leaderobot.com Publishes Article on 106 Embodied Intelligence Data Collection Centers

On November 9, 2026, leaderobot.com published an article about 106 data collection centers for embodied intelligence.

Regarding Apptronik's Apollo, we have reported that Google DeepMind published a method for teaching physical tasks through video observation. When assessing the scale of data collection sites, it is necessary to distinguish not only the number of facilities but also which stages of training use data collected from physical robots.

The publication of an article about data collection sites for embodied intelligence shows that systems for gathering data in real-world environments are a topic of discussion, separate from robot manufacturing capacity and training methods. However, this information comes from secondary reporting, and no primary source was provided to independently substantiate the article's figures or assessments. The number of sites therefore cannot be taken as a direct measure of operational scale or the volume of useful training data. In a related article, we reported that Google DeepMind published a method for teaching physical tasks to Apptronik's Apollo through video observation Google DeepMind Publishes Method for Teaching Apptronik's Apollo New Skills From Video. This is an example suggesting that training data may come from sources beyond recordings of physical robots in operation. When assessing data collection sites, it is therefore necessary to look not only at the number of facilities but also at what they record and which stages of training use the data. On the manufacturing side, we reported that Agility Robotics began trading on Nasdaq after a business combination and presented a manufacturing facility with annual capacity of up to 10,000 units Agility Robotics Completes Business Combination and Begins Trading on Nasdaq. An article about Jabil also discusses the transition from prototyping to design and manufacturing for mass production Jabil Article Discusses Progress in Humanoid Development and Mass Production. Taken together, these examples show that robot supply, data collection and use in training are connected stages, and that it is important not to assess their scale using the same metric. Evidence of manufacturing capacity does not automatically verify actual operations or data quality, and the number of data collection sites alone cannot establish progress in mass production or training. Points to verify include how facilities are counted and how operations are defined; the uses and quality of the collected data; robot models and working conditions; and how the data is used to improve models or physical robots. These cannot be confirmed from the information provided. Refer to the original article for details of its counting method and claims.