Traditional balance of power theory focused on interactions between sovereign states, but balance of power dynamics also play out within relatively weak states. The recognition that failed states do sometimes experience state consolidation suggests, in line with a growing body of work at the interstate level, that a balance of power system is not an effective guarantee against one actor dominating the others. At the same time, balance of power dynamics complicate efforts to consolidate authority in weak states, and help to explain persistent violence. We thus argue that balance of power dynamics constitute a central challenge for state consolidation, but not an insurmountable one. We construct a game-theoretic model of balance of power politics in which a set of secondary actors must choose whether to align with a dominant actor. If enough actors support the leader, political consolidation occurs around the leader; otherwise, a contest ensues that both allocates spoils to the winners and has the potential to unseat the leader. In this environment, either state consolidation or persistent balance of power politics can emerge in equilibrium. Moreover, we show that, independent of the material environment, state consolidation is more likely when secondary actors jointly believe it more likely. This result suggests that aspirants for state consolidation will have an incentive to adopt risky strategies that may produce dramatic victories, which in turn could precipitate a tipping process in which secondary actors rally to their side. We illustrate the logic of this argument with reference to politics in Afghanistan.