• alliances;
  • emerging industries;
  • exploration and exploitation;
  • fuel cell;
  • organizational learning

Abstract This paper presents a new perspective on the ambidexterity debate. Basing our work on the assumption that learning is a multilevel process, we develop and empirically test whether, how, and with what consequences firms can balance explorative and exploitative technology-learning activities across their organizational boundaries. Hypotheses are tested through a longitudinal panel of data tracking the cross-boundary learning strategies of 153 companies involved with innovation in the worldwide fuel cell industry over the period 1999–2006. By juxtaposing intra- and interorganizational exploration-exploitation, we find that the pursuit of cross-boundary ambidexterity is worthwhile. Firms that engage in internal exploitation tend to balance such learning orientation with explorative interorganizational agreements. Consistently, those firms engaged in external exploitation tend to balance it with an internal focus on exploration, at least in the case of exploitative alliances involving familiar partners. Moreover, results confirm that such complementary cross-boundary strategies improve a firm's innovative performance.