A Review of Power System Challenges Stemming from Large AI Data Center Loads

The deployment of Artificial Intelligence (AI) data centers presents significant challenges for electrical power systems. This proposed paper reviews the impacts of AI data center loads on power system operation by systematically analyzing technical documents, utility reports, and regulatory frameworks. Unlike traditional loads, AI data centers exhibit distinct power dynamics, independent protection philosophies, and geographic concentration, which can destabilize both local and system levels. Notable incidents, such as sharp power fluctuations and load disconnections, raise concerns about system stability. This proposed review aims to synthesize knowledge on this emerging topic by developing a multitimescale framework, validating theoretical predictions, classifying AI-induced oscillatory modes (0.5-23 Hz), examining stability effects in Inverter-Based Resources (IBR) systems, and proposing hierarchical mitigation strategies. These insights provide vital guidance for operators and researchers on the impacts of AI data centers on the grid.