Maniformer CEO Maoqing Yao Discusses Scaling Physical AI at Fortune Leaders Forum 2026

Macau, China, September 11, 2026 -- On September 8, The Fortune Leaders Forum 2026 brought together business leaders, innovators and industry experts from around the world to discuss emerging technologies and their impact on business and society.

Dr. Yao Maoqing, Co-founder and Co-President of AGIBOT, President of its Embodied Intelligence Business Group, and Chairman and CEO of Maniformer, joined the panel discussion “Intelligence in Motion: Embodied and Autonomous,” where he shared insights into robot deployment, world models, physical AI data and talent.

Photo: FORTUNE

World Models Need to Understand States and Generate Actions

Discussing world models, Dr. Yao said the concept covers a broad range of architectures and applications. For embodied AI, the goal is not simply to generate realistic visual content, but to understand the state of an environment and how it changes over time.

For example, tying shoelaces is not simply about generating a realistic image of the final result. A robot needs to understand the objects involved, the sequence of actions and the changing physical states throughout the task.

Recent advances in video generation have demonstrated increasingly controllable AI-generated environments and actions. For embodied AI, the challenge is to translate these capabilities into physical actions, such as moving a robotic arm, controlling a joint or operating a gripper.

Scaling Data to Drive New Capabilities in Embodied AI

Dr. Yao said embodied AI could similarly benefit from scaling in data and model training. As data volumes and the scale of model training increase, robots may develop stronger zero-shot generalization.

In the next three to five years, with millions and eventually tens of millions of hours of data, robots may be able to complete more everyday tasks “out of the box.”

Against this backdrop, Maniformer is building integrated physical AI data infrastructure around its MEgo series of data-collection hardware and software. The platform combines real-world, simulation and human-demonstration data to help address data bottlenecks in embodied AI development.

Photo: FORTUNE

Real-world Deployment Requires Sustained Validation

Dr. Yao noted that robot deployment requires long-term testing in real environments.

At a factory in Jiangxi Province, China, where AGIBOT G2 robots were deployed, the robots underwent continuous stress testing during nighttime and non-production hours. The team evaluated success rates, stability, continuous operation and suitability for commercial deployment. After one month of intensive testing, the robots achieved an operational success rate of 99.99%.

Dr. Yao also emphasized that robotics companies should start with real user needs, rather than “bringing an answer to look for a problem.”

Attracting Talent Through a Shared Vision

Discussing talent, Dr. Yao said the company is looking for talent that shares its long-term vision for embodied AI.

He believes that embodied AI will become an important application of multimodal AI, allowing robots to understand and act in the physical world. Talented people want to realize their own value while contributing to companies that create lasting social value.

Looking ahead, Maniformer will continue to bring together data providers, AI developers, robotics companies, research institutions and end users to help bring robots from the laboratory into factories, service environments and everyday settings.

Contact Info:
Name: Michelle Xu
Email: Send Email
Organization: Maniformer
Website: https://maniformer.ai

Release ID: 89203058

In the event of any inaccuracies, problems, or queries arising from the content shared in this press release, we encourage you to notify us immediately at [email protected] (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). Our diligent team will be readily available to respond and take swift action within 8 hours to rectify any identified issues or assist with removal requests. Ensuring the provision of high-quality and precise information is paramount to us.