Abstract: The rise of “AI for Science” has made scientific data a foundational and strategic resource for global technological innovation. This study analyzes three representative cases of trusted scientific data spaces. It maps the relationships among six key participating entities: initiators/supervisory authorities, collaborating/co-managing organizations, platform operators, third-party service providers, data providers, and data users. The study finds that the value networks of trusted scientific data spaces are characterized by four key features: the multi-functional role of initiators/supervisory authorities, the high specialization of participants, the dual nature of data value, and the predominance of public funding. At the same time, four major challenges are identified: weak oversight mechanisms, insufficient adoption of emerging technologies, limited capacity for market value creation, and underdeveloped profit models. In response, the optimization strategies are proposed to provide theoretical guidance for building trusted scientific data spaces in China