The project aims to create a pan-European federated energy data ecosystem and an innovative Energy AI Gym for the development, training, testing and validation of AI Foundation Models specifically tailored to the energy sector. The initiative will combine interoperable energy data spaces, advanced AI and HPC infrastructures, and real-world validation environments across Europe. The objective is to develop reusable and transferable AI models capable of supporting renewable energy integration, flexibility, grid management, sector coupling, electric mobility and industrial energy optimisation. The project will validate the approach through five complementary European pilots in Sweden, France, Finland, Spain/Iberia and the Netherlands, covering different energy systems and demonstrating the transferability of AI Foundation Models between countries, sectors and operational environments.