New York State Governor Kathy Hochul signed an executive order on July 14, suspending statewide environmental permits for new hyperscale data centers, with a maximum duration of one year, applying to large projects with electricity demands of 50MW or more. Former President Trump later criticized that such restrictions would weaken the U.S.'s advantage in AI and data center competition.
For investors, the focus is not on whether New York will block all projects, but on the changing constraints of AI infrastructure. Previously, the market mainly looked at GPUs and capital expenditure budgets, but now it also considers power, water resources, land, and local permits.
Hyperscale data centers can be understood as giant power-hungry factories for AI. Project loads may range from tens of megawatts to over a thousand megawatts, involving cooling water, power transmission access, noise, and land use. The stronger the computational demand, the more challenging it is to keep these costs solely on the company's books.
Hochul's logic is that New York cannot continue to greenlight giant projects without uniform rules in place. Trump's logic, on the other hand, supports data centers as representing investment, employment, and computational advantages, believing that the suspension will push projects to other states. This conflict summarizes the contradictions in the next phase of AI infrastructure.
New York Elevates Community Pushback to Approval Variable
The significance of New York's suspension order is not that it will permanently block data centers, but that it elevates the friction previously scattered across resident hearings, town councils, and environmental litigations to the statewide approval level.
According to the Governor's office, the executive order calls for New York to advance a generic environmental impact assessment and establish review standards around energy, water resources, and community impacts. In project terms, this means hitting the pause button first, and then formulating rules on how giant data centers will use electricity, water, compensate communities, and bear external costs.
The market can no longer blindly extrapolate AI infrastructure based on "strong demand, so it will definitely be built." The demand is still there, but the construction path has narrowed. Whether projects can materialize depends on whether electricity prices are raised, who pays for grid upgrades, if water-stressed areas can cope, and whether tax incentives can garner resident support.
For tech companies, this friction may not necessarily alter their long-term expansion direction, but it will change the delivery pace. The more sensitive variables in the valuation model are not just about a state pausing for a year, but whether similar regulations will spread, whether the approval process and power infrastructure will systematically raise costs.
Michigan Indicates Projects Will Slow Down
New York was a landmark event, but not an isolated case. The Saline Township project in Michigan illustrates that local resistance does not necessarily kick a project out, but it will change the way it moves forward.
OpenAI referred to the local Stargate campus as 1GW in scale. Blackstone and Related Digital previously announced they would provide $16 billion in financing for the Oracle data center project in Saline Township, with participants including OpenAI and Walbridge. The project faced opposition from residents, was initially rejected by the township council, and the developer later pushed the project forward through litigation and settlement.
This case sends a very direct signal to the market. As long as the strategic value is high enough and the capital is deep enough, companies may still be able to push projects through legal processes, compensation arrangements, and design adjustments. The cost is a longer timeline, with political and legal costs factored into the project's return calculation.
AI data centers are no longer just a capital expenditure race between cloud providers and developers. They also have to pass through local governance. Residents see the exchange between electricity, water, noise, roads, and tax revenue, while companies see order delivery, computing power online, and customer commitments.
New Mexico Exposes Energy Infrastructure Hard Constraints
New Mexico's Project Jupiter encounters more of an energy bottleneck. In March, the state's land commissioner rejected Energy Transfer's pipeline permit to cross state trust land, and in July, denied its reconsideration request. The pipeline was originally intended to serve the Oracle-related data center project, with the rejection reasons involving greenhouse gas emissions and water resource pressure.
This detail is harder than resident opposition. A data center doesn't just start running when the servers arrive; it needs a stable power supply, transmission access, and cooling systems. If any permitting phase gets stuck, it will turn AI capital expenditure from financial planning into queued engineering.
On July 1, Oracle stated that the project's power plan had been adjusted to Bloom Energy fuel cells. This indicates that large data centers will continue to seek alternative energy configurations, but the alternative solutions themselves imply that costs, technology, and delivery times need to be reassessed.
Such cases will impact investors' assessment of the quality of AI infrastructure companies. Acquiring servers and chips is just the first step; being able to connect the equipment to a stable, compliant, locally accepted energy system will determine whether a project can transition from announcement to revenue.
Power and Site Selection Capabilities Are Being Revalued
The market implication of this shift is not the end of AI expansion, but a change in the winner's criteria. Chip supply is the first barrier to entry, and power, site selection, and permitting capability are becoming the second-tier moat.
Publicly available data indicates that by 2025, approximately $156 billion worth of U.S. data center projects faced local opposition, litigation, or delays. In the first quarter of 2026, around $130 billion worth of projects were categorized as stalled or delayed. These numbers do not mean that the projects have been entirely canceled; a significant portion may have been delayed, relocated, or renegotiated, but they are sufficient to demonstrate that resource friction has affected the pace of capital deployment.
Under pressure are cloud providers and developers advancing projects in regions where the grid is strained, regulations are strict, and water resources are sensitive. Infrastructure giants like Oracle, Microsoft, Google, Amazon, and Blackstone need to reassess the permitting certainty and power arrangements at project locations.
The beneficiaries are also becoming clearer. Regions with power surplus, policy-friendly states, stable power sources like natural gas and nuclear, dedicated power access, technologies like liquid cooling and closed-loop cooling that reduce water usage, are all gaining higher strategic importance. Power is becoming a precondition for capital expenditure to materialize.
Project Migration Speed Determines Valuation Realization
If New York can establish an executable environmental assessment and community investment framework within a year, the suspension order will resemble a rule reset. Project costs might rise, construction timelines might extend, but the compliance path will be clearer. If the standards are too stringent or if other states follow with stricter limitations, short-term friction might escalate into a nationwide bottleneck.
Whether Trump drives federal-level energy and AI infrastructure support will also impact this tug of war. Even so, grid access, local water resources, and residential electricity prices will not be directly smoothed by political posturing.
AI data centers will continue to be constructed, but from now on, those who can secure power, permits, and community acceptance faster are the ones more likely to turn AI capital expenditure into actual revenue.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
