Jensen Huang's 'Work Report' 2026? Yushu's Wang Predicts Humanoids Will 'Think' in Video — and Run Faster Than Bolt

Bold prediction at Yabuli: robots that imagine, then act

At the 26th Yabuli China Entrepreneurs Forum (亚布力中国企业家论坛) in Heilongjiang, it has been reported that Wang Xingxing (王兴兴) of Yushu Robotics (宇树机器人) made a striking forecast: by mid‑2026 global humanoid robots, especially Chinese models, could run faster than ordinary people and potentially even surpass sprint legend Usain Bolt. The claim stunned the audience and set the tone for a wider argument: robots must move from stage demos to industrial self‑sufficiency.

From performance pieces to factory floors

Wang argued that Yushu has already shifted away from purely theatrical showcases and is piloting robots in real factory environments — notably assembling their own joint modules inside Yushu’s plants. He said the ultimate productivity leap will come when robots help produce robots, closing a manufacturing loop that could accelerate scaling. He was candid about limits, however: current systems still struggle with complex, multi‑step tasks and have low execution success rates, leaving significant optimization ahead.

Video‑first world models and the alignment bottleneck

On technical direction, Wang made a clear bet on video‑based world models as the long‑term path: let the robot’s “brain” imagine high‑quality task videos, then align those videos precisely with physical actions to generate executable commands. He pointed to ByteDance (字节跳动)’s Seedance 2.0 video generation as an important enabling technology, but warned that the global industry faces a common hurdle — precise alignment between generated video and actual robot motion. It has been reported that he predicted solving that alignment problem would trigger a foundational breakthrough for generalist robot models — an embodied‑AI “ChatGPT moment.”

Politics, pragmatism and the public response

New Oriental (新东方) founder Yu Minhong (俞敏洪), who spoke immediately after Wang, joked about generational status and stressed a sober point: technology’s end goal is service to people. Outside the hall, observers note that ambitious robotics programs in China are unfolding amid a broader geopolitical backdrop — export controls and trade tensions that affect access to advanced chips and sensors. The technical alignment issue remains a universal challenge. Can robots truly convert rich internal imagination into reliable factory action? For now, the answer is still being built.