- A SenseTime co-founder said the current bottleneck for the humanoid robot industry is low data volume, making it difficult to meet the demands of complex physical world interactions.
- SenseTime-backed Ace Robotics plans to boost its data volume to 10 million hours by 2027.
The humanoid robotics sector will need about two more years of technological breakthroughs to reach a "ChatGPT moment" similar to that of large language models, according to Wang Xiaogang, co-founder of Chinese artificial intelligence company SenseTime.
Wang made the assessment on Wednesday during a humanoid robotics panel at the Boao Forum. He also serves as the chairman of Ace Robotics, a SenseTime-affiliated company.
The current industry bottleneck lies in data volume, which remains at the 100,000-hour level and is insufficient for the complex interactive demands of the physical world, he said.
Ace Robotics plans to increase this data volume to 10 million hours by 2027, combining world models and simulation technology to achieve highly efficient AI training, Wang added.
Ace Robotics completed its angel round of financing early last month, led by Ant Group with participation from several prominent investment firms.
The newly injected capital will primarily be used to advance a "human-centric" ACE embodied AI research and development paradigm, aiming to solve the "data drought" restricting the industry's growth, the company said.
Traditional machine-centric R&D paradigms are facing severe challenges due to a lack of generalizability, according to the firm.
In mid-December 2025, Ace Robotics launched the industry's first ACE embodied R&D paradigm to help robots understand and adapt to physical laws.
The company's A1 super brain module has already been adapted for various types of robot dogs, and it will continue to work with upstream and downstream industry partners to build an independent and controllable embodied AI application ecosystem, Ace Robotics said last month.
