Prominente Unternehmen im Bereich Embodied Intelligence hat eine weitere Finanzierungsrunde in Höhe von mehreren hundert Millionen US-Dollar abgeschlossen, nur einen Monat nach der vorherigen
Exclusive information obtained by Touzhongwang reveals that Wujie Dynamics has completed an angel - round financing of over $200 million. The latest angel - round financing was jointly invested by several well - known institutions such as JD - affiliated funds, C Capital, Hony Capital, and Shengyu Investment. Old shareholders including Linear Capital, Sequoia China, Huaye Tiancheng, and Yare Capital also participated in the follow - on investment. The funds obtained will continue to be used for the research and development of the embodied general brain, the construction of technical infrastructure, and global large - scale delivery. Meanwhile, the Pre - A round of financing of nearly $200 million is also close to completion.
This is also the second large - scale financing for Wujie Dynamics within a month. In the first half of 2026, the company's cumulative financing reached hundreds of millions of dollars. Previously disclosed investors also included Hillhouse Ventures, Baidu Ventures, Envision Group, CICC Porsche Fund, Beijing Artificial Intelligence Industry Fund, and Shenzhen Capital Group.
Wujie Dynamics was founded in 2025, focusing on the "general brain" and "operational intelligence" of embodied intelligence. In terms of the technical route, it chose the "latent space world model + reinforcement learning". At the same time, the robot body K15 has entered the mass - production stage. In terms of commercial implementation, it has quickly secured global orders worth nearly $100 million from clients such as Envision, ZF LIFETEC, and OMV Group, and has started to deliver hardware - software integrated solutions with general generalization capabilities to global industrial and commercial customers.
Looking back at the wave of embodied intelligence startups, if we classify the embodied companies led by professors' startups around 2023 as the first wave, many new players have entered the market in the past two or three years. Among them, autonomous driving is an important faction, and many teams have migrated from companies such as NIO, XPeng, Li Auto, Horizon Robotics, and Huawei to the embodied intelligence battlefield. An investor told me that the emergence of this wave of companies basically ended in the first half of this year. "When the previous wave of people looked at this battlefield, it was basically a fog. For those who entered later, it's all about cognition and execution."
Although the technical path of the current embodied models has not yet converged, it is difficult to judge which route will ultimately win. However, it is certain that funds are accelerating to concentrate on leading enterprises with both leading embodied general brain technology, higher commercial certainty, and implementation capabilities. Top - tier industrial capital, leading US - dollar and RMB capital are jointly betting on Wujie Dynamics, which is also accelerating on this path.
Wujie Dynamics' robots are continuously evolving and iterating in real production lines
The ideal algorithm leader + the top position of Horizon's intelligent driving
The core advantage of a founding team with intelligent driving background entering the field of embodied intelligence lies in having experienced the test of large - scale mass - production and delivery in the real physical world and possessing top - notch algorithm engineering capabilities. The core founding team of Wujie Dynamics includes Zhang Yufeng, the highest - ranking executive in Horizon's foray into the embodied intelligence track, who is the president of the intelligent driving division, and Xia Zhongpu, the leading figure in algorithm technology at Li Auto.
CEO Zhang Yufeng has a complete experience from core technology R & D to large - scale commercial implementation. He once led a team of a thousand people at Horizon to achieve R & D breakthroughs, rapid engineering implementation, and large - scale delivery of intelligent driving software algorithm products, and grew Horizon's intelligent driving business from scratch to the top in the Chinese market share. He also led the mass - production and delivery of domestic intelligent driving solutions in the tens of millions, and through a 2.4 - billion - euro business cooperation with the Volkswagen Group, achieved China's first technology export of intelligent driving to a global top - tier automobile enterprise.
Zhang Yufeng believes that autonomous driving is the first large - scale application of physical AI in human society, and its development path can be mutually referenced with the current embodied intelligence route. This is also the reason for his involvement in embodied intelligence entrepreneurship.
Co - founder and co - CTO Xia Zhongpu is a key figure in the development history of China's intelligent driving industry, having developed and mass - produced end - to - end models. He once set a record of leading a team of less than 40 people to produce the first version of a demo within one and a half months. At Wujie Dynamics, he leads the R & D of the "latent space world model + reinforcement learning" technology and the construction and upgrading of core technical infrastructure.
An investor in Wujie Dynamics commented that the combination of Zhang Yufeng and Xia Zhongpu is the most battle - tested among the entrepreneurs from Li Auto and Horizon, and they have proven themselves in their previous resumes.
The "latent space" faction in the world model
Embodied intelligence has entered the second half, and the brain is considered the core key to true physical AGI. From the early VLA model to the world model, and then to sub - genres such as video - generative and latent space representation, the technical route has gone through an iteration in just about two years.
Before 2025, the popular concept in this industry was the "VLA architecture". However, due to the need to feed a large amount of data for training and limited generalization ability, this route encountered a bottleneck. Wujie Dynamics, founded in 2025, caught up with the trend of the "world model" and took the path of "latent space world model + reinforcement learning".
Combining the "latent space" world model with reinforcement learning, it constructs the embodied general brain MWA™, which is natively multi - modal and fully pre - trained, endowing robots with the ability to deconstruct physical laws and infer causality. Overseas, Yann LeCun is also an important representative of this route.
Compared with the VLA and Li Feifei's 3D spatial intelligence faction, the advantage of this technical path lies in avoiding the information attenuation of cross - representation mapping between language and actions, and eliminating the noise interference and high - computing - power consumption caused by prediction in the pixel space. With the support of reinforcement learning, the large embodied model can continuously optimize its decision - making strategy through autonomous interaction with the real physical world, thus expanding the practical boundaries of robots in complex, highly open and dynamic scenarios.
On this basis, the self - developed robot body provides strong support for the algorithm model. It is understood that Wujie Dynamics' robot hardware platform has undergone multiple iterations and upgrades. The hardware configuration and software architecture of the self - developed body are deeply optimized for the real application environment. In addition, Wujie Dynamics has also self - developed a large - computing - power platform with 1200 TOPS (INT8) for both the "brain" and "cerebellum" in a full - stack manner, and has standardly deployed it on the robots, continuously refining and iterating it in a closed - loop in multiple real scenarios.
Facing open commercial scenarios, Wujie Dynamics creates broader commercial value
$100 million in orders
The self - developed embodied general brain mentioned above relies on breakthroughs in the model architecture and learning paradigm, "understands physics and can do work", and can greatly reduce data dependence. It should be noted that in the second half of embodied intelligence, it's all about "real work". Wujie Dynamics revealed that it quickly secured global orders worth nearly $100 million within just one year of its establishment, and all the orders are based on real and high - difficulty industrial needs.
In the field of industrial manufacturing, Wujie Dynamics' cooperation with global automotive industry chain leaders such as ZF LIFETEC and OMV Group addresses the pain points in real production lines. In the testing process of key automotive - grade components of a certain enterprise, it faced a two - year dilemma in selecting an automation solution. After extensive research and exploration in the industry, the enterprise finally chose to cooperate with Wujie Dynamics and achieved a breakthrough in complex flexible scenarios relying on its model algorithm technology.
In the energy field, the company signed an order worth over 500 million yuan for the international market with Envision, a global leading green technology enterprise, to explore the construction of an overall solution of "super intelligence + super energy". This is also the first overseas order worth over 100 million yuan in the domestic embodied intelligence operation track.
In addition, in commercial service and physical retail scenarios with higher requirements for generalization and interaction, Wujie Dynamics, through cooperation with domestic and international consumer giants such as chain coffee brands, brings robots into high - dynamic human - machine cooperation physical environments, improves their adaptability, and explores the realization of broader commercial value.
This article is from the WeChat official account "Touzhongwang", author: Liu Yanqiu. It is published by 36Kr with authorization.