The project Green@Cloud aims at developing an energy-aware scheduling framework able to reduce the energy needed for high-performance computing and networking operations in large-scale distributed systems (data centers, clouds, grids). With the advent of new petaflops data centers and the next-generation Internet (“Internet of Things”, “High-Performance Internet”), energy consumption is becoming a major challenge for the IT world. To build up this new energy-aware scheduling framework, the project Green@Cloud will first develop multi-criteria mathematical optimization models (e.g. makespan, energy, robustness) and then design multi-objective optimization methods to solve the problem. These techniques along with statistical and machine learning components will be used to provide autonomous fault-tolerant and robust scheduling paradigms for virtual machines running inside a dynamic environment. A series of time-varying deterministic and stochastic factors will be considered as part of the environment, e.g. renewable energy supply, computational demand or activity of users. Experimentation and validation will be carried on a real test bed using large-scale equipments (e.g. Grid’5000) while relying on distributed scenarios.