New York Governor Kathy Hochul announced a moratorium on the construction of large new data centers on Tuesday [1, 2].

The move represents a significant shift in how the state manages the rapid expansion of artificial intelligence infrastructure. By pausing development, the state aims to prevent potential systemic failures in energy grids and water supplies that could result from the massive resource demands of AI processing.

The executive order establishes a one-year moratorium [1, 3] on permits for new facilities. This restriction specifically applies to data centers larger than 50 MW [4]. According to reports, New York is the first U.S. state to impose such a moratorium [2].

Hochul said the pause is necessary to evaluate the environmental and energy impacts of these facilities. The governor is seeking to avoid a scenario where tech expansion leads to higher utility bills for residents, depleted water supplies, or increased noise pollution [1, 2].

"The bottom line is: Progress shouldn't arrive with a higher utility bill, depleted water supplies, or noise pollution," Hochul said [1].

The state will use the coming year to study how these large-scale operations interact with local infrastructure. This assessment comes as data centers globally face scrutiny for their electricity and water consumption, which are required for cooling the high-density servers used to train large language models.

The administration has not yet detailed the specific metrics that will be used to determine when the moratorium will be lifted, or if certain projects will be granted exemptions during the one-year period [1, 3].

“New York is the first U.S. state to impose such a moratorium”

This policy signals a growing tension between the economic drive to lead in AI infrastructure and the physical limits of state power grids and natural resources. By prioritizing utility stability over immediate industrial growth, New York may create a regulatory blueprint for other states facing similar pressure from tech giants seeking to build massive, energy-hungry compute clusters.