Scheduling algorithms for shared platforms such as grids and clouds granted users of different organizations access to powerful resources and may improve machine utilization; however, this can also increase operational costs of less-loaded organizations. We consider energy as a resource, where the objective is to optimize the total energy consumption without increasing the energy spent by a selfish organization. We model the problem as a energy-aware variant of the Multi-Organization Scheduling Problem that we call MOSP-energy. We show that the clairvoyant problem with variable speed processors and jobs with release dates and deadlines is NP-hard and also that being selfish can cause solutions at most m α − 1 far from the optimal, where m is the number of machines and α > 1 is a constant. Finally, we present efficient heuristics for scenarios with all jobs ready from the beginning.