Chinese embodied AI developer RoboScience has announced a new A2 financing round worth several hundred million yuan, adding fresh capital to its efforts to build general-purpose robot intelligence and the hardware needed to deploy it. The company disclosed the investment on November 2, with Amber Capital, Hongruida Investment and Shanghai Semiconductor Industry Investment among the lead investors. Other participants include 360’s investment arm and several industrial investors.
Founded by Stanford AI Lab alumni Tian Ye and Shao Lin, RoboScience develops a combined software and hardware platform for embodied intelligence. Its products include Visics, a general-purpose robotics foundation model, the RoboMirage physics simulation engine and REX G1, a wheeled humanoid-style robot introduced earlier this year. The new financing will support model development and testing in logistics, retail and industrial environments.
Visics uses what RoboScience calls a Vision-Language-Object-Action architecture. Rather than directly translating visual inputs and instructions into robot movements, the system represents the intended motion of objects in three dimensions before converting those trajectories into commands for different robot bodies. The company is pursuing a model that can transfer manipulation skills between hardware platforms and tasks, although the extent of that generalization in commercial settings remains to be established.
RoboScience is also investing in the data and simulation infrastructure behind its models. According to chief executive Tian Ye, its video training dataset grew from one million examples at the beginning of 2026 to four million by October. The company says its automated data pipeline and simulation tools have reduced the cost of obtaining individual training examples.
The financing follows earlier funding rounds and a series of product releases. RoboMirage was introduced in September 2025, Visics followed in June 2026, and REX G1 was unveiled two months later. RoboScience has also partnered with Tencent Cloud to offer embodied AI capabilities through a cloud service.
The company is positioning itself both as a developer of complete robotic systems and as a supplier of models and controllers for other manufacturers. Its next challenge will be demonstrating that these capabilities can work reliably across different machines and real customer environments, rather than only in controlled evaluations.
