Unitree's Wang Xingxing: Robot-AI Alignment Is Key to Embodied Intelligence's ChatGPT Moment

Speaking at the 2026 World Physical AI Conference and Unitree Partner Ecosystem Conference in Hangzhou on October 22, Wang Xingxing, Unitree Robotics' chairman, general manager and chief technology officer, outlined the gap between rapid progress in large models and the slower path to real-world embodied intelligence.

Wang said leading large-model companies have in some sense already achieved internal model self-evolution: models write their own programs and collect data, while humans only provide limited guidance. Yet the public impact feels less dramatic because a technology breakthrough typically takes five to ten years to spread to everyday users.

In physical AI, Wang argued, early signs of a breakthrough are now visible. The central hurdle for embodied intelligence to reach its own ChatGPT moment is not simply model capability, but aligning AI model inputs and outputs with real robots—ensuring what a model understands and generates matches how physical machines sense, act and are constrained in the real world.

The remarks position robot-AI alignment as a systems challenge spanning data, models and hardware. Unitree used the partner event to frame that challenge as central to the next stage of embodied AI, even as the timeline from technical advances to mass adoption remains measured in years rather than months.