The densification and expansion of wireless networks pose new challenges on energy efficiency. With a drastic increase of infrastructure nodes (e.g. ultra-dense deployment of small cells), the total energy consumption may easily exceed an acceptable level. While most studies focus on the energy radiated by the antennas, the bigger part of the total energy budget is actually consumed by the hardware (e.g., coolers and circuit energy consumption). The ability to shutdown infrastructure nodes (or parts of it) or to adapt the transmission strategy according to the traffic will therefore become an important design aspect of future wireless architectures. Network infrastructure should be regarded as a resource that can be occupied or released on demand. However, the modeling and optimization of such systems are complicated by the potential interference coupling between active nodes. In this article, we give an overview on different aspects of this problem. We show how prior knowledge of traffic patterns can be exploited for optimization. Then, we discuss the framework of interference functions, which has proved a useful tool for various types of coupled systems in the literature. Finally, we introduce different classes of algorithms that have the objective of improving the energy efficiency by adapting the network configuration to the traffic load.