Following the critique of Area Studies and of methodological nationalism, it is often argued that “the state” should not be taken as a unit of study. Departing from discussions within the research network Crossroads Asia that focused not on social and cultural processes contained within certain spatial boundaries but rather on processes that cut across borders, this working paper argues that “the state” should not simply be regarded as a spatial container and that it has not become obsolete for analysis. In particular, anthropological conceptualizations of the state have not only pointed to links and processes that reach across state borders but, following Foucault, also to the dissolution of the boundary between state and society. Suggesting a figurative approach, employing Philip Abram’s contrast of “state system” and “state idea”, and drawing on ethnographic data from Gilgit-Baltistan (northern Pakistan) the paper argues that conceptualizations of the state have to be attentive to local, emic ideas and distinctions like the one between state and government. While the state idea often emphasizes borders, governmental politics often weakens or disregards them. It is concluded that analysis should not reduce the state to a “container” but that we have to come to terms with the actual, empirical manifestations of both state idea and state system in political discourse and practice.