AI data center power fluctuations are severely damaging critical equipment, leading to some facilities operating at 80% capacity and posing risks to grid stability. North American regulators have issued a Level 3 alert.

According to Bloomberg, the power consumption of AI data centers (especially during the model training phase) can surge or plummet dramatically within milliseconds. This severe fluctuation is causing frequent malfunctions or premature obsolescence of critical equipment such as batteries, gas turbines, and internal combustion engines. For example, xAI's Colossus supercomputing facility in Memphis, Tennessee, has already experienced cracks in its gas turbines, and some voltage-stabilizing batteries need replacement within weeks. Sources indicate that the actual operating time of some facilities has dropped to around 80%, far below design expectations. If the problem is not resolved, it will directly impact investors in related projects within the next 12 to 24 months. The North American Electric Reliability Corporation (NERC) has issued a rare Level 3 alert, urging large data centers to immediately address grid stability risks. Companies like NVIDIA and Heron Power Electronics are seeking solutions.