Governor Kathy Hochul signed Executive Order 62, imposing a temporary statewide moratorium on new hyperscale datacenters while New York studies their environmental and community effects [E1]. The order pauses discretionary environmental permitting for as long as one year on facilities that consume or can consume 50 megawatts or more [E1]. Specified research, education and medical uses can qualify for exemptions [E1]. New York describes the measure as the first statewide moratorium of its kind, making an ordinary permitting decision a national precedent for governing AI infrastructure [E2].
New York’s own queue explains the urgency. Datacenter developers had submitted nearly 12 gigawatts of load requests to the NYISO interconnection queue by May 2026, with more than 8 gigawatts entering during 2025 alone [E2]. Hochul paired the pause with a push to repeal sales-tax exemptions for massive facilities and ordered a community-benefits framework within 60 days [E2]. Her demand that large loads “pay their fair share” places utility upgrades, local compensation and tax treatment inside the price of compute [E2].
The 50-megawatt threshold marks where a server campus begins to behave like territorial infrastructure, requiring land, substations, transmission capacity, water and political consent at industrial scale [E1]. A queue measured in gigawatts shows that the constraint has moved beyond the supply of chips and investment capital [E2]. Hyperscalers increasingly arrive before governments as utility-scale petitioners whose projects can alter regional power planning and household cost allocation [E1][E2]. Their negotiations now resemble those of large industrial operators and public utilities, with tax privileges and community benefits settled alongside electricity delivery [E2].
The White House convened a defensive response through its Ratepayer Protection Pledge. Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI agreed to build, bring or buy new power supply and to fund the delivery infrastructure needed for their facilities [E3]. Participating companies also pledged to pay for reserved infrastructure whether or not they ultimately use the electricity [E3]. The terms implicitly accept New York’s central premise: datacenter demand creates public-system costs that hyperscalers must absorb if the buildout is to retain political support [E2][E3].
Residents have begun organizing the same argument nationally, with anti-datacenter protests planned in at least 125 U.S. locations [E4]. Only 14% of Americans surveyed in June said they would welcome an AI datacenter in their own community [E4]. New York lacks northern Virginia’s entrenched cluster, leaving builders room to redirect projects toward states offering faster permits, cheaper power or weaker opposition. The federal pledge remains voluntary and will depend on utility proceedings for practical force [E3]. Protest locations measure organized resistance, not canceled campuses, so the evidence establishes a rising political cost without establishing a halt to construction [E4].
Hochul’s order converts that cost into a state decision with a threshold, a timetable and a permitting freeze [E1][E2]. Each proposed campus must now compete for grid capacity while defending its tax treatment and local bargain [E2][E4]. The White House pledge shows that the industry and federal government already recognize the ratepayer problem, even before regulators determine how firmly the promises bind [E3]. The next AI bottleneck has entered the voting booth [E1][E2][E3][E4].