Core Insights - The intelligent automotive industry in China has entered the AI Driving 2.0 phase, with accelerated technological iteration [1] - The VLA (Vision-Language-Action) technology architecture, represented by companies like Li Auto and Yuanrong Qihang, breaks the limitations of traditional end-to-end models by integrating three modalities to enhance system interactivity and long-term reasoning capabilities [1] - The industry is transitioning from rule-driven to data-driven approaches, although there are still concerns regarding technological pathways, investment returns, and policy implementation risks [1] Industry Developments - Huawei's ADS 4.0 adopts the WEWA architecture, emphasizing cloud simulation and vehicle model collaboration to create a direct cognitive model of the physical world, enabling rapid response and high safety redundancy [1] - Horizon Robotics and Momenta focus on one-stage end-to-end systems combined with reinforcement learning to enhance the human-like experience of intelligent driving systems [1] Investment Opportunities - The Intelligent Automotive ETF (159889) tracks the CS Intelligent Automotive Index (930721), which selects representative listed companies in the intelligent driving and vehicle networking sectors from the A-share market [1] - The index aims to reflect the overall performance of securities related to intelligent automotive companies, showcasing the industry's diversity and broad characteristics [1]

Investment Rating - The report maintains a "Strong Buy" rating for the industry [1] Core Insights - The evolution of intelligent driving has entered the AI 2.0 phase, focusing on scalable capabilities and the ability to autonomously handle extreme edge scenarios, which will further enhance the commercial viability of intelligent driving systems [1] - Major players in the high-level intelligent driving sector are accelerating their entry into the Robotaxi business, leveraging mass-produced vehicles to optimize model training and performance in extreme scenarios [1][79] - The report highlights the importance of diverse real-world data and robust R&D resources as key competitive advantages for players in the AI driving space [81] Summary by Sections 1. Tesla's Software and Hardware Iterations - Tesla's FSD (Full Self-Driving) software has achieved significant milestones, with over 60 billion miles driven cumulatively, showcasing its leading position in intelligent driving [7] - The next-generation AI5 chip is expected to greatly enhance the performance and energy efficiency of Tesla's driving systems [8][9] 2. Development Stages of High-Level Intelligent Driving in China - The industry has transitioned from a rule-based system to a fully data-driven approach, marking the arrival of the AI driving era [15][12] - The current phase emphasizes end-to-end models that utilize extensive data for improved driving performance and user experience [18][19] 3. Technical Architecture: Mainstream Player Directions - The VLA (Vision-Language-Action) model integrates visual, language, and action modalities, enhancing the system's ability to understand and interact with the physical world [27][28] - Huawei's ADS 4.0 emphasizes a scene-driven approach, utilizing cloud simulations to train AI drivers without relying on large language models [49][50] 4. Business Model: Acceleration of Robotaxi Initiatives - The Robotaxi business is seen as a critical avenue for data collection and model optimization, with major players planning to leverage mass-produced vehicles for this purpose [65][66] - The report outlines two main technological routes for Robotaxi operations: the "crossing route" represented by Waymo and the "gradual route" represented by Tesla, each with its own advantages and challenges [67][68] 5. Investment Recommendations - The report recommends investing in companies such as Seres, Horizon Robotics, Great Wall Motors, Li Auto, and Xpeng Motors, which are well-positioned to capitalize on the advancements in AI driving technology [81]