BridgeDP Robotics, a Shenzhen developer of a general-purpose robot operating system, has closed a Pre-A+++ funding round. The round was led by China Mobile’s chain-leader fund, with returning investors Fosun Capital and Shenzhen Capital Group adding capital, at a scale reaching the 100-million-yuan level.
The company said the fresh money goes to four areas: building the ecosystem around its RoboCraft AI general motion-capability platform, scaling its cross-body whole-body motion data factory, iterating its motion-control models, and laying out an overseas technical service and compliance system. BridgeDP will keep extending from “motion control” into manipulation, world models and other key fields, accelerating a global general-purpose robot operating system.
The quiet force behind the demos
In a crowded embodied-AI field, BridgeDP is an unusual animal. It does not build robots. Yet many of the flashy moves you see from various machines may run on motion software from this roughly hundred-person startup, making it a low-profile but essential enabler.
Founded in May 2023, BridgeDP set out to build a general robot OS. Early on it entered through the robot “cerebellum”, supplying motion-control solutions for quadrupeds, wheeled-legged and humanoid bodies. Founder Shang Yangxing took his bachelor’s at Huazhong University of Science and Technology, then a master’s at Southern University of Science and Technology in Shenzhen under LimX Dynamics founder Zhang Wei, working on perception, planning and control of legged robots before leaving to start up.
The firm first served body makers on a project basis: clients brought hardware, BridgeDP sent engineers to model, train, debug and deploy. That validated demand and brought early revenue and profit, but every new body meant relearning the tuning. The insight was blunt: project work is a services business, not a scalable product.
From projects to a platform
RoboCraft AI is the answer, a general motion-development platform best described as an “action mall” for robots. Traditional teams re-solve “how to stand, how to walk, how to deploy, how to tune” for every new machine. RoboCraft AI turns expert-only motion development into a platform-assisted software flow, starting with the Motion layer and banking reusable building blocks: walking, running, turning, obstacle crossing, performing.
BridgeDP sets itself apart by stressing third-party, cross-body use: the same action or skill should migrate across brands and form factors. Tasks that once needed five reinforcement-learning engineers and three months, it claims, can shrink to one software engineer in one week, provided the hardware behaves.
The longer roadmap runs from “action mall” to “skill mall” to “application mall”: parametric atomic skills like box carrying, door opening, sorting and stair climbing, then full task solutions for factory inspection, warehouse handling, mall guiding and showroom reception. Underneath sits a three-layer split: Motion wraps hardware difference, Skill offers composable capabilities, Task dispatches skills per job.
Why stay independent
Asked about in-house motion teams at robot makers, Shang is unruffled. His structural edge is neutrality: by not building hardware, BridgeDP becomes the independent third party that serves a cross-platform developer base. With five or six surviving body makers, no developer will learn a separate toolchain for each, just as no one builds a separate Android for every phone brand. A unified platform is the job of an independent third party.
The moat, he argues, is not a web page but hardened experience: cross-body deployment, Sim-to-Real loops, real-failure data, motion models, a unified abstraction interface and a safety-deployment system. BridgeDP is also building a joint-module database to fight widespread spec inflation in the actuator market, planning to publish verified performance data on its platform.
On money, RoboCraft AI charges subscriptions, billing per deployed model per robot. This year’s focus is the platform product, not revenue scale; the firm expects revenue in the tens-of-millions-yuan range, up from last year, with profitability already proven on project work.
Editor’s note: This is an adapted translation of the original zhidx report. It has been trimmed and restructured for readability for an international business audience.