The present paper studies emissions abatement in a game of technological investments. Players invest in competing technologies. One technology is cheap and dirty, the other expensive but green, and technological investments are strategic complements. The paper makes three contributions. First, it resolves complications due to equilibrium multiplicity in games of technological investment by addressing equilibrium selection through the use of global games. The unique equilibrium may be inefficient, in which case policy interventions are needed. This leads to my second contribution, which is to introduce network subsidies. Network subsidies virtually guarantee efficient adoption of the green technology but do not, in equilibrium, require any payments to be made. Third, a two-stage extension of the game sheds a new light on the (mostly experimental) literature on institutional choice. In particular, players decide in stage 1 whether or not to make technological investments strategic complements in stage 2; that is, they choose between playing a prisoners’ dilemma or a coordination game in the second stage. In contrast to the existing literature, my global games approach allows me to characterize a unique equilibrium irrespective of the type of game chosen by the players. The two-stage game has a unique perfect Bayesian equilibrium and allows for sharp predictions.