Chinese scientific simulation and AI data startup SciVisual has completed Angel and Angel-Plus funding rounds totaling CN¥ tens of millions, according to a report from ChinaVenture. Investors include Shanghai Future Industry Fund, Pudong Venture Capital, and Houxue Capital, which continued its support from the Seed round. The Hefei-based company converts complex scientific equipment into structured digital assets that AI systems can interpret and use.
From Research Visualization to AI Data Infrastructure
Founded by University of Science and Technology of China researcher Yang Yuanfan, SciVisual began with efforts to make advanced scientific achievements easier to understand. Its technology combines 3D models, component relationships, material properties, and physical parameters so that AI tools can process scientific and engineering data. The company has since evolved from scientific visualization into what it calls a scientific 3D AI infrastructure business.
Proving the Model in Complex Facilities
SciVisual has worked on projects involving fusion reactors, particle accelerators, and other major research facilities, including the EAST tokamak and the HALF advanced light source in Hefei. According to the company, it has served more than 60 academicians and over 800 scientists across more than 20 national large scientific facilities and 300 frontier research projects. These environments provide a demanding training ground for accurate modeling, data standardization, and lightweight rendering.
Solving the Data and Cognition Gap
Yang Yuanfan explains that AI can only enter scientific research when it first understands real scientific objects. That requires addressing three separate layers: scattered data in CAD files, simulations, papers, and sensor logs; cognitive information such as component function, spatial relationships, and material parameters; and application workflows that allow data to enter AI training and inference. SciVisual has focused its long-term research on the cognitive layer, which it sees as the foundation for any later AI use.
Why Structured 3D Data Matters Now
Five years ago, a structured 3D model was mainly valuable for human viewing, scientific communication, and production efficiency. With the rise of AI agents, the same data can now help models understand tasks, call up relevant objects, and organize workflows. SciVisual is therefore repositioning the same underlying assets from content and visualization infrastructure to AI infrastructure.
Expanding into Industrial Applications
The startup is now applying its research-proven capabilities to industrial customers in sectors such as consumer goods and high-end equipment. Yang Yuanfan describes the company as building the data infrastructure for AI for Science, translating physical devices from geometry, components, and materials into structured information that AI models can call and reason over. SciVisual reported revenue of tens of millions of Chinese yuan this year and currently has a team of nearly 40 people.
Investor Confidence in Scientific Data
Shanghai Future Industry Fund said scientific data is a foundational production factor for future industries and that SciVisual has closed the loop from data entry to structured assets. Pudong Venture Capital described the digital and intelligent expression of scientific objects as a missing piece in the national future industry ecosystem. Houxue Capital added that the company helps researchers turn scientific installations and engineering prototypes into standardized AI digital assets.
SciVisual is positioning itself as a missing data layer between general AI models and the specialized realities of scientific and industrial environments. By focusing on structured 3D data and domain-specific workflows, the company aims to help AI systems move beyond text and images into high-stakes research and engineering applications. Its ability to convert large-scale facility work into reusable infrastructure will be the central test as it scales commercial adoption.