US Electric Grids Confront AI Power Crisis, Eye Emergency Cut-Offs for Data Centers

U.S. electricity grids are grappling with an unprecedented surge in demand, primarily fueled by the explosive growth of energy-hungry data centers. In response, policymakers and grid operators are now advancing controversial strategies, including proposals to disconnect these facilities during power emergencies to safeguard grid stability.

Texas has emerged as a frontrunner in this ‘tough-love’ approach. Following a deadly 2021 winter blackout, state lawmakers passed a bill in June mandating standards for utilities to disconnect large electricity users during peak demand. This move aims to prevent widespread outages on the few days annually when extreme temperatures push the grid to its limits.

The trend is rapidly expanding. PJM Interconnection, the nation’s largest grid operator serving 65 million people across 13 mid-Atlantic states, has just released a proposal that could deny guaranteed electricity to new data centers during emergencies. Similarly, the Southwest Power Pool, covering Great Plains states, anticipates expanding power-reduction programs for its biggest users.

The push for these measures comes as power plant construction struggles to keep pace with data center expansion, particularly with the late 2022 debut of generative AI technologies like OpenAI’s ChatGPT, which demand vast computing power. Analysts warn that data center load could overwhelm the grid, leading to rising electricity bills for average Americans.

Big Tech and data center operators are pushing back, advocating for flexible standards and financial incentives for voluntary shutdowns. While Google recently struck a confidential deal in Indiana to reduce its data center’s electricity use during grid stress, the broader industry is wary of mandatory disconnections, citing potential market instability and investment deterrence.

The debate highlights a fundamental shift in energy policy: whether to build billions of dollars worth of new power plants to serve data centers for a few hours a year, or to implement demand-side management that prioritizes grid resilience and potentially saves money for regular ratepayers.

阅读中文版 (Read Chinese Version)

Disclaimer: This content is aggregated from public sources online. Please verify information independently. If you believe your rights have been infringed, contact us for removal.