英伟达与发那科宣布合作开发可从语音指令执行任务的实体AI工业机器人，依托英伟达虚拟工厂进行训练，以加速实体部署。系统将内置英伟达计算单元，兼容ROS 2，支持以Python控制动作，并加入避免撞击人类的功能。发那科占全球工业机器人近20%份额，累计出货量超过100万台；当前全球在役工业机器人为466万台，年出货量约50万至55万台。行业预测到2050年工厂与仓储机器人可能增至10亿台，引发制造商战略转向实体AI。

实体AI旨在让机器人实时分析数据并反馈到物理动作，突破传统需预置空间、精细设定与依赖熟练工程师的限制。美中多家初创公司正开发人形机器人，Tesla计划在工厂逐步部署Optimus。软银以53.7亿美元收购ABB机器人业务，强调结合人工超级智能与机器人技术。安川电机计划在美国生产搭载英伟达GPU的AI机器人，用于挖掘隧道与采收蔬菜，并依靠本地化研发追赶美国AI水准。

发那科认为其精密硬件优势与英伟达AI能力结合具关键意义。多家制造商正导入英伟达半导体与服务推进实体AI。发那科管理层强调实体AI只有在高性能AI与高精度机械能力结合后才能实现。

Nvidia and Fanuc announced a partnership to develop physical-AI industrial robots capable of executing tasks from verbal commands, trained within Nvidia’s virtual factory for rapid deployment. Systems will embed Nvidia computers, support ROS 2, allow Python-based motion control, and incorporate human-avoidance functions. Fanuc holds nearly 20% of the global robot market with over 1 million units shipped; 4.66 million robots operate worldwide, with annual shipments of 500,000–550,000. Forecasts project up to 1 billion robots in factories and warehouses by 2050, accelerating strategic shifts toward physical AI.

Physical AI enables robots to analyze real-time data and translate it into physical actions, surpassing constraints of traditional robots requiring preset spaces, detailed configuration, and skilled engineers. U.S. and Chinese startups are developing humanoids, and Tesla plans a phased factory rollout of Optimus. SoftBank purchased ABB’s robotics unit for USD 5.37 billion, aiming to merge artificial superintelligence with robotics. Yaskawa plans U.S.-made AI robots with Nvidia GPUs for tasks such as tunnel digging and vegetable picking, using localized development to close the AI gap with the U.S.

Fanuc argues its hardware precision combined with Nvidia’s AI forms a decisive advantage. Many manufacturers are adopting Nvidia semiconductors and services to advance physical AI. Fanuc leadership states that physical AI becomes viable only when high-performance AI is fused with high-precision mechanical capability.