Gasgoo Munich- On March 17, Gasgoo learned that JD.com recently announced plans to leverage its massive supply chain advantages—spanning retail, logistics, healthcare, industrial sectors, food delivery, and home services—to build the world's largest and most comprehensive embodied intelligence data collection center.
Embodied intelligence is rapidly accelerating from the laboratory to industrial application. The latest forecast from IDC projects that the global market for intelligent robotics hardware will approach $30 billion by 2026, with China poised to surpass $11 billion, cementing its role as a core engine of global growth.
On the policy front, "embodied intelligence" has been written into the government work report for a second time—following its inclusion in 2025—marking it as a key area for future industrial development.
Capital is following suit: since the start of 2026, disclosed financing in the sector has totaled nearly 15 billion yuan, and the number of domestic companies valued at over 10 billion yuan has grown to seven.
Yet, beneath this industry boom, a scarcity of real-world data is emerging as the critical bottleneck holding back commercial deployment. Several industry insiders note that the sector still relies heavily on manual teleoperation for data collection. Pieced together, the entire industry has likely accumulated only a few hundred thousand hours of data—a drop in the bucket compared with the tens of millions, or even hundreds of millions, of hours required to achieve intelligent emergence.
Image source: JD.com
JD.com has reportedly established an industry-leading robotics data center, building a full-process pipeline that covers collection, annotation, training, and validation. Spanning five core scenarios—logistics and warehousing, industrial manufacturing, healthcare, home services, and urban operations—the facility captures multidimensional data including visual, tactile, and spatial trajectory information. Most notably, JD plans to mobilize hundreds of thousands of participants. This includes over 100,000 internal employees across various roles and up to 500,000 external workers, covering more than 100 specific scenarios such as homes, offices, factories, logistics hubs, stores, restaurants, medical facilities, and sanitation sites. The company emphasized that all data collection will be conducted in strict compliance with laws and regulations.
The evolution of technical roadmaps underscores the strategic importance of JD's data play. Currently, the focus of capital in embodied intelligence is shifting from hardware "body" stunts to a "brain" revolution. Vision-Language-Action (VLA) multimodal large models have emerged as the key to liberating robots from preset programming and endowing them with generalization capabilities. Yet, training these VLA models depends critically on massive, multi-modal datasets drawn from real-world scenarios.
Under its roadmap, JD aims to accumulate more than 10 million hours of high-quality data within two years. The target includes 5 million hours of real-world human scenario video data within the first year, surpassing 10 million hours by the second year, alongside 1 million hours of robot body data. This initiative is designed to resolve the industry's "data drought" at the source.
This move by JD represents more than just a breakthrough in data scale; it promises to accelerate the co-evolution of the robot "brain" and "cerebellum." By helping embodied models transition from merely "seeing and moving" to truly "understanding the real world," the initiative will provide a critical engine for the healthy and rapid development of the entire industry.
![[Gasgoo Express] Xiaomi's next-generation SU7 to launch on March 19](https://gascloud.gasgoo.com/production/2026/03/7aa7b031-ad2e-4c97-81ba-3cad96724db4-1773780503.png)
