英伟达业务版图再拓展。 黄仁勋曾多次公开强调，自动驾驶汽车不仅是"机器人技术的第一个主要商业应用"，更是一个"数万亿 美元级别的产业"。 36氪从多处获悉，英伟达内部正在孵化Robotaxi项目，这项决定在近期的一项All hands meeting（全员 大会）上宣布，将交由就职多年的高级总监Ruchi Bhargava负责。 知情人士告诉36氪，新项目将采用全新的一段式技术路线。该技术路线仅使用一个"端到端"神经网络， 核心是通过仿真技术形成的世界模型对神经网络进行强化训练，与特斯拉FSD采用的路线相似。 英伟达今年1月发布了Cosmos世界基础模型。该平台通过整合文本、图像、视频及传感器数据，生成遵 循物理规律的高质量合成视频数据，并已经经过2000万小时数据的预训练。 Cosmos世界基础模型的意义之一，便是能够拓展现实场景难以产生的复杂数据，来提升自动驾驶系统 的能力上限。 这条路线已经得到行业的初步认可，理想、小鹏等企业均已着手打造自己的世界模型。 消息人士向36氪表示，英伟达发力Robotaxi，其逻辑并非简单的业务扩张，而是想推出一个"Robotaxi 的技术样本"。 此前，英伟达已与通用、奔 ...

Core Insights - Nvidia is incubating a new Robotaxi project led by senior director Ruchi Bhargava, utilizing an end-to-end technology approach that leverages simulation to enhance neural network training, similar to Tesla's Full Self-Driving (FSD) strategy but with a stronger technical foundation [4][5] - The project is supported by Nvidia's Cosmos world model, which integrates various data types to generate high-quality synthetic video data, having completed 20 million hours of pre-training [4] - Nvidia plans to invest $3 billion in the Robotaxi project, contrasting with Waymo's cumulative investment of approximately $12 billion to achieve its current operational scale [5] Company Strategy - The aim of Nvidia's Robotaxi initiative is not merely business expansion but to validate its full-chain engineering capabilities from GPU chips to physical AI models, thereby defining the infrastructure and ecological standards for the next generation of "physical AI" [5] - The US Robotaxi market is expected to accelerate by 2025, with Waymo operating in seven cities and Tesla launching services in Austin and the Bay Area, achieving significant download numbers [5][6] Competitive Landscape - Despite facing challenges in autonomous driving software, Nvidia possesses significant advantages, including its self-developed DRIVE Thor chip with a computing power of 2000 TOPS, enhancing end-to-end model inference efficiency [6] - Industry experts note that the Robotaxi market is still in its early stages, with Waymo operating around 700 vehicles and Tesla deploying only a few dozen in Austin, indicating that Nvidia's entry is timely and offers opportunities to compete for technological leadership [6] Financial Outlook - Nvidia's net profit for Q2 2025 is projected to reach $26.4 billion, providing substantial funding for long-term research and development [6] - Jensen Huang, Nvidia's CEO, emphasizes that autonomous vehicles represent a major commercial application of robotics and a multi-trillion-dollar industry [6]