The current framework for data center siting was not designed to fail communities. It was designed for a different era, when data centers were smaller, less energy-intensive, and fewer in number. The permitting processes, zoning categories, utility rate structures, and subsidy programs that now govern data center development were built over decades for purposes that did not anticipate the scale and pace of growth that AI demand has produced.
The result is a framework with structural gaps: communities that have no meaningful opportunity to participate in siting decisions before they are made, local governments that lack legal authority to require the information they need, utilities that recover costs from residential ratepayers that should be borne by large industrial customers, and state incentive programs that operate as unconditional subsidies regardless of community impact. Each of these gaps has been identified by researchers, legislators, advocacy organizations, and communities themselves. Each has corresponding proposals for correction.
What follows is a survey of those proposals, organized by level and type. For each, the proposal is described, the evidence or precedent supporting it is identified, and its current status is noted. The picture that emerges is of a reform landscape that is active, contested, and producing real results in some jurisdictions while stalling in others.
State Siting Legislation
Virginia has been the most active state legislature on data center reform, partly because it hosts more data center capacity than any other state and partly because the community impacts of that concentration have become politically visible in ways that can no longer be managed through ordinary economic development advocacy. In 2026, the legislature considered sixty-one data center bills in a single session. Fifteen reached Governor Abigail Spanberger’s desk, with significant legislation addressing both energy cost allocation and siting.
On the siting side, Virginia’s HB 153 established a permit process for high energy use facilities — defined as those requiring 100 megawatts or more — and required applicants to conduct sound assessments examining a facility’s noise profile on homes and schools within 500 feet. Localities were given express authority to require assessments of potential impacts on water resources, agriculture, and historic resources. This was a meaningful change from the previous framework, in which localities lacked clear legal authority to require this information and faced litigation risk for conditioning approvals on it.
The Virginia Joint Legislative Audit and Review Commission’s 2024 study of data centers provided the factual foundation for much of the 2026 legislation. The study documented that “inadequate local planning and zoning have allowed some data centers to be located near residential areas” and that “some localities have zoned industrial areas next to residential areas, even though land use principles state that industrial uses and residential uses should not be zoned next to each other.” It recommended that localities be expressly authorized to require sound modeling studies, to establish maximum allowable sound levels using low-frequency noise metrics, and to require water use estimates before approving data center projects.
Maryland has been developing legislation addressing data center environmental review and community impact. Earlier proposals — including SB 978 in 2025, which would have required full environmental review and impact mitigation plans with community engagement and benefit agreements — failed to advance, but the legislative effort continues. Arizona, Texas, and Georgia have seen legislative activity on both the incentive reform and siting side, with outcomes varying by the political dynamics of each legislature.
What stronger state siting laws would include, based on the Virginia experience and advocacy organization recommendations: mandatory environmental review triggered by scale thresholds (by energy demand, water use, or facility size); community notice requirements before permitting decisions rather than concurrent with them; noise standards specified in legislation rather than left to local discretion; water use disclosure requirements as a condition of permitting; grid impact assessment requirements; and the authority for localities to require community benefit agreements as a condition of approval.
Local Zoning Reform
The reclassification of data centers from commercial uses to industrial uses is perhaps the most fundamental local zoning reform available. Many communities have permitted data centers in commercially zoned areas based on the erroneous characterization of data centers as office-like facilities — a characterization that was more plausible for the small data centers of the 1990s than for the warehouse-scale industrial operations being built today. Reclassifying data centers as industrial uses subjects them to the setbacks, noise standards, and environmental review requirements that apply to other industrial facilities.
Aurora, Illinois, a suburban city west of Chicago, drew national attention in 2026 for imposing some of the country’s strictest local restrictions on data centers, requiring developers to comply with new zoning requirements, energy use rules, water consumption limits, and noise standards. Aurora’s approach represented a local government using its zoning authority aggressively to impose conditions that state law had not required. It is a model that other local governments can follow independently of state legislative action.
Specific elements that local zoning reform can address include: setback requirements measured from residential property lines and from schools; noise limits specified as conditions of approval rather than as general nuisance standards; height and massing standards that limit the visual impact of large-scale facilities in areas adjacent to residential neighborhoods; landscaping and acoustic buffer requirements; restrictions on the siting of data centers in areas zoned residential or mixed-use; and requirements for sound modeling studies before approval rather than after complaints are received.
The JLARC study{target=”_blank”} recommended that localities be expressly authorized to establish maximum allowable sound levels for both new and existing data centers, using alternative low-frequency metrics that better capture the kind of noise data centers actually produce. Standard decibel measurements do not adequately capture low-frequency hum; zoning ordinances that specify measurements using low-frequency metrics are more effective at controlling the specific type of noise that affects neighbors.
Tax Abatement Reform
The reform of data center tax abatement programs has emerged as a significant legislative priority in several states. Good Jobs First has recommended that states cancel data center tax exemptions entirely, on the grounds that they are unnecessary for an extremely profitable industry, uncapped, often indefinite, and producing revenue losses that states cannot accurately project. Short of cancellation, it recommends capping the annual amount of exemption any facility or company can receive and requiring annual public disclosure of revenue losses.
Virginia’s legislature engaged in extended debate over whether to end the state’s data center sales tax exemption, which is scheduled to expire in 2035. The Virginia Senate supported early termination on January 1, 2027, citing $1.6 billion in lost revenue. The House disagreed. The dispute delayed the state budget.
What stronger abatement frameworks would require: community benefit agreements as a mandatory condition of receiving any public subsidy; clawback provisions with specific triggers and enforcement mechanisms; job quality standards (wage floors, benefits requirements, local hiring preferences) tied to the abatement rather than aspirational; annual public reporting on subsidy costs, jobs created, and compliance with conditions; and prohibition on non-disclosure agreements that prevent elected officials from sharing deal terms with constituents.
States with stronger disclosure requirements — those that systematically report data center subsidy costs in annual tax expenditure reports and break out state and local portions — provide a model. Georgia, Washington State, and Tennessee systematically report both state and local foregone revenue. That level of transparency should be the floor, not the ceiling.
Noise Standards
No federal noise standard specifically governs data center operations. The EPA issued community noise guidelines in 1974 that establish recommended levels for residential areas — 55 decibels for outdoor exposure and 45 decibels indoors — but these are guidelines, not enforceable regulations. The EPA’s noise program has been essentially defunct since the Reagan administration eliminated its budget.
In the absence of federal standards, state and local noise ordinances provide the primary framework. These vary widely in their approach, metrics, and enforceability. The core problem with existing ordinances is that they typically measure noise in A-weighted decibels, which underweight low-frequency sound. Data center cooling equipment produces low-frequency hum that A-weighted measurements understate; residents near data centers often experience significant disturbance at measured levels that appear to comply with ordinances.
Local ordinances that have worked better specify measurements using C-weighted metrics or specific low-frequency sound pressure levels, apply both daytime and nighttime standards, and establish a clear process for measuring compliance and imposing penalties for violations. The JLARC study’s recommendation that Virginia localities be authorized to use “alternative low frequency metrics” in their noise ordinances reflects this understanding. The Aurora, Illinois restrictions represent another local model.
State noise standards that apply to industrial facilities — where they exist — provide another potential framework. Some states have enacted industrial noise standards through environmental or public health statutes that could be extended to data centers explicitly.
Water Use Disclosure and Limits
Water use disclosure is among the most actively contested reform areas. As of 2025 and 2026, at least eight states introduced legislation requiring data centers to report water use. The outcomes illustrate both the political difficulty and the achievability of disclosure requirements.
California passed water use reporting legislation through both chambers; Governor Gavin Newsom vetoed it, citing concerns about rigid requirements on an important industry. New Jersey passed similar legislation; its governor also declined to sign. Virginia passed a bill authorizing localities to require water use assessments; Governor Glenn Youngkin vetoed it. Despite these setbacks, Utah enacted HB 76, which requires mandatory water use reporting for large data centers. Minnesota’s HF 16, enacted in 2025, prohibits permit approval for data centers if the facility will cause adverse impacts to watershed health or supply.
Climate XChange’s analysis of state water policy options identifies the range of approaches: banning NDAs between developers and local governments on water use information (California’s AB 1370, enacted 2025); requiring water source, withdrawal, consumption, and discharge reporting; requiring public notification of disclosures; and establishing water efficiency standards targeting values below the average water use effectiveness for U.S. data centers.
Proposed data center growth in Texas could increase statewide water use by an estimated 3 percent by 2030 — an amount not accounted for in the state’s water plan, according to the Houston Advanced Research Center. States that are not requiring water use disclosure are making resource planning decisions in the dark.
The Netherlands provides an international reference point. Amsterdam’s municipality imposed a moratorium on new data centers in July 2019, citing concerns about space and energy. The moratorium ended in 2020 with an agreement allowing construction in designated areas subject to stringent energy efficiency requirements. The Dutch experience shows that democratic governments can impose binding limits on data center development — a reference point for communities and legislators considering whether stronger action is within the range of legitimate policy choices.
Grid Planning Reform
FERC’s interconnection reform proceedings address the question of who pays for the transmission infrastructure that data center growth requires. Public Citizen has called on FERC to reconsider proposals to federalize data center interconnection that would limit state authority over energy aspects of data centers, to accept a moratorium request on new data center connections to the grid, and to clarify that federal reliability standards apply to large data center loads.
PJM’s Expedited Interconnection Track, filed with FERC in early 2026, proposes to accelerate interconnection for large loads — a measure designed primarily to benefit data center developers. The cost allocation dimensions of interconnection reform — determining whether data centers pay the full costs of transmission upgrades they require — are being worked out in FERC and PJM proceedings whose technical complexity makes community participation difficult.
State integrated resource planning requirements provide a complementary lever. When utilities must file long-term resource plans with state PUCs, those plans incorporate data center load projections and proposed generation and transmission investments. IRP proceedings allow consumer advocates to contest the assumptions behind those projections and to challenge resource plans that impose disproportionate costs on residential ratepayers.
Proposals for data centers to bear a larger share of transmission upgrade costs are gaining traction. The Virginia legislation requiring that costs for increased generating capacity be passed on to customers with 25 megawatts or more in electric demand — rather than spread across all ratepayers — represents a concrete implementation of this principle at the state level. The Pennsylvania PUC’s model tariff proposal, approved 3-2 in November 2025, proposes a similar approach for Pennsylvania’s utilities.
Environmental Justice Requirements
Applying meaningful EJ review to data center siting would require, at minimum, cumulative impact assessments that evaluate the full burden on communities already bearing disproportionate environmental loads; community engagement requirements that go beyond notice-and-comment; disclosure requirements that give communities the information they need to evaluate proposals; and the authority to deny or condition approvals based on adverse EJ impacts.
The Illinois EPA’s application of its EJ and Title VI review framework to the Prime Data Centers permit application in Elk Grove Village shows what a state-level EJ review process looks like in practice. It is not a veto — it is enhanced analysis, expanded public outreach, and documented consideration of impacts on overburdened communities. That model could be extended to data center siting generally in states that have EJ review frameworks.
At the federal level, the weakening of EJ requirements under the current administration — the rescission of DOE EJ guidelines, the rollback of NEPA implementing regulations — has reduced the federal framework. State EJ requirements operate independently and can fill some of that gap.
The argument for EJ requirements in data center siting is not only about racial and economic equity, though that equity argument is sound on its own terms. It is also about the self-reinforcing dynamic that results when communities with less political power are systematically selected as locations: they are selected because they cannot resist, and the pattern becomes entrenched. Breaking it requires requirements that apply regardless of a community’s political capacity to resist.
The Direction of Reform
The reform landscape is moving faster than at any prior point. The scale of hyperscaler investment — and the visibility of its community impacts — has made data center policy a live political issue in states where it was invisible five years ago. Legislatures that ignored the issue are now considering dozens of bills in a single session. PUCs that never examined data center load are opening public proceedings.
The direction of reform is reasonably clear: toward mandatory disclosure, toward data centers bearing more of the costs they impose on the grid and on local infrastructure, toward local government authority to require meaningful assessment before approval, and toward community benefit requirements as a condition of public subsidy. The speed and strength of adoption will depend on whether the communities most affected are able to participate in the legislative and regulatory processes where those decisions are made.
This article was researched and drafted with AI assistance under human review. See our full AI and editorial practices.