Modern scientific progress in fields such as fusion energy, materials research, climate science, and biomedical imaging depends on researchers' ability to find and reuse the vast amounts of data produced across the national research ecosystem. National investments such as Globus and domain-specific data repositories have made scientific data transfer and storage efficient and reliable, but they were not designed to help researchers find data by its scientific meaning, and they do not provide the descriptive information that artificial intelligence tools need to interpret datasets and recommend related work across disciplines. As a result, much valuable scientific data remains underused. By laying the groundwork for an intelligent, AI-ready scientific data discovery ecosystem, this project advances the progress of science and supports national prosperity through faster, more open scientific discovery. The project will assess current data discovery practices and requirements across diverse scientific communities, and identify the metadata and semantic information needed to support concept-driven and AI-driven data discovery. The project will develop a community-informed architectural plan for a scalable, AI-ready metadata service that interoperates with existing software and hardware ecosystems. The plan will define representative scientific use cases, technical requirements, design principles, and a governance model. The findings are expected to benefit not only the engage