Humanoid Robot Brains Could Reach a Turning Point
Humanoid robot brains could experience a major breakthrough by the end of 2027, according to ACE Robotics chairman Wang Xiaogang.
Wang said embodied artificial intelligence could reach its own “ChatGPT moment” by late next year.
He expects world models and environmental data to drive the breakthrough.
However, he believes widespread commercial use could take several more years.
“Even if we reach that inflection point by late 2027,” Wang said, commercial adoption could still take another four to five years.
AI Robots Still Face Major Challenges
ChatGPT and other large language models have quickly become part of everyday life.
Robots, however, face a much harder challenge.
They must understand physical spaces and respond to changing conditions.
They also need to complete complex tasks without constant human guidance.
That makes real-world training data extremely important.
World Models Could Transform Robotics
Embodied AI models give robots the ability to understand their surroundings.
These systems combine perception, physical simulation and action planning.
They can help robots understand what is happening around them.
They can also predict future actions over longer periods.
ACE’s open-source Kairos-4B model uses these capabilities.
The company says the model currently ranks highly on public benchmarks.
Chinese Robotics Industry Seeks Real-World Results
China’s humanoid robotics sector has attracted significant investor interest.
Companies are now looking beyond demonstrations involving dancing or athletic movements.
Investors increasingly want evidence that robots can perform useful commercial tasks.
That shift could accelerate the development of practical humanoid robots.
Unitree founder Wang Xingxing recently predicted a major breakthrough in robot brains within two or three years.
He and other robotics executives have also highlighted the shortage of high-quality training data.
ACE Robotics Raises More Than $100 Million
ACE Robotics launched in July 2025.
Ant Group and SenseTime back the Chinese startup.
The company raised more than $100 million during the first half of this year.
Wang said ACE hopes to launch an initial public offering as soon as regulations allow.
Chinese listing rules generally require companies to operate for at least three fiscal years.
Training Data Remains a Major Bottleneck
ACE believes the robotics industry does not yet have enough real-world data.
Wang estimated that companies have collected around 100,000 hours of relevant data over several years.
He said that amount remains far too small for training large embodied AI models.
ACE is now expanding its data collection efforts.
The company uses workers equipped with lightweight sensors on real production lines.
Wang expects the approach to generate tens of millions of hours of data within two years.
Robots Learn Through Physical Training
Many robotics companies use physical tele-operation to train humanoid systems.
Workers wear special equipment and controllers.
They then repeat physical movements hundreds of times.
The process generates data that helps robots learn how to perform tasks.
ACE is also pursuing this approach while expanding its real-world training operations.
ACE Plans Thousands of Robot Deployments
ACE is already testing its AI systems in commercial environments.
These include unmanned retail stores, hotels and instant-delivery warehouses.
The company works with humanoid robots from Chinese manufacturers such as Unitree, AgiBot and Fourier.
Wang said ACE plans to deploy its technology in at least 1,000 stores over the next year.
He expects that number to reach 10,000 stores within two years.
Nvidia and Chinese Chips Power ACE Models
ACE uses artificial intelligence chips from Nvidia for its models.
The company also works with Chinese chipmakers, including Rhino Tech and Digua Robotics.
Wang said this approach helps ACE diversify its supply chain.
He expects demand for intelligent hardware to increase as robotics technology develops.
At the same time, he said hardware costs must fall significantly before humanoid robots can achieve broad commercial adoption.
Commercial Robotics Could Take Years
The predicted humanoid robot brains breakthrough could arrive by late 2027.
However, Wang does not expect immediate mass adoption.
He estimates that broad commercial implementation could take another four to five years after the breakthrough.
That timeline reflects the difficult transition from laboratory demonstrations to reliable robots working in everyday environments.
If ACE and other companies solve the data and hardware challenges, humanoid robots could eventually become a much more common part of commercial operations.
