Mifeng Technology has raised a new round worth several hundred million yuan to expand the data infrastructure it says is needed for embodied AI. The financing is notable not only because of its size, but because China Telecom led it. The involvement of a major state-owned telecommunications group places a data-platform company at the intersection of robotics training, industrial infrastructure, and a broader push to commercialize physical AI.
Cailian Press reported that China Telecom led the investment, with Zhangjiang Group and existing investors also participating. The company said the funding will support its embodied-AI data-platform infrastructure, its MEgo product line for collecting data without a robot body, and capabilities for data collection, governance, and evaluation. The report described the round as worth several hundred million yuan.
The announcement is a reminder that embodied AI requires more than a capable model and a robot chassis. Training systems to work in changing physical environments depends on how data are collected, labeled, managed, tested, and reused. Mifeng is trying to build tools around those steps. Its approach puts the company in a fast-growing segment where the practical bottleneck is often not an individual algorithm, but the supply of useful data that can be converted into learning and evaluation processes.
Mifeng’s Funding Focuses on Embodied AI Data Infrastructure
Mifeng’s stated use of proceeds is specific. It plans to expand data-platform infrastructure for embodied AI, including MEgo products that collect information without a robot body. That focus matters because physical AI projects commonly need data about real-world tasks, objects, movement, and environments. A company that can capture and organize such material may become part of the operating layer that robotics developers depend on before they can train or validate systems.
The Sina Finance report published by China Business Journal also identified China Telecom as the lead investor and Zhangjiang Group as a participant. It described Mifeng’s work around embodied-AI data and the company’s planned expansion of collection, governance, and evaluation capabilities. The overlap between the two reports supports the core facts of the financing, while the company’s intended technology and spending remain management plans rather than independently measured outcomes.
The funding also shows why data work has become a distinct investment category. Robotics builders need hardware, motion systems, and models, but their systems also need records of real tasks. Mifeng’s MEgo line is designed to gather data without attaching the collection process to a robot body. The company says that can support its broader data platform. The reports reviewed for this article do not provide a verified count of data hours, robots, customers, or deployed sites, so those metrics should not be inferred.
China’s physical AI sector already includes companies working on robots, world models, and industrial deployments. EastFrontier recently examined how Guangdong’s embodied-AI industrial clusters are taking shape. Mifeng’s position is different. Rather than presenting a finished robot as its main product, it is pitching a data layer that can support multiple robotics and embodied-AI programs.
China Telecom Adds a Strategic Investor to the Data Race
China Telecom’s role as lead investor is important because telecommunications companies bring infrastructure expertise and large enterprise relationships. The financing announcement does not specify the operational arrangements that may follow from the investment, and it should not be read as proof of a commercial deployment. But the lead role signals that a major telecom group sees value in the data infrastructure Mifeng is developing.
Zhangjiang Group’s participation adds another connection to Shanghai’s technology ecosystem. The two reports do not disclose the precise investment amount from either China Telecom or Zhangjiang Group. They identify the round only at the several-hundred-million-yuan level. That is enough to establish the scale of the financing while leaving the cap table, valuation, and individual commitments unknown.
For an embodied-AI data company, the identity of investors can matter as much as the headline amount. A strategic investor may offer a route into industrial settings, cloud resources, or enterprise customers, but those outcomes are not guaranteed by a funding announcement. Mifeng will still have to demonstrate that its collection and governance tools can supply data that are useful across different physical tasks.
The company is entering a field in which data are increasingly treated as a shared asset rather than a byproduct of a single robot. EastFrontier’s coverage of Jijia Vision’s world-model financing highlighted how investors have backed systems intended to understand physical environments. Mifeng’s data-platform strategy addresses a related, earlier-stage problem: making real-world information available for training and evaluation.
Reusable Data May Decide Which Robotics Systems Scale
The central commercial question is whether Mifeng can make embodied-AI data reusable at sufficient quality. Collecting a video or sensor record is not the same as making it valuable for training. A data platform has to handle capture, organization, governance, and evaluation in a way that developers can integrate into their workflows. Mifeng says it will use the new funding to strengthen all four areas.
The financing is therefore not simply a vote on a single product. It is a wager on the proposition that China’s physical-AI market will need specialized infrastructure companies alongside robot manufacturers. Mifeng’s MEgo products and data-platform plans are designed for that layer of the market. The company has not yet disclosed performance or commercial-scale data to show whether the model works across customers.
For now, the clearest facts are the financing, the investor group, and the stated use of proceeds. China Telecom led a several-hundred-million-yuan round; Zhangjiang Group and existing investors participated; and Mifeng says it will expand embodied-AI data tools spanning collection, governance, and evaluation. As more robotics companies look for ways to train systems on physical tasks, the quality and portability of that data may become as consequential as the hardware that ultimately performs the work.