Integration and interaction of next-generation AI-focused data centers with smart grids and district energy systems: The state-of-the-art, opportunities and challenges

The rapid evolution of artificial intelligence (AI) and high-performance computing (HPC) has significantly increased the demand for data center capacity, particularly for Graphics Processing Unit (GPU) data centers. These data centers offer enhanced computational capabilities, but they also consume significantly more electricity than traditional data centers. However, existing reviews primarily focus on the role of traditional data centers for general-purpose computing concerning energy aspects. This paper rethinks the role of next-generation AI-focused GPU data centers as prosumers-both producers and consumers of energy, when integrated with and interacting within smart grids and district energy systems. First, we systematically review the existing strategies and methods to enhance the energy flexibility of data centers within the smart grids and highlight unique computing workload characteristics and AI-driven flexibility of GPU data centers in comparison with traditional data centers. Second, we comprehensively summarize transformative cooling technologies, particularly liquid cooling, the higher-grade waste heat recovery potential of GPU data centers and their various applications. Third, we thoroughly discuss the opportunities and technologies for renewable energy integration and curtailment as key strategies for GPU data center decarbonizations. Furthermore, this study elaborates on potential challenges and future perspectives of GPU data centers within smart grids and district energy systems as prosumers.