177 billion yuan to strengthen its position in the smart new energy heavy truck sector [21] - Chery's Rely brand has launched the R08 EV, its first pure electric pickup truck, priced between 127,800 and 158,800 yuan, marking its entry into the new energy pickup market [22] - The White Rhino has begun testing a 24-cubic-meter unmanned vehicle in Shenzhen, aiming to expand its commercial scale and capture strategic opportunities in urban logistics [23]

据介绍，依托生成式AI技术，WeRide GENESIS能够在虚拟世界中快速"生成世界"——只需几分钟即可 构建高度真实的仿真城市环境，并精准复刻现实道路中难以高频获取的极端长尾场景，使自动驾驶系统 得以在虚拟环境中完成高强度训练与验证。同时，通过不断将仿真生成的经验反哺真实道路运行， WeRide GENESIS显著提升了自动驾驶算法应对真实物理世界复杂场景的能力与迭代效率，并大幅降低 传统道路测试所需的时间与成本。 消息面上，据文远知行-W官微消息，1月28日，公司正式发布自研通用仿真模型——WeRide GENESIS，构建起物理AI(Physical AI)与生成AI(Generative AI)之间的融合桥梁，贯通"现实物理世 界"与"虚拟仿真世界"，将加速自动驾驶汽车的规模化研发、训练与商业落地。 智通财经APP获悉，文远知行-W(00800)尾盘涨近3%，截至发稿，涨2.81%，报23.44港元，成交额 1583.07万港元。 ...

（文章来源：每日经济新闻） 中信建投指出，当前汽车板块处于淡季弱势表现，但市场对销量悲观预期反馈逐步钝化，悲观预期见 底。结构性看反内卷及出海预期改善，自动驾驶政策催化落地，人形机器人特斯拉V3临近推出节点。 2026年汽车以旧换新政策支撑内需，商用车或更为受益；结构性看多端侧AI（无人驾驶及机器人）商 业化0-1突破带来估值弹性。具体来看，自动驾驶方面，26年有望成为无人驾驶商业化元年，政策宽松 预期强化。人形机器人板块进入基本面兑现阶段，后续需关注一系列量产节点。商用车板块，26年补贴 政策全额延续，看好内需托底与出口放量继续支撑行业景气。 香港汽车ETF（520720）跟踪的是港股通汽车指数（931239），该指数从港股通范围内选取涉及整车制 造、零部件及智能驾驶等业务的上市公司证券作为指数样本，并重点配置新能源与智能化相关方向，以 反映港股市场汽车产业链相关上市公司证券的整体表现。 ...

Core Viewpoint - WeRide has launched its self-developed general simulation model, WeRide GENESIS, aimed at bridging the gap between physical AI and generative AI to accelerate the development, training, and commercialization of autonomous vehicles [2][3]. Group 1: Technology and Innovation - The WeRide GENESIS simulation model utilizes generative AI technology to quickly create highly realistic simulated urban environments in just a few minutes, allowing for intensive training and validation of autonomous driving systems [2]. - The model can replicate extreme long-tail scenarios that are difficult to capture frequently in real-world conditions, significantly enhancing the capability and iteration efficiency of autonomous driving algorithms in complex real-world scenarios [2]. - By continuously feeding back experiences generated in simulation to real-world operations, the model greatly reduces the time and costs associated with traditional road testing [2]. Group 2: Market Challenges and Solutions - The development of the GENESIS simulation model addresses the challenges faced in the global commercialization of autonomous driving, particularly the significant differences in road network structures, traffic habits, infrastructure, weather, and regulations across different cities [3]. - Real-world testing has limitations in efficiency and scene coverage, making it difficult to support large-scale validation needs across multiple cities and scenarios, thus necessitating a comprehensive simulation platform [3]. - The simulation model is compatible with diverse urban road elements, sensor perspectives, and configuration setups, allowing for training and testing of various vehicle levels from L2++ ADAS to L4 Robotaxi on a single platform [3]. Group 3: Company Positioning - WeRide currently holds autonomous driving licenses in eight countries and has deployed autonomous vehicles in over 40 cities across 11 countries [3].

Core Viewpoint - The article discusses the development of SplatSSC, a novel framework for Semantic Scene Completion (SSC) that addresses the limitations of traditional dense grid representations by utilizing a depth-guided approach and decoupled aggregation mechanism to enhance performance and efficiency [3][4]. Group 1: Challenges in Traditional Methods - Traditional dense grid representations in SSC have been limited by two main issues: low utilization rates of randomly initialized Gaussian primitives (approximately 3.9%) and the generation of erroneous semantic fragments known as "Floaters" due to isolated outliers [3][4]. - The existing methods often rely on large-scale random distributions of Gaussian primitives, leading to significant computational redundancy and wasted model capacity [6]. Group 2: SplatSSC Framework - SplatSSC introduces an innovative depth-guided strategy and a decoupled aggregation mechanism, resulting in a significant leap in performance and efficiency [4]. - The framework employs a parallel branch strategy, integrating a learnable image encoder for multi-scale semantic extraction and a pre-trained Depth-Anything model for stable depth features [10]. Group 3: Core Technologies - The Group-wise Multi-scale Fusion (GMF) module in SplatSSC replaces random initialization with precise guidance using geometric priors, requiring only 1,200 Gaussian primitives (about 7% of previous methods) to effectively cover spatial distributions [11][13]. - The Decoupled Gaussian Aggregator (DGA) is designed to combat the "Floaters" issue by decoupling occupancy probability from semantic contributions, ensuring clean scene boundaries [15][19]. Group 4: Experimental Validation - SplatSSC achieved state-of-the-art (SOTA) performance on the Occ-ScanNet dataset, with an Intersection over Union (IoU) score of 62.83% and a mean IoU (mIoU) of 51.83%, surpassing previous SOTA methods by 6.35% and 4.16% respectively [22][23]. - The model demonstrated superior fine-grained perception capabilities, particularly in recognizing intricate objects like chair legs and table surfaces [22]. Group 5: Efficiency and Resource Management - SplatSSC's design allows for a significant reduction in inference latency (approximately 9.3% to 115.63 ms) and memory consumption (approximately 9.6%), while maintaining a stable parameter scale with only a 0.19% increase [34]. - The framework's efficiency is highlighted by its ability to achieve high-quality scene reconstruction with fewer Gaussian primitives, demonstrating that the "quality" of primitives is more critical than their "quantity" [32][33].

Core Viewpoint - The article discusses the rise and fall of Haomo Zhixing, a subsidiary of Great Wall Motors, highlighting its initial promise and subsequent failures due to internal conflicts, management issues, and technological missteps [3][6][60]. Group 1: Company Background - Haomo Zhixing was established on November 29, 2019, as part of Great Wall Motors, with its actual control held by Chairman Wei Jianjun, who owns approximately 37% of the shares [8]. - The company aimed to become a global parts supplier, allowing Great Wall Motors to focus on vehicle manufacturing, inspired by the operational models of global automotive giants [8][12]. - Initially, Haomo Zhixing experienced rapid growth, with Great Wall Motors achieving a sales volume of 1.1116 million vehicles in 2020, marking a 4.8% year-on-year increase [9]. Group 2: Strategic Initiatives - In 2020, Great Wall Motors launched its "331" strategy for intelligent driving, aiming to become a leader in the autonomous driving sector within three years [9][12]. - The company planned to achieve L3 autonomous driving capabilities and the first deployment of laser radar in 2021, further enhancing its technological ambitions [12]. - By December 2021, Haomo Zhixing secured nearly 1 billion yuan in Series A funding from various investors, indicating strong initial market confidence [18]. Group 3: Challenges and Failures - Internal management issues became apparent when a whistleblower reported recruitment violations and a lack of collaboration between teams in Beijing, Shanghai, and Baoding, leading to a loss of trust from Great Wall Motors [4][6]. - Haomo Zhixing faced repeated failures to deliver on promises, including the inability to launch its city NOA (Navigation on Autopilot) system as scheduled, which was attributed to internal chaos and poor technical direction [32][46]. - The company's dual leadership structure, with both a chairman and CEO making decisions without a CTO, led to confusion and inefficiencies, ultimately hindering its technological progress [32][36]. Group 4: Market Position and Future Prospects - By the end of 2023, Great Wall Motors began collaborating with other companies like Yuanrong and Zhuoyu for advanced driving projects, effectively sidelining Haomo Zhixing [49][50]. - Despite having opportunities for an IPO in 2023, the process was halted due to external shareholders' concerns over market conditions and valuation expectations [51][52]. - As of late 2024, Haomo Zhixing's operations were severely impacted, leading to a freeze on its bank accounts and uncertainty for its employees, reflecting a significant decline from its initial promise [54][58].

Group 1 - Anta Sports plans to acquire 29.06% stake in PUMA SE for €1.5055 billion (approximately ¥12.278 billion) to enhance its global market position and brand influence [2] - Shandong Agricultural University Fertilizer Technology Co., Ltd. is set to be listed on the Beijing Stock Exchange on January 28, 2026, with an initial public offering of 16 million shares priced at ¥25.00 per share, raising a total of ¥400 million [2] Group 2 - DeepWay has completed a Pre-IPO financing round of ¥1.177 billion, marking the largest single financing in the autonomous driving new energy heavy truck sector [3] - Turing Quantum has secured several hundred million in Series B financing, continuing its trend of raising significant capital within six months [4] - AI medical innovation company Xu Zhi Shi has completed an A+ round of financing exceeding ¥50 million, with plans for a B round to be initiated after the Spring Festival [5] Group 3 - Elon Musk announced that SpaceX's Starship V3 will undergo its first flight test in six weeks, aiming to launch the next generation of Starlink V3 satellites [6] - Ying Shi Innovation is set to launch a dual-camera handheld gimbal camera, with product details leaked prior to the official announcement [7] - Apple has released a new version of AirTag, featuring a wider tracking range and louder speaker, priced at $29 for a single unit and $99 for a four-pack [10]

四年间毫末智行获得7轮融资，估值一度达到百亿元。但由于种种原因，2025年毫末智行突然陷入经营困境，顾维灏也随之离开毫末。 公开资料显示，顾维灏是自动驾驶领域资深技术专家，2022年9月进入清华大学智能产业研究院（AIR）攻读工程博士学位，师从张亚勤教授。 他于2003年加入百度，历任百度MP3和视频搜索主架构师、百度语音识别技术负责人、百度地图副总经理、百度智能汽车事业部总经理等职，主导L3级自 动驾驶技术研发。 2019年他从百度离职，并在2021年2月出任长城汽车旗下毫末智行CEO，推动形成"车企+技术公司"双基因发展模式。其团队研发的城市辅助驾驶系统搭载于 多款长城车型，并与美团、阿里达摩院合作开发末端物流无人车。顾维灏提出"数据智能驱动自动驾驶"理念，构建包含数据治理、模型研发到安全验证的完 整技术闭环，并建立三项核心能力体系架构。 来源：猎云网；文/孙媛 据AI科技评论，前毫末智行首席执行官顾维灏已开启创业征程，将目光投向酒店服务机器人领域，并已组建起一支实力强劲的团队。该团队核心成员多来 自百度和自动驾驶公司Momenta，目前顾维灏正积极与众多投资方接触洽谈。 现如今，他再次创业，在另一条赛道 ...

Group 1 - A 32-year-old programmer's sudden death has raised concerns about the company's health management practices, as the deceased had not undergone the required health check-ups, which the company does not acknowledge [5][6] - The company sent internal emails addressing the situation, stating that the employee had never completed health checks at their facility, despite having annual check-ups at a top-tier hospital [5][6] - Following the incident, the company announced a return to normal operations, canceling celebratory events and emphasizing the need to resume regular business activities [6] Group 2 - Tencent's CEO, Ma Huateng, announced the company's exit from investments in Meituan and JD.com, citing unsustainable cash-burning competition in the food delivery sector [8] - Tencent's investment in Meituan began in 2014, and by the end of 2022, the company had reduced its stake significantly, ultimately exiting completely by early 2026 [8] Group 3 - Li Xiang, CEO of Li Auto, faced criticism from employees after a lengthy all-hands meeting focused primarily on AI trends, with many employees expressing confusion and a desire for more concrete company direction [9] - The company reported a 19% year-over-year decline in vehicle deliveries for 2025, completing only 63.4% of its adjusted sales target [9] Group 4 - DeepSeek released a new open-source model, DeepSeek-OCR 2, which utilizes an innovative method to dynamically rearrange image components based on their meaning, enhancing AI's visual understanding capabilities [11][12] - The model's design allows for more accurate interpretation of complex documents, integrating visual understanding with reasoning and programming abilities [12] Group 5 - NIO has strengthened its intelligent driving team by hiring a former algorithm expert from Alibaba, indicating a commitment to advancing its AI and autonomous driving technologies [24][25] - The company plans to increase its computational resources and aims to regain its position among the top players in the industry through significant updates in the current year [25] Group 6 - Baidu's smart cloud division has raised its AI revenue growth target from 100% to 200%, aiming to capture a larger share of the rapidly growing AI cloud market [40] - The company has been successful in securing a significant number of AI-related projects, leading the market in both project count and total value [40] Group 7 - The automotive industry is witnessing a surge in electric vehicle models, with companies like Volkswagen and Xpeng introducing new models that emphasize advanced features and competitive pricing [42][34] - Xpeng's flying car, expected to be delivered by 2026, showcases innovative design and technology, aiming to capture market interest in the emerging flying vehicle segment [34]

Investment Rating - The industry investment rating is "Hold" [6] Core Insights - The automotive sector experienced a weekly increase of 2.20%, with the best-performing sub-sector being automotive services, which rose by 6.87% [3] - The average daily wholesale volume for domestic passenger car manufacturers was 51,200 units, a year-on-year decrease of 28.00%, while the average daily retail volume was 50,200 units, down 22.00% year-on-year [3] - In January, the narrow passenger car retail market is expected to reach approximately 1.8 million units, with new energy vehicles accounting for around 800,000 units, resulting in a penetration rate of 44.4% [4] - CATL launched a light commercial vehicle solution featuring the industry's first mass-produced sodium battery, which can operate in extreme temperatures [4][11] - Tesla plans to sell the Optimus robot to the public by the end of next year, indicating a significant future business line for the company [5] Summary by Sections Automotive Sector Performance - The automotive sector's weekly performance was +2.20%, ranking 15th among 31 primary industries [3] - The top five companies in terms of stock performance included New Coordinates (+36.30%) and Iron Flow Co. (+32.97%) [3] Sales and Market Trends - The expected retail sales for January are 1.8 million units, with a month-on-month decrease of 20.4% but a year-on-year increase of 0.3% [4] - The penetration rate for new energy vehicles in January is projected at 44.4% [4] Technological Developments - CATL's new sodium battery solution is designed for various vehicle types and can function effectively in extreme cold [4][11] - The first prototype of the Hongqi all-solid-state battery has successfully passed critical testing, marking a significant technological advancement for the company [11] Investment Recommendations - Focus on companies involved in intelligent vehicle technology and those with potential overseas sales [12] - Recommended companies include Beiqi Blue Valley and several component manufacturers [14]