WeRide Spins Off Data Business, Subsidiary JingSuo Bets on Embodied Intelligence
0xBroomberg
WeRide is spinning off a business unit for the first time — wholly owned subsidiary JingSho will sell data services for embodied AI, targeting RMB 300 million in 2026 revenue. This means WeRide is repackaging its autonomous-driving data capabilities into a standalone business aimed at robotics companies.
Who is JingSho, and why spin it off now?
JingSho was founded in May 2024 and operated as a support function inside WeRide's autonomous-driving R&D system.
Now independent, it covers two lines: an AI data platform (raw data through model deployment) and embodied-intelligence data services (data synthesis, teleoperation capture, data processing).
This means → WeRide is turning its internal "data factory" from an R&D cost center into a revenue-generating business that sells to outside clients.
Where is the money, and how fast is it growing?
WeRide's intelligent-data-services revenue was roughly RMB 55.8 million in 2024 and an estimated RMB 159.6 million in 2025. JingSho's 2026 target is RMB 300 million — close to tripling in two years.
JingSho recently closed a Series A round. WeRide sees strong demand internally, and investors are reportedly confident in the outlook.
In plain terms = this segment used to be a small line buried in the earnings report. Spinning it out and raising outside capital signals that management believes it can stand on its own.
What is JingSho actually selling?
The core product is not the volume of data samples — it is the capability to turn real-world data into training-ready material.
Once real data enters the system, it can generate further synthetic data. Data collection is only one pathway; the real-world data JingSho has accumulated also extends beyond automotive scenarios.
This means → for robotics companies, the value hinges on whether JingSho can lower data-production costs and funnel data from different sources into a single training pipeline.
Can autonomous-driving data be used directly for robots?
Not directly. Autonomous driving and embodied intelligence differ in sensors, action spaces, and task objectives — robots need large volumes of grasping, locomotion, manipulation, and human-interaction data.
What can genuinely transfer across scenarios is the process layer: data governance, scenario mining, simulation generation, model evaluation, and engineering delivery — not the vehicle data itself.
In plain terms = JingSho is not selling "dashcam footage." It is selling the factory capability to convert any real-world scenario into machine-learnable training material.
Where are the competition and risks?
The market is already moving: Rui Qi Mobility launched its own embodied-intelligence data platform in June 2026, similarly extending Robotaxi data capabilities into robotics.
A March 2026 Frost & Sullivan report noted that the physical-AI simulation and data-platform market is still in early-stage growth, with platform value contingent on downstream applications' technical maturity.
Nomura flagged in a July report that data-as-a-service can monetize quickly by the hour or by project, but suppliers lacking model-evaluation and application capabilities risk being vertically integrated away by robotics firms.
This reflects a key validation point: whether JingSho can bundle real data, synthetic data, scenario libraries, model evaluation, and delivery workflows into a recurring-purchase product — rather than a project-fee staffing service — will determine if the economics work.
Content is for reference only, not financial advice.