The exponential growth of artificial intelligence infrastructure poses a significant threat to grid stability, as data centers' power consumption patterns become increasingly volatile. Utilities are struggling to accommodate the anticipated growth of hyperscale facilities and high-density compute clusters, which are projected to account for 3 to 4 percent of global electricity demand by the end of the decade1. The issue lies not only in the sheer amount of power consumed but also in the unpredictable behavior of large-scale compute systems. This volatility can quietly test grid limits, making it challenging for utilities to maintain a stable power supply. The emerging concern is that the unstable power usage of AI infrastructure can have far-reaching consequences, potentially disrupting the entire grid. This matters to practitioners because understanding the volatile power use of AI systems is crucial to developing strategies for mitigating grid instability and ensuring a reliable power supply.