Intestinal villi are luminal protrusions that are essential for efficient nutrient absorption. In mammals, villus formation initiates during embryogenesis with evagination of the intestinal epithelium and underlying mesenchyme into the lumen. This event is tightly coupled both temporally and spatially with the emergence of PDGFRa-expressing mesenchymal cells immediately below the epithelium that condense to form aggregates. While several signaling pathways have been linked to mesenchymal cell clustering and subsequent villus initiation, the physical and cellular mechanisms of villus cluster formation remain unclear. To uncover these mechanisms, we performed live imaging of intestinal explants and discovered that dynamic mesenchymal motility drives the formation of mesenchymal cell clustering and initiates epithelial morphogenesis. To identify the molecular underpinnings of this process, we performed single-cell RNA sequencing at multiple stages of villus morphogenesis, which revealed the emergence of cell adhesion and contractility programs within clustering PDGFRa-expressing cells. Further, pharmacological inhibition of these programs prevents clustering and villus formation, implicating that they are required for this process. A computational vertex model is used to simulate the clustering dynamics, demonstrating that differential motility and adhesion are both necessary and sufficient for the clustering. The computational model directly connects cluster formation and villus initiation to the cellular properties. The model predictions and results obtained from explant tissue and in vitro engineered system display close correspondence, suggesting the robustness and applicability of our proposed model. Together, these results provide a new understanding and a cellular mechanism describing how mammalian intestinal villi are initiated.