The computing ecosystem continues to grow at a breakneck pace and consumes a substantial portion of the world’s electricity. Artificial Intelligence's (AI) recent successes have further exacerbated this computing-driven surge in energy demand. As a result, there is a growing need to integrate various energy sources in the energy grid. However, volatility of some energy sources creates new challenges for the power grid operators, who must continuously balance electricity supply and demand. Achieving the energy efficient data centers would not only help stabilize the power grid but also ensure that AI’s rapid growth remains supported. This project aims to realize that vision by integrating AI data centers into emerging smart grid programs. To this end, the project will develop a data center control framework to regulate data center power consumption with power grid and carbon constraints, all while providing necessary performance guarantees to data center users. By doing so, the project will help stabilize the grid and provide electricity cost incentives for data centers. This project aims to improve fundamental knowledge in AI data center management and optimization. While there is a rich body of research in large-scale computing system management and recent advancements in computing demand response, the state-of-the-practice is substantially lagging due to science and engineering challenges. To enable AI data centers to perform “flexible computing", this project will in