推理内容（再强调一下，观点与猜测可能是错误的，事实是经过仔细考证的）： 理想自动驾驶芯片最核心的是数据流架构与软硬件协同设计，相当于一种很适合跑VLA的NPU芯 片。 按谢炎的说法，从项目立案到交付上车，要花三年左右时间，如果是明年上车的话，那应该是2023年 左右开始设计的，猜测理想VLA架构具体如何搭建思路可能主要在2024年下半年。 所有专门为加速神经网络计算而设计的芯片都可以叫NPU，它不像GPU那样追求通用性，而是把硬件 资源集中用于神经网络中最常见的运算，矩阵乘法与加法。NPU会在内部集成大量MAC(乘加单元)， 采用数据流架构，让数据传输路径更短、更搞笑。 本文内核是一篇推理文，推理素材核心基于理想汽车CTO谢炎在25Q2电话会议上的表示，贾鹏在 GTC2025上介绍的理想VLA架构，各种关于TPU/NPU/Thor的介绍。 本文事实部分经过仔细考证，观点部分只能保证经过认真思考，不能保证一定正确。 本文得以成文，离不开一位群友的思路启发与交流。 参考资料： 理想贾鹏英伟达GTC讲VLA 1228字省流版/完整图文/完整视频 为什么Thor芯片要保留GPU，又有NPU？ TPU是谷歌专门为期机器 ...

Core Viewpoint - The article discusses the product launch strategies and future plans of three major electric vehicle manufacturers: NIO, Xpeng, and Li Auto, highlighting their new models, technological advancements, and market competition dynamics [3][5][7]. Company Overview - NIO plans to release a total of 9 models by 2025, with 6 under its brand, including the recently launched ET9, ES6, and EC6 [7]. - Xpeng has 7 models planned for this year, with G6 and G9 already launched, while other models like G7 and new P7 are yet to announce pricing [7][8]. - Li Auto's key models include the i8 and i6, with the i8 set to launch in July [7][8]. Product Planning - NIO's product lineup includes 4 refreshed models and 1 new model, while Xpeng's includes several refreshed versions and new models [5][7]. - Li Auto's L series has undergone significant updates, with a focus on enhancing features and introducing new models [7][12]. Launch Strategies - NIO has opted for online launch events to reduce costs, while Xpeng utilizes social media for pre-launch marketing, extending the promotional period [9][12]. - Li Auto prefers to synchronize product highlights with user experiences during launches, minimizing pre-release information [9][12]. Sales Targets - NIO aims for a sales target of 440,000 units, Xpeng targets 380,000 units, and Li Auto's target is disputed, with claims of 700,000 units being denied [9][12]. Technological Advancements - All three companies are focusing on self-developed technologies, including operating systems and autonomous driving chips [20][24]. - Li Auto introduced its self-developed Star Ring OS, which is open-sourced to encourage community contributions [21][22]. - NIO's NX9031 chip has been integrated into its latest models, showcasing its technological capabilities [14][24]. AI and Autonomous Driving - Li Auto is shifting towards a VLA driver model for intelligent driving, while Xpeng focuses on end-to-end AI solutions [25][26]. - The competition in AI capabilities is intensifying, with each company adopting different strategies to enhance their offerings [25][26]. Market Competition - The competition among NIO, Xpeng, and Li Auto is becoming more pronounced, with NIO and Li Auto facing direct competition, while Xpeng's lower price point provides some buffer [27][28]. - The overall market is expected to become increasingly competitive as more models are introduced [28].