“物理AI”被用来描述自动化系统在现实世界中感知、理解、推理并执行复杂动作的能力，在汽车行业中被视为一项关键转向。该概念在2026年CES上被集中推广，反映出汽车、机器人与芯片产业的加速融合。对芯片公司而言，这一趋势对应一个到2032年规模可达1230亿美元的市场，相比2023年增长约85%，使汽车成为继消费电子之后最具吸引力的计算平台之一。

在实际应用层面，多家车企和科技公司公布了明确的时间表和量化目标。福特表示，计划在2028年前销售允许驾驶员无需注视前方道路的系统。索尼与本田合作的纯电品牌Afeela已发布量产车型与原型车，目标是在多数场景下实现自动驾驶。英伟达将为吉利的“智能驾驶系统”提供芯片，并参与梅赛德斯-奔驰在美国于当年推出的免手持驾驶系统，后者最终目标是实现通勤路径的全程自动驾驶。

这些功能对车载算力提出了数量级提升的要求。咨询公司AlixPartners指出，未来车辆的“中央大脑”规模将是当前的数百倍。正因如此，英伟达和ARM等芯片商成为“物理AI”叙事中最积极的推动者：ARM在CES设立专门的物理AI部门，英伟达推出面向自动化系统的开源模型。营销术语的背后，是算力需求和产业价值的同步放大。

“Physical AI” is being used to describe autonomous systems that perceive, understand, reason, and act in the real world, and it signals a strategic shift in the automotive industry. The term was heavily promoted at CES 2026, reflecting the rapid convergence of cars, robots, and chips. For chipmakers, this shift represents a market projected to reach $123 billion by 2032, about 85% growth from 2023, positioning vehicles as one of the most valuable future computing platforms.

In practice, companies are attaching timelines and scale to these claims. Ford says it plans to sell a system by 2028 that allows drivers to operate vehicles without watching the road. Sony and Honda’s electric Afeela models aim to handle most driving situations autonomously, though without a fixed launch date. Nvidia will supply chips for Geely’s intelligent driving system and is involved in Mercedes-Benz’s US launch of a hands-off driving feature, with the long-term goal of full home-to-work automation.

These ambitions require orders-of-magnitude increases in onboard computing power. Consultancy AlixPartners estimates that the vehicle’s “central brain” will be hundreds of times larger than today’s systems. This explains why chipmakers are the loudest promoters of physical AI: ARM launched a dedicated Physical AI division at CES, and Nvidia introduced open-source models for autonomous machines. Behind the marketing, rising computational demand and economic scale are tightly aligned.