中国车企正在智能驾驶领域争逐。"十五五"规划提出我国要坚持智能化、融合化方向，加快建设制造强 国、交通强国。这将推动汽车产业向智能制造、绿色制造转型，巩固制造业支柱地位。 "智能驾驶正迎来它的'GPT时刻'。"清华大学计算机系教授邓志东向新京报贝壳财经记者表示，当前行 业处于白热化竞争状态，城区领航辅助驾驶（NOA）已成为顶级玩家的"必争之地"，从一线城市向更 多区域覆盖的速度，成为检验车企技术实力的试金石。 不仅如此，智能化打破了传统汽车产业的边界，新的玩家、新的合作模式、新的利润点正在涌现。 AI能力成车企估值新锚点 "十五五"规划提出我国要推动技术改造升级，促进制造业数智化转型，发展智能制造。在北京社科院副 研究员王鹏看来，这将促进AI与制造深度融合，催生"软件定义汽车"、服务型制造等新业态。他认为， 如今汽车的价值核心正从机械性能转向智能化体验。 他举例表示，从资本市场上来看，AI能力正成为估值的关键点，与华为深度合作的赛力斯的市值曾经 在短期内飙升，"如果车企具备智能驾驶能力，投资者对它潜在的软件收入、未来可能的生态服务收入 和技术壁垒带来的长期定价权都会有更大想象力。"王鹏表示。 不仅如此，"十五 ...

Core Insights - The Chinese automotive industry is in a competitive race towards intelligent driving, with the "14th Five-Year Plan" emphasizing the need for smart and integrated development, accelerating the transition to intelligent and green manufacturing [1][2] - AI capabilities are becoming a new valuation anchor for automotive companies, shifting the focus from mechanical performance to intelligent experiences, with significant implications for investment and market perception [2][4] - The industry is witnessing a shift from hardware competition to software competition, with companies increasingly focusing on self-developed chips and partnerships with major chip manufacturers [3][4] Group 1 - The "14th Five-Year Plan" aims to promote technological upgrades and the digital transformation of manufacturing, leading to the emergence of new business models such as "software-defined vehicles" [2][4] - The competition in intelligent driving is characterized by a "computing power arms race," with companies needing substantial AI computing resources to develop and train intelligent driving models [2][3] - The automotive industry is evolving into a comprehensive sector that integrates advanced technologies from AI, energy, chips, and digital services, reflecting a deep interconnection with various industries [7][8] Group 2 - The relationship between automotive and chip industries is becoming increasingly symbiotic, with new business models and value growth points emerging from this integration [8] - Companies are exploring the potential of utilizing excess computing power from vehicles as distributed computing nodes within the broader digital economy [8] - The competitive landscape is marked by different collaboration models, with companies like Huawei providing full-stack solutions, while others focus on self-research to maintain control over core technologies [7][8]

Core Insights - The smart driving industry is experiencing a transformative moment akin to the "GPT moment," driven by the maturity and commercialization of "world model agents" technology [1] - The current technological phase is marked by the successful mass production and commercialization of systems like Tesla's FSD V13.2 and Huawei's ADS 4.0 [1] - The challenge of data collection for autonomous driving safety can be addressed through "digital twin" technology, which generates vast amounts of synthetic data [1] Group 1 - The concept of "world model agents" is identified as the future direction of smart driving, moving beyond the traditional "end-to-end" approach [1] - The safety of autonomous driving systems must exceed that of human drivers, requiring AI to accumulate significantly more driving experience [1] - Companies providing high-quality simulation platforms and data services will hold greater value in the future automotive industry [1] Group 2 - A competitive "computing power arms race" is underway, occurring simultaneously in cloud and vehicle environments [2] - In the cloud, constructing world models from vast amounts of real and synthetic data necessitates substantial resources, including hundreds of thousands of AI accelerator cards and EFLOPS-level computing power [2] - On the vehicle side, the demand for computing power in smart chips is increasing from 500-600 TOPS to over 2500 TOPS, highlighting the need for innovation in chip design and system integration [2]