Investment Rating - The report rates the transportation industry as "Outperform" [2] Core Insights - The disruption of navigation in the Hormuz Strait has led VLCCs to reroute to the Yanbu Red Sea passage, with West African routes compensating for the export gap [3][12] - The escalation of the Middle East situation has caused tight air cargo capacity between Asia and Europe, with Cathay Pacific canceling flights to Dubai and Riyadh until March 31 and increasing capacity to Europe [3][16] - NVIDIA announced an expansion of its collaboration with Uber and Lyft, launching the Robotaxi plan in multiple U.S. cities starting in 2027, which has positively impacted related stocks [3][25] - WoFei ChangKong held a supply chain conference in Chengdu, unveiling a 10 billion opportunity list and receiving a 10 billion yuan credit support from ICBC [3][27] - The first "road-air integration" automotive test site in China has commenced operations, marking a significant step in low-altitude vehicle testing infrastructure [3][28] Industry Dynamics Shipping and Logistics - The Baltic Air Freight Price Index increased by 2.6% month-on-month but decreased by 0.7% year-on-year [30] - The container shipping price index (SCFI) rose by 29.38% year-on-year, while dry bulk freight rates increased by 25.75% year-on-year [41] - In February 2026, the express delivery volume decreased by 10.90% year-on-year, while revenue remained relatively stable with a slight decrease of 0.01% [53] Investment Recommendations - Focus on opportunities in oil transportation, dry bulk shipping, and container shipping sectors due to the evolving Middle East situation, recommending companies like China Merchants Energy and COSCO Shipping [4][15] - Attention to coal transportation-related stocks such as Daqin Railway and Jiayou International [4] - Investment opportunities in high-speed rail and highways, recommending companies like Beijing-Shanghai High-Speed Railway [4] - Emphasize low-altitude economy and autonomous driving trends, recommending companies like CITIC Offshore Helicopter [4] - Monitor international market expansion opportunities in express logistics, recommending SF Holding and Jitu Express [4]

Report Industry Investment Rating No information about the report industry investment rating is provided in the given content. Core Viewpoints - Last week, the stock market continued to adjust, with major indices oscillating downward and daily trading volume shrinking to the 2 - 2.4 trillion range. Geopolitical conflicts and weakened Fed rate - cut expectations suppressed risk appetite. The bond market showed a weak and oscillating pattern, with the 10 - year Treasury yield rising. The convertible bond market mostly declined, with the CSI Convertible Bond Index down 3.15% for the week [1][6][7]. - Amid the escalating US - Iran conflict and rising oil prices, market risk - aversion persisted. The Shanghai Composite Index fell below 4000 points, and convertible bonds adjusted with the underlying stocks. The valuation of debt - biased convertible bonds was significantly compressed, while the premium rate of convertible bonds with a par value above 120 yuan rebounded. In the short term, convertible bonds present individual - bond opportunities. It is recommended to focus on sectors such as semiconductors, computing power leasing, energy storage, and defensive sectors [2][17]. Summary by Relevant Catalogs Market Trends (2026/3/16 - 2026/3/20) Stock Market - The overall market continued to adjust, with major indices oscillating downward and daily trading volume shrinking to 2 - 2.4 trillion. Geopolitical conflicts and weakened Fed rate - cut expectations suppressed risk appetite. The technology sector fluctuated more, and upstream resource products showed divergent trends. Defensive sectors were relatively resilient [1][6]. - By industry, most Shenwan primary industries declined. Communications (2.10%), banks (0.36%), and food and beverages (-0.48%) performed relatively well, while non - ferrous metals (-11.82%), basic chemicals (-10.53%), and steel (-10.29%) performed poorly [7]. Bond Market - Despite balanced and loose liquidity, bond yields generally rose due to factors such as improved economic data, inflation pushed up by geopolitical conflicts, and liquidity concerns. The bond market showed a weak and oscillating pattern, and the 10 - year Treasury yield closed at 1.83% on Friday, up 1.56bp from the previous week [1][7]. Convertible Bond Market - Most convertible bond issues declined. The CSI Convertible Bond Index fell 3.15% for the week, the median price dropped 3.31%, and the arithmetic average par value fell 5.54%. The overall conversion premium rate increased by 2.71% compared with the previous week. The arithmetic average conversion premium rates of convertible bonds in the par value ranges of [90,100), [100,110), and [110,120) changed by -3.51%, -4.74%, and -2.68% respectively, and were at the 95%, 91%, and 78% quantiles since 2023 [7]. - By industry, convertible bonds in most industries declined. Banks (-0.72%), commerce and retail (-1.21%), and beauty care (-1.71%) performed relatively well, while petroleum and petrochemicals (-10.05%), social services (-9.69%), and non - ferrous metals (-7.23%) performed poorly [10]. - In terms of individual bonds, Yubang (photovoltaic welding tape), Hongbai (organosilicon), Jinhong (helium), Haiyou (photovoltaic film), and Songlin (smart home) convertible bonds led the gains; Zhongchong Zhuan 2 (pet food), Jize (wind power), Baichuan Zhuan 2 (fine chemicals), Hengyi (chemical fiber), and Hongqiang (concrete admixture) convertible bonds led the losses [1][11]. - The total trading volume of the convertible bond market last week was 316.723 billion yuan, with an average daily trading volume of 6.3345 billion yuan, a decrease from the previous week [15]. Valuation - As of March 20, 2026, for equity - biased convertible bonds, the average conversion premium rates of convertible bonds in the par value ranges of 80 - 90 yuan, 90 - 100 yuan, 100 - 110 yuan, 110 - 120 yuan, 120 - 130 yuan, and above 130 yuan were 49.44%, 41.36%, 28.57%, 19.24%, 16.43%, and 14.53% respectively, at the 97%/96%, 97%/97%, 94%/93%, 87%/81%, 92%/93%, and 97%/95% quantiles since 2010/2021 [18]. - For debt - biased convertible bonds, the average YTM of convertible bonds with a par value below 70 yuan was -2.95%, at the 4%/11% quantiles since 2010/2021 [18]. - The average implied volatility of all convertible bonds was 46.08%, at the 92%/96% quantiles since 2010/2021. The difference between the implied volatility of convertible bonds and the long - term actual volatility of the underlying stocks was 4.55%, at the 88%/88% quantiles since 2010/2021 [18]. Primary Market Tracking Newly Announced Issuances and Listings - Last week (2026/3/16 - 2026/3/20), Shang 26 and Boshi convertible bonds announced issuances, and Tonglian convertible bonds were listed [25]. - As of the announcement on March 20, there were no announcements of convertible bond issuances or listings for the next week (2026/3/23 - 2026/3/27) [29]. Issuance Progress - Last week, Shenergy Co., Ltd. was approved for registration by the exchange. Shengde Xintai, Baotai Co., Ltd., Tonghe Technology, and Huaxiang Co., Ltd. passed the review of the listing committee. Runbei Hangke and Xianfeng Jinke passed the general meeting of shareholders. Shentong Express had a board of directors' plan. There were no newly accepted enterprises by the exchange [29]. - As of now, there are 100 convertible bonds to be issued, with a total scale of 167.22 billion yuan. Among them, 4 have been approved for registration, with a total scale of 6.43 billion yuan; 12 have passed the listing committee, with a total scale of 11.89 billion yuan [29].

Core Viewpoint - The article discusses Elon Musk's announcement regarding the "TERAFAB" project, which aims to produce over 1 terawatt of computing power annually, with 80% allocated for space applications and 20% for terrestrial uses [2][10]. Group 1: Project Overview - The TERAFAB project will establish a vertically integrated semiconductor manufacturing facility in the U.S., covering logic chips, memory chips, and advanced packaging [3]. - The project is expected to require an investment of approximately $25 billion, surpassing existing super factory scales [4]. Group 2: Chip Development Roadmap - Tesla's chip evolution includes the AI5, AI6, and D3, targeting cutting-edge 2nm process technology with a production goal of 100 to 200 billion custom AI chips annually [4]. - The AI5 chip is designed for Full Self-Driving (FSD) and Optimus, with performance improvements of 40-50 times over the previous generation [7][8]. Group 3: Strategic Goals - The distribution of computing power from TERAFAB reflects ambitious goals, with a focus on creating a distributed interstellar computing network through AI satellites [12]. - Musk emphasizes the need for internal chip production to mitigate geopolitical risks and supply chain vulnerabilities, as existing suppliers like TSMC and Samsung cannot meet Tesla's demands [16][21]. Group 4: Challenges and Considerations - Building the TERAFAB facility presents significant challenges, including high capital costs and the complexity of advanced chip manufacturing processes [19][21]. - The timeline and location for the factory remain uncertain, with Tesla planning to utilize its substantial cash reserves for funding [18].

Core Viewpoint - The U.S. National Highway Traffic Safety Administration (NHTSA) has intensified its investigation into Tesla's Full Self-Driving (FSD) system due to potential safety defects that may arise in low visibility conditions, affecting approximately 3.2 million vehicles manufactured since 2016 [1][2]. Investigation Details - NHTSA is examining whether the FSD system is safe to use in conditions such as fog, heavy rain, and glare, which could impair camera visibility [1][3]. - The investigation includes an "Engineering Analysis" to assess the effectiveness of the degradation detection system, which is designed to alert drivers when visibility is compromised [5][7]. - Previous incidents have shown that the degradation detection system failed to recognize conditions that impaired camera visibility, leading to accidents, including a fatality [6]. Potential Outcomes - If NHTSA finds that the degradation detection system is inadequate, it could lead to a recall, which may require Tesla to implement a software update to address the issues [9][10]. - There is a possibility that NHTSA may deem the camera-only system unsafe for self-driving applications, potentially requiring Tesla to make significant changes to the system or its marketing [10][11]. - A recall or significant changes to the FSD system could have substantial financial implications for Tesla and impact its reputation, especially in light of recent incidents involving its robotaxi fleet [11].

Core Viewpoint - The competition in China's autonomous driving is shifting from merely adding functionalities to focusing on system architecture and capability organization, as the industry approaches a limit in functional advancements [3][4][26]. Group 1: Shift in Competition Focus - The current competition is moving towards how well systems can organize and utilize their capabilities rather than just stacking functionalities [3][4]. - As the complexity of driving scenarios increases, the ability to recognize, judge, and act continuously becomes crucial, indicating that system architecture will determine future performance limits [3][11][26]. Group 2: Ideal's Strategic Response - For Ideal, the transition to architecture competition is critical, requiring the company to demonstrate its capability in defining the next generation of autonomous driving systems, not just in creating user-friendly vehicles [6][30]. - The introduction of MindVLA-o1 represents Ideal's answer to the architectural challenge, aiming to unify understanding, judgment, and action within a single "driving brain" framework [7][14]. Group 3: Technical Innovations and Challenges - MindVLA-o1 aims to address several key challenges: understanding the physical world, predicting changes in dynamic environments, and converting judgments into actions [15][19][20]. - The architecture integrates visual, language, and trajectory models into a unified framework, enhancing the system's ability to understand three-dimensional relationships and respond to real-time changes [14][18][21]. Group 4: Future Directions and Industry Implications - The future of autonomous driving is seen as a starting point for broader physical AI applications, with MindVLA-o1 serving as a foundational model for a unified intelligent system [30][31]. - The competition will evolve from functional capabilities to model capabilities, with a focus on who can continuously train and iterate on the "driving brain" [32][35].

Domestic News - The Ministry of Industry and Information Technology, the National Development and Reform Commission, and the State Administration for Market Regulation are accelerating efforts to address shortcomings in automotive chips and basic software, aiming to enhance application scale and quality performance [7] - The State Administration for Market Regulation has released the revised "Automotive Recall User Guide (2026 Edition)", which includes a new section for interpreting recall announcements to assist vehicle owners [8] - NIO's self-developed chips have exceeded 550,000 units in mass production, addressing challenges in the automotive semiconductor industry [9] - Xiaomi plans to invest over 60 billion yuan in AI over the next three years [10] - FAW has announced a humanoid robot patent that enhances the robot's adaptability to internal disturbances [11] - Lantu Automotive has officially listed on the Hong Kong Stock Exchange, becoming the first high-end new energy vehicle stock from a central state-owned enterprise [12] - WeRide has entered a strategic partnership with ELEVATE Slovakia to launch Slovakia's first autonomous driving project [13] - CATL has opened 30 battery swap stations in Hangzhou, with plans to expand to nearly 70 stations by the end of the year [14] International News - Infineon has announced an expansion of its collaboration with NVIDIA to advance physical AI system architecture, focusing on humanoid robotics [15] - Stellantis electric vehicles can now access Tesla's Supercharger network in North America, marking a significant milestone for EV brands [16] - Uber has partnered with Rivian, investing up to $1.25 billion to develop a fully autonomous taxi fleet based on Rivian's R2 SUV platform [17] - The Delhi government in India plans to introduce a subsidy policy for scrapping old vehicles in exchange for electric vehicles to accelerate the transition to electric mobility [18] Commercial Vehicles - In 2025, China's commercial vehicle exports surpassed 1 million units for the first time, with a significant increase of 86.8% in new energy vehicle exports [21] - Foton's Thailand factory has begun exporting light trucks to Indonesia, marking a significant step in its localized production strategy [22] - Iveco's natural gas heavy-duty truck has won an international award, highlighting the competitive edge of natural gas technology in the commercial vehicle sector [23] - Weichai has launched its new series of small trucks in Chengdu, aiming to become a leading brand in the new energy commercial vehicle market [24]

Core Insights - The event on March 20 showcased the first approved large-scale comprehensive testing ground in China for "road-air collaboration" [1] - Didi Autonomous Driving and GAC Aion presented a new generation of Robotaxi models, which have undergone successful durability tests in autonomous driving mode [2] - The Robotaxi integrates Didi's L4 full-stack software and hardware technology with GAC Aion's intelligent manufacturing capabilities, designed specifically for driverless scenarios [2] Group 1 - The new Robotaxi model has been tested in complex conditions, demonstrating stable performance and autonomous driving capabilities [2] - The vehicle features a new hardware platform with 33 sensors, including LiDAR and cameras, enabling comprehensive environmental perception [2] - The central computing unit of the vehicle boasts over 2000 TOPS of GPU computing power, with improved performance and reduced costs by 74% compared to the previous generation [2][3] Group 2 - The vehicle meets the stringent safety standards of both China and Europe, incorporating a multi-layered safety redundancy system [3] - The development process adheres to functional safety standards, ensuring rigorous testing across various conditions to guarantee reliable operation in urban environments [3] - Didi Autonomous Driving emphasizes responsible technological innovation and aims to contribute to the high-quality development of the autonomous driving industry [3]

Core Viewpoint - The article discusses the emergence of "world models" in AI, highlighting the significant advancements made by the Chinese startup InSpatio, which has developed a groundbreaking model called InSpatio-World that surpasses existing models in efficiency and performance [1][2][30]. Group 1: World Model Development - The global tech industry is heavily investing in world models, which are essential for AI to understand and reconstruct the physical world [1]. - InSpatio has released InSpatio-World, which outperforms other real-time interactive models in the authoritative WorldScore-Dynamic ranking, demonstrating superior performance with a significantly lower training cost of only a few million yuan [2][18]. - The model achieves a real-time inference speed of 24 FPS on a single professional-grade graphics card, showcasing a major leap in efficiency compared to traditional 2D video paths [2][10]. Group 2: Technical Innovations - InSpatio-World utilizes a 3D spatial architecture, moving away from the inefficient 2D video approach, thus enabling real-time interaction and dynamic world modeling [2][13]. - The model's architecture allows it to achieve high performance with a minimal parameter size, proving that innovative modeling paradigms are more crucial than sheer scale [19][26]. - InSpatio's approach includes state-anchored world modeling, which allows for a more accurate representation of the physical world by decoupling observation perspectives from physical entities [20][23]. Group 3: Industry Impact - The advancements in world models are expected to redefine multiple industries, including autonomous driving, virtual reality, and embodied intelligence, by enabling AI to observe, remember, predict, and plan in a manner closer to real-world interactions [30][31]. - InSpatio aims to create a self-evolving digital world that empowers intelligent agents to transition from mere interaction to proactive world modification [31][32]. - The company has received significant interest from global leaders in robotics, autonomous driving, and content creation, indicating a strong demand for its world model technology [32].

大家好我是大鱼 Rivian突然 丢出一记重磅 联手Uber 砸下最高 12.5%亿美元 目标部署高达 5万台全自动 R2 Robotaxi！ 他们直接放弃 2027年获利预期 把所有资源 all in自驾 豪赌Level 4 无人监督的未来 同一时间 Waymo已累积 1.7%亿英里纯 rider-only 里程 安全数据 碾压人类驾驶 Robotaxi 赛道瞬间 杀声四起 Tesla不再 独占鳌头 这场战争 才真正开打 大家不要错过 今天的精彩内容 OK let's go 特斯拉的股票 周四收盘是380.30% 全天下跌了 12.48%美元 跌幅是3.18% 成交量是 6654万股 第一部分 Tesla Robotaxi 最大杀手出现！ Rivian近期 刚把R2推出来 大家还在讨论 它要怎么跟 Tesla Model Y 硬碰硬 结果一转眼 他们直接宣布 跟Uber合作 要搞高达 50000台 全自动Robotaxi！ 更狠的是 同一天他们还说 2027年调整后 核心获利 不指望了 全砸进R&D加速 自驾路线图 我必须告诉你 一个最残酷的事实 这就是Rivian 现在的选择 他们正把未来 全押在这一搏上 ...

Core Viewpoint - The Hong Kong stock market is experiencing a rebound, particularly in the automotive sector, with significant inflows into the Hong Kong Stock Connect Automotive ETF, indicating a positive sentiment towards automotive stocks [1][3]. Group 1: Market Performance - As of March 20, the Hong Kong Stock Connect Automotive ETF (159210) rose by 2.6%, potentially ending a three-day decline, with over 11 million yuan in inflows over the past three days [1]. - Major component stocks of the ETF saw substantial gains, including Horizon Robotics up over 6%, Geely Automobile up over 5%, and CATL up over 3% [3]. Group 2: Key Company Highlights - Geely Automobile reported a total revenue of 345.23 billion yuan for 2025, a 25% year-on-year increase, marking a historical high, with a net profit of 16.85 billion yuan, a slight increase of 0.2% year-on-year [5]. - Geely's cumulative deliveries for January and February reached 476,300 units, a slight increase of 1% year-on-year, surpassing BYD to become the sales champion, with 241,700 units of new energy vehicles sold, a 10.1% increase year-on-year [5]. Group 3: Industry Trends - The automotive industry is undergoing dual transformations centered around electrification and AI integration, with expectations for significant advancements in smart driving and vehicle technology [11]. - The demand for automobiles is expected to remain resilient despite policy changes, with a notable recovery in mid-to-high-end demand driven by shifts in consumer habits and product competitiveness [6]. - The global expansion of Chinese smart electric vehicles is anticipated, with projections of nearly 10 million units in overseas sales within five years [7]. Group 4: Investment Opportunities - The Hong Kong Stock Connect Automotive ETF tracks the China Securities Hong Kong Stock Connect Automotive Industry Theme Index, covering the entire automotive supply chain, including electric vehicles and related technologies [11]. - The ETF's top holdings include leading companies such as BYD, Xpeng Motors, and Li Auto, focusing on opportunities in smart driving and energy storage sectors [11].