Generating innovative solutions for large-scale multifaceted problems increasingly requires the carefully orchestrated coordination and collaboration of complex collectives composed of multiple teams. However, there are many difficulties inherent in collaborative work, which are often exacerbated when individuals hail from multiple fields, perspectives, cultural backgrounds, and geographical locations. Although collective creativity can be maximized when teams leverage functionally diverse information, often residing outside the boundary of the team, this is only true to the extent that teams can effectively reconcile often-competing perspectives. Resolving these countervailing pressures requires leadership networks - patterns of emergent influence - that enable organizational teams to explore and exploit diverse informational sets. In this thesis, I turn to leadership networks in order to understand how the social structure of influence within cross-functional multiteam systems (i.e., MTSs) holds the potential to catalyze innovative new ideas. I evaluate hypotheses about the structure of leadership networks and resulting creative output in a sample of geographically distributed cross-functional MTSs formed using students completing linked semester-long projects across two universities in the US and France. Findings reveal the structure of leadership networks, both during early exploration and later exploitation phases, has important downstream consequences for innovation. First, my results suggest that throughout exploration and exploitation, innovation arises in those MTSs who exhibit leadership networks high in bridging ties and whose leaders have strong mutual influence on one another. Second, I find innovation arises in those MTSs whose leadership networks are highly concentrated around a relative few members during the exploitation phase.