Skip to content

Data Center Moratorium Wave Expands Across the U.S. in 2026

August 5, 2026 • Garrett Beane
Rows of server racks in an AI data center illuminated by green status lights and an amber warning glow.

The data center moratorium has become a growing policy tool as state and local governments slow or tighten approvals for large facilities. Regulators are examining their effects on electric grids, water supplies, utility rates, land use, and surrounding communities.

The actions do not amount to a coordinated nationwide moratorium. They instead reveal an expanding patchwork of grid reviews, environmental studies, zoning restrictions, and temporary local pauses. For an industry accustomed to competing jurisdictions offering tax incentives and expedited development, that change is significant.

Artificial intelligence is an important part of the demand surge, but it is not the only factor. Cloud computing, streaming, enterprise services, cryptocurrency operations, and conventional data storage also contribute to the growing pipeline of large-load projects.

Texas Freezes Progress for Data Centers Seeking ERCOT Connections

Texas Governor Greg Abbott issued a directive on August 3, 2026, requiring the Public Utility Commission of Texas and the Electric Reliability Council of Texas to verify and audit data centers advancing through ERCOT’s interconnection process before additional projects are approved to move forward.

Projects that do not comply with the verification requirements can be denied a connection to the ERCOT grid. The requested information includes projected electricity and water consumption, cooling technology, on-site generation plans, public incentives, ownership interests, and measures addressing noise, traffic, lighting, setbacks, and emergency response.

Abbott’s directive says ERCOT is considering approximately 474 gigawatts of connection requests—more than five times the grid’s record peak demand—and estimates that data centers represent about 90% of the new power requests.

ITD Insight

An interconnection queue is not a construction forecast. It can include speculative, overlapping, delayed, or abandoned proposals, and the same developer may evaluate multiple possible sites. The 474GW figure demonstrates the scale of regulatory uncertainty facing ERCOT; it does not mean Texas is preparing to add 474GW of confirmed data-center demand.

The Texas action is substantial, but its scope matters. It applies to data centers seeking to advance through ERCOT’s grid-connection process. It is not a blanket suspension of every land-use, construction, or data-center approval across the state. Projects outside ERCOT territory require separate analysis, while facilities planning on-site generation may have less dependence on the public grid but still face local zoning, water, emissions, and community requirements.

The order follows an earlier June directive requiring data centers to bear the infrastructure costs created by their projects rather than shifting those expenses to residential customers. Industry representatives have said clearer verification standards could distinguish viable projects from speculative requests, while critics continue to call for broader rules governing water use, tax incentives, and local impacts.

New York Pauses Certain State Environmental Permits

New York Governor Kathy Hochul signed Executive Order No. 62 on July 14, 2026. The order directs the Department of Public Service to prepare a generic environmental impact statement examining data-center energy demand, water use, air quality, noise, and effects on disadvantaged communities.

While that review is underway, the Department of Environmental Conservation must hold certain incomplete applications for discretionary state permits in abeyance. The order generally concerns facilities capable of consuming 50 megawatts or more, with exclusions for qualifying manufacturing, research, education, and medical facilities.

This is narrower than a simple statewide ban on every new data center. Applications already determined complete by the state environmental agency are treated differently, and the order explicitly says its permitting pause does not apply to approvals issued by local governments.

New York is also considering a grid-acceleration funding mechanism, large-load interconnection reforms, water-withdrawal rules, and a community investment framework. The policy direction is therefore broader than a temporary pause: the state is developing a structure intended to make large users pay more directly for the infrastructure and risks associated with their projects.

Oregon Reviews Statewide Policy as Hillsboro Pauses Applications

Oregon Governor Tina Kotek created a statewide data-center advisory committee in January to study energy, water, land-use, tax, and community issues. According to Oregon Public Broadcasting, the committee is expected to submit its report to the governor no later than October.

The committee’s work is a policy review rather than a statewide freeze. It nevertheless reflects growing concern about how data-center development affects residential electricity customers, agricultural land, grid planning, natural resources, and the state’s enterprise-zone incentives.

Hillsboro, one of Oregon’s largest technology and data-center markets, has taken more direct action. The city enacted a 120-day land-use moratorium on new data-center and battery-storage applications beginning July 27. The pause gives officials time to study potential changes to development standards and local policy.

That moratorium should not be confused with a permanent ban or with Oregon’s separate state-level review. Existing facilities and previously submitted applications may also be governed by different rules.

Local Governments Test the Limits of Their Authority

Municipalities and counties are increasingly using zoning law and temporary moratoriums to gain time before approving additional hyperscale projects.

In Kentucky, communities have considered a range of approaches. Louisville officials have debated both a temporary moratorium and new zoning standards following controversy over a previously approved 1.6-million-square-foot development. A committee vote can advance a proposal, but it should not be described as a fully enacted citywide moratorium until the required council process is complete.

Potential Louisville standards have addressed industrial zoning, building scale, acoustics, infrastructure, and protection against utility-cost shifting. Other Kentucky communities have adopted temporary pauses, while some local governments have rejected moratorium proposals in favor of case-by-case regulation.

Texas illustrates the legal uncertainty surrounding local restrictions. San Marcos used its zoning code to exclude new data centers from permitted uses within the city, becoming the first Texas municipality reported to take that approach. Hill County adopted a temporary pause in May but rescinded it in June after a developer filed a $100 million lawsuit. The county replaced the moratorium with a development-review checklist.

These disputes demonstrate that local opposition does not automatically translate into durable legal authority. State preemption laws, property rights, existing entitlements, procedural requirements, and differences between city and county powers can all determine whether a restriction survives a challenge.

Why Grid Operators Are Becoming More Cautious

Large data centers can operate at high and relatively steady loads, particularly when supporting AI training, inference, or cloud services around the clock. Connecting such facilities may require new substations, transmission lines, generation, and reserve capacity.

The central ratepayer concern is not merely that a large customer uses substantial electricity. It is that utilities may build expensive infrastructure for projects that arrive late, operate below their requested capacity, or never materialize. Without suitable contracts, deposits, minimum payments, or cost-allocation rules, some of that risk can fall on households and small businesses.

Data centers are not necessarily inflexible loads. Some operators can schedule portions of their computing work, curtail demand during grid emergencies, use battery storage, or build dedicated generation. The practical value of those options depends on enforceable agreements, the workload, the facility design, and whether claimed generation is available when the grid needs it.

Water Use Depends on Location and Cooling Design

Water impacts also vary considerably. Some facilities use evaporative cooling supplied by municipal systems, groundwater, or surface-water sources. Others rely on closed-loop liquid systems, reclaimed water, or air cooling that can reduce direct water consumption but may carry different energy or efficiency trade-offs.

For regulators, the useful questions are therefore site-specific: How much water will the facility consume during average and peak conditions? Where will it come from? How much will be reused? What happens during drought restrictions? Does the local water and wastewater system require expansion, and who pays for it?

The Economic Case Is More Complicated Than a Simple Ban

Data centers can produce meaningful property-tax revenue, construction employment, demand for local contractors, and investment in power infrastructure. They may also attract related technology projects or provide an anchor customer for new generation.

Those benefits are highly dependent on the agreement. Permanent staffing can be modest relative to the facility’s capital cost, generous tax abatements may reduce local revenue, and infrastructure obligations can outlast a canceled or downsized project. Communities therefore need to evaluate net benefits rather than relying on headline investment figures.

The emerging regulatory model is less about rejecting data centers outright and more about assigning costs and risks clearly. Developers that provide credible load forecasts, fund necessary infrastructure, disclose water requirements, mitigate neighborhood impacts, and commit to enforceable community benefits are likely to face a more predictable path than projects built around secrecy or speculative capacity reservations.

InsightTechDaily Analysis: The Frictionless Expansion Era Is Ending

The recent actions in Texas, New York, Oregon, Kentucky, and individual municipalities do not establish a national shutdown of AI infrastructure. They do show that large data centers are increasingly being treated as major industrial developments rather than routine commercial buildings.

That shift changes the development equation. Access to inexpensive land and a favorable tax package is no longer enough. Projects must increasingly demonstrate how they will obtain electricity and water, who will finance supporting infrastructure, how neighboring communities will be protected, and what happens if projected demand never arrives.

Technology companies are responding with on-site generation, direct power agreements, battery storage, nuclear partnerships, renewable procurement, and more efficient cooling. None of those measures automatically resolves local concerns, and each brings its own permitting, emissions, reliability, or land-use questions.

The likely outcome is not the end of data-center construction. It is a more selective market in which credible projects advance under stricter conditions while speculative proposals encounter longer reviews, higher financial guarantees, or rejection. For the AI and cloud industries, infrastructure planning is becoming as important as access to processors.

Sources include the Office of the Texas Governor, New York Executive Order No. 62, Oregon Public Broadcasting, the City of Hillsboro, and The Texas Tribune.

Related Reading


Mini PC and full-size desktop PC comparison

The Mini PC Has Grown Up: When a Tiny Desktop Can Replace a Full Tower

See how compact systems are taking on work that once required a full-size desktop, including private computing and local AI workloads.


Apple M7 Ultra local AI workstation concept

Apple’s 1.5TB M7 Ultra Rumor Points to a New Class of Local AI Workstation

A rumored 1.5TB unified-memory ceiling points toward powerful local alternatives for organizations that do not want every AI workload running in a hyperscale data center.