The 2026 World Physical AI Conference and Unitree Partner Ecosystem Conference was held in Hangzhou today. According to ChainCatcher, Unitree Robotics Chairman, General Manager and Chief Technology Officer Wang Xingxing said in his keynote that the core challenge to embodied AI reaching its “ChatGPT moment” is aligning AI models' inputs and outputs with real robots. Wang said that leading AI model companies have, in a sense, largely achieved internal model self-evolution: models can write programs and collect data, while people provide guidance and participate as needed. He added that it typically takes five to 10 years for technological breakthroughs to become widely accessible, which may explain why their impact is not yet deeply felt by the public. Physical AI, he said, is already showing signs of a breakthrough.
