How far is embodied AI from the real technological singularity? At the 2026 Yabuli Forum Annual Meeting held on March 17, Autonomous Intelligent Technology founder Wang Xingxing gave a calm prediction. He believes that the true ChatGPT moment for embodied AI is coming soon, but it still needs some refinement. It is estimated conservatively that it will take two to three more years.

In Wang Xingxing's definition, the "ChatGPT moment" of embodied AI is seen as a key performance threshold: that is, one day in the future, an AI model of a robot will be able to independently complete 80% of tasks through only verbal and written instructions in about 80% of unfamiliar scenarios.

Although this prediction seems slightly pessimistic, Wang Xingxing also added that the process of technological breakthroughs could be accelerated beyond expectations due to certain key algorithmic breakthroughs. Looking at industry dynamics over the past year, he has repeatedly tracked the turning point of embodied intelligence before:

September 2025: He once predicted that general humanoid robots could achieve "autonomous work" as early as late 2026.

End of 2025: He believed the turning point of embodied intelligence might occur within the next one or two years.

January 2026: He even explicitly stated that those who can create large-scale robot models are worthy of winning the Nobel Prize.

Currently, robot large models are still in the early stages. Although the average growth rate of domestic intelligent robot companies is impressive, they still face challenges in the deep integration of algorithms and hardware before truly entering households as "living partners." When a robot's brain can understand the complex physical world as well as ChatGPT understands text, that will be the starting point of the true breakthrough of embodied intelligence.