en.Wedoany.com Reported - Recently, China's world model company Manifold AI completed a new round of financing worth several hundred million yuan. The new investors in this round include China Reform Fund under China Reform Holdings Corporation, InnoVen Capital under Temasek, as well as industrial capital BAIC Capital and Xinneng Venture Capital, among others. Meanwhile, four existing shareholders made additional oversubscribed investments. To date, Manifold AI has completed six rounds of financing since its establishment about a year ago, with the total amount of the Pre-A round reaching nearly 1 billion yuan.
This round of financing will primarily be used for the development of a new generation of technical frameworks, multimodal world models, and the construction of related infrastructure platforms. The company will also promote the application of embodied intelligence in fields such as e-commerce logistics, 3C manufacturing, and automotive manufacturing, and explore MR mixed reality pan-entertainment scenarios. The use of funds indicates that Manifold AI's current focus is not only on model development but also on extending to real-world scenarios such as robotics, industrial manufacturing, and spatial interaction.
Founded at the end of May 2025, Manifold AI focuses on world models and embodied intelligence. The company has previously completed multiple rounds of financing, with historical investors including Junlian Capital, Cowin Capital, Inno Fund, and Huawei Hubble, among other venture capital institutions and industrial capital. Unlike artificial intelligence models that solely generate text, images, or videos, world models emphasize the ability to understand physical space, object states, action outcomes, and environmental changes, making them a crucial underlying capability for embodied intelligent systems to enter the real world.
In terms of technical approach, Manifold AI conducts research and development around multimodal world models, embodied intelligence, and infrastructure platforms. World models need to enable intelligent agents to understand "what they see," "what they can do," and "what happens after an action," which is vital for autonomous decision-making in robots performing tasks such as warehousing, assembly, handling, inspection, and interaction. If a model can predict real-world changes in a virtual environment, it can reduce the reliance of robot training on large amounts of real-world data, improving generalization capabilities and deployment efficiency.
The composition of investors in this round also reflects the industrialization characteristics of embodied intelligence. The simultaneous participation of state-owned funds, international capital, and industrial capital indicates that this direction has moved from a laboratory concept to a stage closer to industrial application. Industrial capital such as BAIC Capital and Xinneng Venture Capital focuses on applications related to automotive manufacturing, chip computing power, and smart terminals, which may provide Manifold AI with more industrial resources for subsequent collaboration in manufacturing scenarios, robotic systems, and computing power platforms.
The difficulty of implementing embodied intelligence still lies in real-world scenarios. Although e-commerce logistics, 3C manufacturing, and automotive manufacturing have high-frequency operations, standardized processes, and automation upgrade needs, the on-site environment is complex. Robots need to handle issues such as occlusion, grasping errors, object variations, equipment coordination, and safety boundaries. For world models to enter these industries, they must pass verification of stability, real-time performance, low-cost deployment, and compatibility with existing production line systems.
MR mixed reality pan-entertainment represents another application direction. Multimodal world models can be used to generate interactive spaces, virtual character behaviors, and dynamic scene feedback, enabling a more natural integration of virtual content with real-world spaces. Compared to industrial scenarios, pan-entertainment applications emphasize immersion, real-time interaction, and content generation efficiency, but they also require models to possess spatial consistency and action continuity.
For Manifold AI, the nearly 1 billion yuan Pre-A round provides a financial foundation for subsequent research and scenario validation. Going forward, the company needs to transform world model capabilities into deliverable products and establish stable customer cases in embodied intelligence and MR applications. Completing the financing is just the starting point; subsequent progress in model effectiveness, engineering capabilities, scenario replication speed, and commercial revenue will determine whether it can evolve from a high-valuation startup into a true industrial platform.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
