The AI revolution has officially collided with the physical realities of the world’s energy infrastructure. By enacting a one-year freeze on new hyperscale data slot online facilities, New York has become the first US state to formally halt development of data centres using 50MW of power or more for up to one year.
The order was signed by New York Governor Kathy Hochul on 14 July, with the goal of establishing a comprehensive environmental regulatory framework.
While the moratorium legally blankets all commercial data storage facilities of this scale, the freeze was overwhelmingly catalysed by the unprecedented power demands of the generative AI boom.
“We’re in the midst of one of the most significant economic upheavals in generations… perhaps ever,” Kathy says. “These hyperscale AI data centres consume enormous amounts of power, truly threatening to outpace our grid’s capacity. They drive up costs for local ratepayers and I refuse to let those costs get passed down to New Yorkers.”
Previous US President Joe Biden (far left) with New York Governor Kathy Hochul, looking at a 3D rendering of a future Micron factory in 2022. Credit: Mangel Ngan/Getty Images
During the temporary pause, state officials will draft a Generic Environmental Impact Statement to rigorously evaluate how these facilities affect regional power grids, water reserves and air quality.
While this moratorium is a local policy action, its implications are global. The decision sends a warning to the tech sector that the era of unchecked digital infrastructure expansion is over.
What is unfolding in New York serves as an example of how governments worldwide may manage, tax and regulate the physical footprint of the AI boom moving forward.
Why New York has paused 50MW data centres
The primary driver behind New York’s intervention is the unprecedented, volatile nature of AI data centre power consumption. Conventional data facilities draw a relatively steady, predictable baseline of electricity from local grids.
AI infrastructure, however, operates in drastically fluctuating phases. The intense computational training required for foundation models demands vast amounts of energy over multi-week spikes, which then drop to highly variable levels during live user deployment.
This unpredictability is catching utility providers off guard. More than two thirds of utilities (77%) are struggling to forecast the energy demand required by the expansion of AI-driven data centres, according to Capgemini’s report, AI meets the grid: shaping the data center power play.
This forecasting crisis is intensified by the sheer volume of power these facilities require. Capgemini projects that electricity consumption from AI training and inference will rise from 25% to 60% of total data centre power demand within the next three to five years, rapidly pushing aside traditional digital workloads.