The importance of geography for innovation has long been recognized. Regions like Silicon Valley help companies generate ideas in part by facilitating knowledge exchanges among local organizations. Despite the central place of geography in many explanations of innovation, existing theories remain limited in several respects. Importantly, because they have sought to explain knowledge acquisition, current perspectives do not account for how firms that are proximate to many industry peers internalize, process, and use the volumes of information available to them locally. Further, contemporary theories do not address how, despite lacking the advantages of proximity, some geographically remote firms produce important innovations. The author connects insights from macro and micro level theories of innovation to propose that geography can enhance or constrain firms' performance, but such effects are moderated by intraorganizational network structures. Data on collaborations among inventors and the geographic locations of 454 U.S. firms active in nanotechnology R&D between 1990 and 2004 are used to show that as proximity to industry peers decreases---and spillovers become less common---inefficient networks are beneficial because they create and sustain diversity internally. For firms with high proximity, more cohesive network structures that facilitate information processing are desirable.