That is a timely proposition for communities that want investment but do not have a mature playbook for bargaining with sophisticated developers. It also matters to technology users. Remote computing infrastructure may feel abstract to the people using online services, but its physical footprint is governed through local contracts, ordinances, utility arrangements, and public-budget decisions. The quality of those arrangements can determine whether promises about resources, mitigation, and public value can be measured later.
What the FAS guide establishes—and what it does not
Published on September 10, 2026, Before Breaking Ground: A Local Government Guide to Better Data Center Policy and Community Benefits is based, according to FAS, on primary research, public-records requests, a scan of 42 local ordinances and 12 state actions, and a comparison of data-center development and community-benefit agreements.
FAS’s central assessment is not that every local government has failed in negotiations. Rather, the organization says data-center development has moved faster than many governments’ capacity to respond. Its interviewees also reported no structured channel through which governments could learn from one another’s deals. That distinction matters: it is evidence of a reported capacity problem and information gap, not proof that local officials are incapable of protecting their communities.
There is a second reason to read the guide as a framework rather than a settled scorecard. FAS’s own materials contain an unresolved counting inconsistency. The report describes a comparative review of 10 agreements and identifies 10 jurisdictions, while FAS homepage and publication-listing material calls the project an analysis of eight executed community-benefits agreements. The report further notes that only the Lancaster document is explicitly named a “Community Benefits Agreement”; it analyzes nine differently titled contracts as functionally similar arrangements.
That does not invalidate the practical lessons. But it does mean readers should not casually repeat a single figure as though the scope were unambiguous. FAS should clarify whether the difference reflects a distinction between executed agreements and a broader set of comparable documents, or some other methodological choice.
The key shift: make promises measurable
The guide’s strongest contribution is its focus on terms that are concrete enough to monitor. Community benefits can be politically attractive in broad language, but vague pledges can be difficult to enforce. A useful agreement identifies the baseline, the limit, the reporting method, the party responsible, the consequences of noncompliance, and a process for disputes.
FAS recommends that future agreements address several areas that are easy to mention but harder to specify:
- Electricity: Agreements should account for sudden demand surges and state whether power will be self-generated or grid-connected.
- Water: Use should be reported in gallons, with procedures that apply when demand exceeds normal levels.
- Noise: Terms should establish measurable thresholds and require annual testing.
- Remedies: Parties should consider consequential damages and mediation or other alternative-dispute-resolution mechanisms.
These are recommendations, not a universal legal template. A city may have different authority, utility arrangements, environmental conditions, or procurement rules than another. Still, the direction is sound as a negotiating discipline. “Manage water responsibly” is difficult to verify. A requirement to report gallons and define what happens when usage rises above normal is a term a government and its residents can inspect.
The same is true for power. A data center’s connection to the grid or use of self-generation can have very different practical implications, so treating electricity supply as a background technical detail risks leaving the most consequential questions outside the agreement. The FAS approach is not a guarantee that a project will have no resource impact. It is an attempt to ensure that the relevant terms are disclosed, negotiated, and enforceable.
Lancaster shows how specificity can look
A City of Lancaster, Pennsylvania document offers a concrete illustration of the kind of detail at issue, though it must be described carefully. The document is labeled a draft, and its effective-date field is blank, so the available material does not establish that it is a fully executed final agreement.
Even so, its provisions demonstrate the difference between an aspiration and a negotiable obligation. The draft states that municipal-water use may not exceed 20,000 gallons per day at either campus. That is not the same as a claim about all water associated with the sites, nor does it resolve every local water question. It is specifically a cap on municipal water use. But it creates an identifiable ceiling rather than leaving the subject to a general commitment.
The Lancaster draft also ties transmitted-noise limits to pre-construction ambient levels at receiving properties. Again, the important feature is the baseline. Without a baseline and a testing method, a noise promise can devolve into conflicting perceptions after construction. A pre-construction reference point gives the parties a way to assess whether a limit has been exceeded.
Finally, the draft requires either a $20 million letter of credit or an acceptable corporate guarantee to secure Article 8 obligations. The available excerpt does not establish every obligation contained in that article, so it would be wrong to characterize the security as a guarantee for every promise in the document. It does, however, show an effort to attach financial backing to a specified class of commitments.
For municipalities, this is a practical lesson. A developer commitment has more force when the agreement says what is measured, who verifies it, and what financial or legal mechanism supports compliance. For residents, it creates clearer questions to ask at public meetings: What is the actual cap? Where is the baseline? Who receives the reports? What remedy applies if a commitment is missed?
Tax breaks should not be written as permanent assumptions
The guide also challenges the idea that a tax incentive should remain untouched for the life of a project. FAS recommends that state and local governments avoid granting tax exemptions or economic incentives without an opportunity to reopen negotiations after five years. Its rationale is uncertainty: property values, employment needs, and the economics of a deal can all change.
A reopening provision is not the same as an automatic tax increase or a rejection of incentives. It is a recognition that a long-running agreement can be made on assumptions that no longer hold. A scheduled review gives both sides a formal moment to assess whether promised public value, economic conditions, and fiscal costs still justify the arrangement.
The fiscal stakes are potentially large, but they must be described with care. Good Jobs First reported in June 2026 that four states were known to be losing $1 billion or more annually through data-center tax exemptions. Its reported figures included projected losses of $2.5 billion for Georgia in fiscal year 2026, $1.94 billion for Virginia in fiscal year 2025, and $1.6 billion for Ohio in 2025.
Those figures support a claim about reported or projected revenue losses in particular states. They do not establish a precise national total, and they should not be treated as perfectly comparable across every jurisdiction. Good Jobs First also said disclosure is incomplete: 14 of 37 states with relevant sales- and use-tax exemptions do not publish official revenue-loss figures in a timely way.
That uncertainty itself strengthens the case for better deal terms and review points. If the public cost of an incentive cannot be consistently measured, officials should be cautious about making irreversible commitments based on estimates that cannot later be checked.
Decommissioning is an untested part of the bargain
The guide raises a less visible issue: what happens when a data center eventually closes, changes ownership, or reaches the end of its useful life. FAS reports that multiple interviewees were unaware of a data center that had ceased operations or been formally decommissioned. It therefore describes enforcement provisions in current agreements and ordinances as theoretical to date.
That statement is narrower than it may first appear. It reports what interviewees knew; it does not prove that no U.S. data center has ever been formally decommissioned. The absence of a familiar example in the interviews should be treated as a warning about limited operational precedent, not as a nationwide factual conclusion.
FAS’s own recommendation is also more limited than some retellings suggest. It calls for an inventory, information about facility lifetime, and a decommissioning timeline to support e-waste planning. Separately, one local-government interviewee recommended approval of a decommissioning plan before approving a data center. That may be a sensible policy idea, but it should not be misrepresented as the institutional FAS recommendation.
The distinction matters because end-of-life requirements can impose different responsibilities: documenting equipment, removing it, remediating a site, funding the work, and planning for electronic waste are not interchangeable obligations. Communities should ask which of these duties is actually covered rather than assuming that the word “decommissioning” solves all of them.
Transparency is more valuable than unverified claims
Water reporting is especially important because public arguments can quickly outrun verified facts. Separate reporting has described a lawsuit by a former AWS water-sustainability program manager alleging false or misleading AWS statements about water use and sustainability at Northern Virginia data centers. AWS reportedly declined substantive comment.
These remain allegations in the available reporting, not findings that AWS made false statements. No primary complaint, response on the merits, or judicial decision was available in the supplied record. The responsible conclusion is not to use the case as proof about AWS or the industry as a whole.
Its relevance is narrower: it illustrates why agreements and policies should rely on reporting requirements, defined metrics, and auditable obligations rather than broad sustainability claims alone. Clear disclosure helps communities evaluate performance without requiring them to resolve disputes through competing marketing language.
A more durable bargaining checklist
The FAS guide does not argue that every proposed data center should be rejected, nor does it show that a single agreement can eliminate every infrastructure or budget risk. Its more useful message is that communities should negotiate with an expectation of uncertainty.
Before ground is broken, local officials and residents can press for answers that fit the evidence in the guide:
- Are electricity demand surges, grid connection, and self-generation addressed explicitly?
- Is water use reported in gallons, and is above-normal use governed by a defined process?
- Is there a pre-construction noise baseline, a measurable threshold, and recurring testing?
- Are remedies specific, including a viable dispute process and, where appropriate, financial security?
- Can tax incentives be reconsidered after five years as conditions change?
- Does the project provide a credible equipment inventory, lifetime information, and a timeline relevant to e-waste planning?
These questions will not produce identical answers in every municipality. But they move the debate from whether a data center sounds beneficial to whether its commitments can be tested, funded, and enforced. In an era when computing demand is physically concentrated in fewer places, that may be the most valuable form of local leverage available.