In the data economy, organizations are increasingly engaging in data ecosystems, managing and sharing data as valuable assets to address business and societal challenges. Whereas organizations have traditionally exchanged data vertically along the value chain, we observe a growing trend where they share complementary data assets with others, even at times with their competitors. Despite the rising significance of data ecosystems, research on this emerging form of horizontal data sharing is still scarce. Building on the theory of communities of practice, we study a pioneer data sharing community involving over 40 multinational companies that developed an innovative approach to pool data management efforts. We derive a design theory for horizontal data sharing, structured around eight design principles that align with key dimensions grounded in theory: domain of interest, shared practice, and community. By offering prescriptive design knowledge, our findings make an important contribution to the emerging literature on cross-company data sharing. Our research also provides practitioners with actionable insights on how to establish and operate data sharing communities effectively.