别认为小鹏还只是个造车的

Core Viewpoint - Xiaopeng Motors is transitioning from being merely a car manufacturer to creating a "super intelligent entity" with its second-generation VLA (Vision-Language-Action) model, which aims to revolutionize autonomous driving technology and enhance user experience [2][4][10]. Group 1: Technological Advancements - The second-generation VLA model features a disruptive architectural redesign that eliminates the "language translation" step, allowing for intuitive responses akin to human driving [2][5]. - The model integrates visual, auditory, and reading capabilities, enabling it to function as a foundational model for various applications, including Robotaxi and humanoid robots [5][9]. - Xiaopeng's approach to leap directly from L2 to L4 autonomous driving aims to clarify responsibility and streamline the development process, avoiding the complexities associated with L3 [7][8]. Group 2: Market Positioning and Strategy - Major investment banks, including Morgan Stanley and Bank of America, have issued buy ratings for Xiaopeng Motors, highlighting its potential to compete with Tesla in the global market due to advancements in autonomous driving technology [2][3]. - The company aims to double its overseas sales by 2026, with a target of selling 1 million vehicles abroad by 2030, contributing significantly to its profits [11][13]. - Xiaopeng's Robotaxi initiative is set to begin public road testing, with plans for operational deployment in Guangzhou by the end of 2026, marking a significant step in expanding its market beyond traditional passenger vehicles [14]. Group 3: Research and Development - The second-generation VLA has been trained on 50PB of data, with each model version utilizing 40 trillion tokens, showcasing Xiaopeng's commitment to data-driven development [9]. - The establishment of a "General Intelligence Center" consolidates the autonomous driving and smart cockpit divisions, emphasizing the integration of driving decisions and human-vehicle interaction [11]. - The company is investing heavily in the foundational model, which is seen as crucial for the future of various embodied intelligent systems, indicating a long-term vision for technological leadership [11][16].