This paper enters the debate on the islands of innovation through the lens of the standard Lucas (1988) growth model. It begins with a review of the theoretical details of the model and of the ensuing main empirical results, which can be identified when estimating such model on a sample of 261 EU27 NUTS2 regions. Next, empirical results are interpreted in the light of recent EU innovation and education policies. Our results point to the paramount importance of taking into account patterns of connectivity between “islands” of innovation and other regions. On the basis of our empirical estimates, we claim that future further concentration of innovative activity could achieve maximum returns by enhancing connectivity between spatial innovation leaders and lagging regions. This situation may be characterised as targeting “hubs”, rather than “islands”, of innovation, and is in agreement with “open innovation policy”.