We explore the uncertain dimension induced by the complexity of energy systems, analyzing whether and under which conditions low-carbon transitions can effectively take place. By accounting for social and environmental considerations, heterogenous single utility-maximizing agents optimally decide whether to adopt a green technology which reduces carbon emissions, allowing eventually for a green energy transition. We characterize the determinants of the success of such a transition, emphasizing that even if the favorable conditions are met the low-carbon transition may not result in long run environmental improvements due to the path-dependency and metastability phenomena which characterize the complexity arising from agents’ interactions. Public policy may solve these issues by increasing the incentive for single individuals to adopt, ensuring thus the achievement of a permanent low-carbon state. By extending the analysis to a spatial network characterized by multiplexity due to the social and environmental interconnections, we show that spatial interactions negatively affect agents’ adoption incentive and reduce the effectiveness of public policy by interacting in a complex way with path-dependency and metastability. In particular, spatial interactions may require a larger subsidy to support a permanent low-carbon transition, thus neglecting their effects on agents’ behavior and environmental outcomes may compromise our chances to achieve a greener future.