Based on the case of the biopharmaceutical industry, the aim of this paper is to challenge the core conviction now widespread within the "spatial clustering theory", which devotes a key (if not exclusive) role to geographical proximity in explaining clustering dynamics of innovation activities within specific territories. Our argument is threefold. First, mere geographical proximity is not enough; in many cases, cognitive, organizational and strategic forms of proximity are often at least as crucial as the topological closeness among innovation actors. Second, our idea is that clusters are fundamentally the territorialized outcome of combinations of inter-organizational and social networks among actors pursuing common goals, each of these actors having a specific territorial and social embedding that allows him or her (or not) to operate and interact at different spatial scales. These networks are socially and territorially embedded, but they can operate at various spatial scales. Third, sector-driven dynamicsas is in the case of biopharmaceuticalsmay structurally frame the way the actors interact and collaborate in R&D projects and innovation processes. Indeed, the dynamics underlying the emergence, structuring and evolution of biopharmaceutical clusters are both multi-actor and multiscalar. In this perspective, clusters and networks appear to be intertwined phenomena, consubstantial one to each other, and co-evolving organizational modes of biopharmaceutical innovation.