T heories of network evolution frequently focus on "network endogeneity," which stresses predictable, path-dependent evolution rooted in previous network structure. However, theories of technological evolution and innovation remind us that networks may undergo significant change as technological discontinuities exert pressures on existing relationships and firms engage in exploratory search. How can we incorporate sources of change into our theories of network evolution instead of focusing so squarely on sources of inertia? By using recent advances in graph theory, we develop a more flexible theory of network evolution by examining two patterns of partner selection that have the potential to change networks: "shortcut" formation between relatively unconnected partner clusters, and the entry of new firms into the "main component" of incumbent partners. Our findings suggest an important contingency for the endogeneity perspective: structural homophily predicts shortcut formation but not alliance formation within clusters. Furthermore, we demonstrate that the pattern of alliance formation between incumbents and new entrants to the alliance network is driven by a combination of endogenous and exogenous mechanisms. New entrants attach to more prominent incumbents, but they are more likely to attach with an alliance deal that comprises multiple partners. We demonstrate these findings in an industry where systemic technology encourages cooperation and where network entry is prevalent-the mobile communications industry from 1993-2002.