ACE Robotics, a Chinese robotics start-up company, predicts that smart people may be at a turn in a industry similar to ChatGPT by 2027. Reuter quoted Wang Xiaogang, Chairman of the company, as saying that the key to this change lies in the progress of the world ' s models and in the improved capacity to collect real environmental data.

The smart goal is to enable robots not only to show action, but also to perform their tasks in a complex and changing reality. Many man-made robots are now able to walk, dance or even fight, but there is still a clear gap from taking on useful work on a large scale.

Training data remain a major bottleneck

Wang Xiao has just indicated that the industry has accumulated data for the past few years for approximately 100,000 hours, which is far from sufficient to train a general base model. The reliance of robotics training on action, environment and feedback data in the real world is more dependent on big text than on language models, which also limits data acquisition.

ACE Robotics plans to collect tens of millions of hours of data over the next two years to close this gap. According to the company ' s judgement, robotic models are only likely to increase their generalization capacity once environmental perception, action decision-making and physical feedback data continue to accumulate.

Company pushes down and finance expansion

ACE Robotics was founded in July 2025 to develop AI models for human robots. According to reports, the company received support from ant groups and merchant soup and completed over $100 million in financing in the first half of 2026.

  • Founded: July 2025
  • Financing in the first half of the year: over $100 million
  • Target for deployment in the next year: 1,000 stores

The company is also planning to deploy the technology to 1,000 stores in the coming year as a source of training and commercialization in the real scene. According to Reuters, ACE Robotics is also prepared to move forward with the first public collection unit when conditions permit.

The robot industry accelerated the model competition.

Competition around the robot AI model is warming. The report mentions that in January of this year, Boston power released a quantitative version of Atlas human robots and said that AI progress brought business deployment closer. In June, Ali Baba launched Qwen-Robot Suite to help robots navigate, perform physical tasks and simulate the real environment.

At the same time, researchers are experimenting with new methods of data collection. Last October, a team released HumanoidExo to train human robots by recording human movements.

In terms of industry, robotics are still far from being widely used, but the integration of model capabilities with real-world data is becoming the central direction of the next phase of competition.