Data mesh and data spaces are emergent paradigms for scalable data sharing within and across organizations. Despite different motivations and trust assumptions, both rely on data catalogs to provide visibility and discovery of data products. This paper studies data catalogs as a convergence point between the two paradigms. First, we analyze data access and usage policy placement by distinguishing policy source of truth, policy binding, and policy enforcement. We show that in data meshes, catalogs can serve as a single source of truth for policy bindings and enable governance automation through workflow-driven or reconciliation-based approaches integrated with enterprise platforms, whereas in data spaces, catalogs primarily support discovery and contract initiation, with enforcement remaining boundary- and contract-driven at participant connectors. Second, we examine the feasibility of modeling catalogs as knowledge graphs to represent relational metadata, improve discovery, enable semantic interoperability, and support context-aware classification with conservative propagation along lineage. The results clarify how catalogs, policies, and enforcement components align in each paradigm and provide a basis for future design science guidelines for data-sharing system design.