Aiming at the robotic brain, MouDeep Intelligence secures another 300 million yuan in financing.

```

On May 14, the embodied intelligence technology company Mosheng Intelligence announced the completion of a 300 million RMB Pre-A round of financing.

This round of investors showcased significant diversity, including institutions with state-owned backgrounds such as Lvjihang Capital and Minzhao Fund, publicly listed real economy companies like LEO Group, as well as financial investors such as Huafu Capital, Yingjue Honor, and Gengxin Capital.

Notably, this is the fifth round of financing completed by this company, which was founded in January 2025, within the past half year.

According to publicly disclosed information, this round not only achieved oversubscription, but the subsequent two rounds of financing are already in the closing stage. In the current relatively cautious macro environment of the primary market, a single company maintaining a high-frequency financing pace reflects the capital market’s revaluation of profit margins in the embodied intelligence industry chain—funds are accelerating their shift from heavy-assets “hardware manufacturing” toward high-value-added “brain R&D.”

The support for such high valuations and frequent financing lies in its core team structure, which balances business, algorithms, and computing power adaptation. In the technology-intensive track of embodied intelligence, the team’s genes often determine the company’s technological ceiling and the path to commercialization.

The three core co-founders of Mosheng Intelligence all carry the “Fudan” label and possess distinctly complementary abilities.

Founder Mu Zelin, an experienced entrepreneur, has previously led the founding and acquisition exit of AI companies, endowing the team with early business acumen and pathways for primary market operations.

Co-founder Chen Tao, currently Director of the Deep Learning Laboratory at Fudan University, led the development of Huawei’s first-generation AI image algorithm engine. His longstanding research in AI image algorithms, particularly large model compression and chip adaptation, is key to Mosheng’s solution to edge-side computing bottlenecks.

The other co-founder, Zhang Yimin, formerly Chief Scientist at Intel China Research Institute, brings nearly thirty years of AI research and industrialization experience, filling in the team’s global perspective and underlying software-hardware collaborative architecture.

This combination seeks a practical balance among the industry’s core needs: commercialization, edge computing power limitations, and cutting-edge model architecture.

Stripping away the capital layer’s hype, Mosheng Intelligence has avoided the hyper-competitive humanoid robot hardware manufacturing route, instead asserting its position as a “lightweight embodied brain provider.” Its core business model is “One Brain, Multiple Forms”—building an underlying operating system adaptable to diverse hardware entities such as humanoid, cleaning, logistics, and industrial robots.

Technologically, the company’s system is based on a “World Action Model.” Its early released MotionGPT model has already proven the technical feasibility of coupling vision and motion generation.

Entering 2026, its technical focus has clearly shifted to post-deployment capabilities. Its latest released T²MB embodied large model highlights robots’ continual learning abilities in offline states; disclosed data indicate its offline absolute performance can be increased by up to 25%.

If this performance metric can be stably reproduced in industrial settings, it means robots could break free from dependence on costly cloud computing and high-bandwidth real-time communication, greatly reducing overall edge-side operating costs.

The intense capital competition for Mosheng Intelligence is essentially a bet on the shifting bottleneck within embodied intelligence industries.

In the past two years, the industry's focus has been on reducing costs in hardware bodies such as servo motors. But once the hardware cost curve is clarified, the lack of generalization across scenarios and tasks becomes the core barrier to large-scale application.

The large model Mosheng focuses on is designed to solve the generalization challenge. Investors hope it can, through the diminishing marginal cost of software, secure a central position in the industry ecosystem. Additionally, early commercialization verification has heightened its valuation premium. Recently, Mosheng signed a strategic partnership with Yijia Elder Care to jointly launch 10,000 home care robots capable of human-machine collaboration.

This cash flow expectation from definite B-end orders largely offsets the high risks inherent in underlying large model R&D.

Looking ahead, the commercial potential of Mosheng Intelligence lies in whether it can build a strong data flywheel through its first-mover advantage. Once the number of hardware entities connected to its system reaches a critical point, the emergence of real-world physical data will feed back into the model, forming a technical moat.

However, from the perspective of objective industry development laws, one should still beware of the valley of death between technical demos and large-scale deployment.

On one hand, five rounds of financing in half a year expose the extremely high capital consumption rate of this technological route. The training costs of multimodal embodied large models are exceptionally high, and whether the company can achieve positive cash flow within the capital window remains unknown.

On the other hand, the ten thousand robots within the strategic agreement will face severe tests from a vast array of long-tail scenarios when deployed in complex real-world environments. Mosheng Intelligence now holds the core chips for the embodied brain track, but the next challenge will be extremely tough—large-scale engineering deployment and closed-loop real-world scene data.

Risk Warning and DisclaimerThe market carries risk; investment requires caution. This article does not constitute personal investment advice and does not consider the specific investment goals, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular situation. Invest based on this at your own risk. ```