公众号记得加星标⭐️，第一时间看推送不会错过。 来源：内容 编译自 notebookcheck 。 特斯拉在准备下一代AI5 FSD计算机以集成到车辆的同时，透露了一些AI6芯片的细节。AI5/HW5计算机将搭载一款内部 开发的AI芯片，该芯片专为特斯拉自动驾驶汽车的需求而量身定制，功能强大，不适合出口。 特斯拉将不得不限制其为下一代使用3nm工艺打造的HW5 计算机开发的"惊人"人工智能芯片的功率，该芯片将于 2026 年 底投入量产。这个发布时间已推迟整整一年，原定于"2025 年底/2026 年第一季度初发布。 特斯拉现在称之为AI5的硬件 5.0 FSD 计算机在人工智能计算方面将非常强大，它显然会与美国政府出于国家安全考虑对 人工智能芯片实施的出口管制相冲突。 埃隆·马斯克表示，特斯拉在将计算能力与自动驾驶汽车软件融合方面，拥有最佳的AI芯片设计。正因如此，该公司再次 内部开发HW5计算机，尽管其最新的HW4计算机也是几年前开发的， 但"目前还没有一款芯片是我们愿意用在汽车上 的"。 特斯拉 Model Y自动驾驶出租车目前搭载的 AI4 芯片 ，其无人驾驶全自动驾驶（FSD）功能显然比即将推出的 ...

Core Insights - The automotive industry in China is entering the second half of intelligent transformation, with advanced driver-assistance systems (ADAS) becoming a key battleground [1] - The previous approach of maximizing features with minimal hardware has revealed limitations, leading to performance bottlenecks and user trust erosion [2][3] - A shift is needed from a focus on cost-effectiveness to prioritizing safety and system performance as the foundation for user experience [4][5] Group 1: Safety and Performance - Safety is now a prerequisite for ADAS, not just an added benefit, marking a new phase in the industry [4][5] - The introduction of national standards for ADAS safety requirements indicates a regulatory shift towards compliance rather than innovation [7][8] - High-performance computing and redundancy in models and decision-making are becoming industry consensus for ensuring safety and reliability [9][10][12] Group 2: System Integration and Cost - Achieving scalable deployment of ADAS requires a balance of safety, performance, and cost [15][16] - The industry is moving towards a "triangle" framework that integrates safety systems, high-performance support, and cost efficiency [16][17][18] - The need for high bandwidth and efficient data transmission is critical, as merely increasing computing power is insufficient to overcome system bottlenecks [23][24] Group 3: Advanced Models and Real-World Application - The integration of large models is essential for enhancing the system's ability to understand complex driving scenarios [24][25][28] - The evolution of ADAS is transitioning from a focus on functionality to a focus on safety, where the ability to handle extreme scenarios is paramount [29][30] - The ultimate goal is to create a trustworthy and continuously evolving ADAS platform that prioritizes user safety and experience [28][33]