Core Viewpoint - The article discusses the advancements and potential of Horizon Robotics' HSD (Highway Smart Driving) system, emphasizing its transition from testing to near mass production, and its implications for the smart automotive industry in China [1][4][36]. Group 1: Technological Advancements - Horizon Robotics has launched its self-developed computing hardware, the J6P, and the UniAD software, marking significant milestones in autonomous driving technology [2][4]. - The HSD system features a one-stage end-to-end architecture, which improves the driving experience by integrating various functionalities without relying on traditional rule-based systems [19][36]. - Key innovations include dense modal information processing, joint optimization of lateral and longitudinal control, and a robust safety verification module to ensure safe vehicle operation [20][21][24]. Group 2: Performance and User Experience - The latest version of HSD demonstrates improved speed control and smoother driving experiences, particularly in complex scenarios like traffic lights and merging [6][10]. - The system's ability to navigate through obstacles and make real-time decisions showcases its advanced cognitive capabilities, allowing it to handle various driving conditions effectively [11][36]. - Despite its advancements, some issues were noted, such as misidentifying stationary vehicles, indicating areas for further refinement [13][36]. Group 3: Industry Implications - Horizon Robotics aims to establish a comprehensive solution for the automotive industry, moving away from isolated implementations for specific manufacturers [41]. - The company's strategy focuses on a gradual approach to autonomous driving, with a clear timeline for achieving hands-off and eyes-off driving capabilities [37][39]. - The competitive landscape suggests that companies failing to keep pace with these advancements may risk obsolescence in the rapidly evolving smart automotive sector [37][39].

Core Insights - The automotive industry is experiencing a shift from traditional E/E architecture to a centralized architecture, enabling companies to take control of software development and redefine vehicle core functionalities [1] - The trend towards self-developed chips is driven by the desire to capture profits from the high-margin ends of the "smile curve" (chips and algorithms) [1] - The "smart driving equality" movement initiated by BYD presents both opportunities for automakers and a historic chance for smart driving suppliers [2] Industry Dynamics - In the context of intensified competition and the saturation of smart driving solutions, automakers prioritize getting their products to market rather than focusing solely on self-development [2][3] - The window for automakers to capitalize on this trend is limited to approximately two years [3] Competitive Landscape - NVIDIA currently dominates the smart driving market with a 45.4% share, primarily due to its Orin-X chip, which is highly regarded for its AI acceleration capabilities [4] - New entrants in the market, such as Horizon and Black Sesame, have made significant advancements in chip capabilities, allowing them to compete with NVIDIA's offerings [5][8] - The validation timeline for domestic chips is only 2-3 years behind NVIDIA, indicating a narrowing gap in competition [6][8] Chip Development and Cost Analysis - The development of chips is a long-term endeavor, with significant time required for accumulation and iteration [10] - Companies need to achieve a minimum production scale and possess strong iteration capabilities to make self-developed chips economically viable [11] - Cost analysis shows that the manufacturing costs of chips vary significantly based on technology nodes, with the total cost for advanced chips reaching up to 3915 yuan [12] Data and Talent Acquisition - Automakers have a substantial advantage in data acquisition for training smart driving systems, with companies like Tesla and Huawei leading in vehicle deployment [15] - The scale of talent and internal collaboration is crucial for the development of advanced smart driving technologies, with major players employing large teams dedicated to this field [16] Conclusion - BYD, while relying on external suppliers for chips and algorithms, is actively working on self-developed smart driving solutions, indicating a trend among leading automakers to balance self-development with external sourcing [17]

另一方面则是自家系统的硬件降本，归功于两个关键零部件的国产化。 今年年初，围绕智驾渗透率有机构给了一个乐观预测： 高速和城区NOA未来每年都将增长5%，到2030年能达到55%和25%。 一个是激光雷达，一路从爱马仕跌成了拼多多，要感谢国产厂商的奋起直追；另一个是芯片，为了应对 地平线征程6M，英伟达推出了Orin Y的降本平替方案，售价比Orin X还低100美金。 但与中国品牌占据全球92%份额的激光雷达市场不同，在智驾芯片市场国产化才刚刚冒头。而在这块市 场，也上演着一出从入门到高端阶梯追赶的替代戏码。 01 由小见大 黑芝麻智能CEO单记章不久前也公开唱多，称今年底乘用车NOA渗透率将达到20%。 而涨势喜人的渗透率，则要感谢拼命内卷的车企们，把智驾系统的价格打下来了。 一方面是靠规模化采购，头部车企能将整套智驾成本控制在4000元以下，有券商调研纪要显示，比亚迪 天神之眼C的智驾域控制器成本大概为3000元，加上传感器后整体硬件成本不到5000元； 按算力大小智驾芯片可分为三类：低于30TOPS的小算力芯片、30到150 TOPS的中算力芯片以及150 TOPS以上的大算力芯片。 在智驾芯片国产化的 ...