The Rising Cost of Fighting Data Centers

Local battles over proposed AI data centers are no longer just about noise, water, or zoning. They have become expensive contests of outreach, process, and messaging — and the bill does not stop when the campaign does.

Residents, advocacy groups, and industry organizations are spending real money on texts, mailers, legal filings, advertising, and organizing. Sometimes it is a few thousand dollars. In at least one county it is well over a million. And when opposition succeeds, communities forfeit some of the largest local revenue streams on offer anywhere in the economy. The cumulative cost across communities — money spent to fight, and money lost by winning — is growing fast.

A note on what follows. The figures below come from three very different kinds of source: published vendor price lists, official government records, and projections produced by parties with a direct financial stake in the outcome. We label which is which throughout, because on this subject the label frequently matters more than the number. Where a figure is derived by arithmetic rather than reported by anyone, we say so. Where a claim rests only on an advocacy group’s own website, we say that too.

What the Tools Actually Cost

Mass text messages. Political and advocacy SMS platforms publish per-message rates in a narrow band. Scale to Win charges 1.5 cents per outbound SMS segment on a long code; GetThru charges 3.5 cents, plus a one-time $300 setup fee credited back at $5,000 of spend; Hustle charges 4 cents plus a $250 platform fee. A targeted neighborhood blast of 5,000 to 10,000 texts therefore runs roughly $75 to $400 in message costs alone.

Political senders must also clear carrier registration first. The Campaign Verify token costs $95 for all committee types, once per entity per two-year election cycle. On top of that sit 10DLC brand and campaign registration fees, which one vendor itemizes at an average of $114 for a political use case. But this is vendor-dependent: Scale to Win states that it covers every 10DLC fee except the Campaign Verify token, which puts a group’s one-time compliance cost as low as $95. Call it $95 to $115, depending on who you sign with.

Direct mail. A political postcard or letter — design, printing, list, and postage included — commonly runs $0.35 to $1.10 per piece, with 4x6 postcards at the low end and self-mailers at the high end. Oversized formats run roughly $0.85 to $1.40 all-in. Mailing 5,000 to 10,000 households therefore costs roughly $2,000 to $10,000 through standard mail classes. USPS Every Door Direct Mail can cut that substantially by saturating carrier routes at $0.26 per piece in postage, with no address list required — pieces are simply addressed to “Postal Customer.” Industry-backed efforts that carpet an entire county push well past any of these figures, as the Virginia numbers below show.

Flyers, door hangers, and yard signs. Printing in volume is cheap. Discount printers quote flyers at about 6 cents each at 5,000 copies and 16 cents at 1,000; full-service and union shops run two to three times higher. Door hangers range from 9 cents each at 1,000 down to 4 cents at 10,000. Standard double-sided 18x24 corrugated yard signs run about $5 to $6 each in quantities of a few hundred to a couple thousand, and stakes are 58 to 67 cents each in bulk against $1.70 to $2.34 at retail. A few hundred signs is therefore roughly $1,200 to $1,800 in materials.

The real cost is distribution. Paid canvassing vendor Knock AI publishes a rate card of $2.00 per door urban, $2.50 suburban, $3.50 rural, and $4.00 very rural — we present that as one vendor’s published pricing rather than an industry standard, because no other vendor we could find publishes per-door rates at all. Canvassers themselves average about $22 an hour by one aggregator’s estimate and $21.32 by another. There is no Bureau of Labor Statistics occupation for “canvasser,” so anyone citing federal wage data for this figure is citing something else. Volunteer efforts do not eliminate the cost; they convert it into unpaid time.

Legal appeals and process. Filing fees are modest. In Colorado Springs, the city’s own appeal form requires a $176 check to appeal an administrative decision to the Planning Commission. Six such appeals cost about a thousand dollars in pure filing fees.

The legal work is the expense. Land-use attorneys bill in the range of $200 to $600 an hour, and Clio’s 2025 rate data puts the national real estate practice average at $377 against a $349 all-practice average. Expert witnesses are better documented than the lawyers: SEAK’s 2024 survey of 1,633 experts reports a median of $450 an hour for file review, $500 for deposition and trial testimony, and a median retainer of $3,000.

What we cannot tell you is the total. There is no bar association survey, no American Planning Association figure, and no academic study establishing what a contested land-use appeal actually costs start to finish. The $10,000-to-$30,000 range that circulates widely traces back to a real-estate marketing blog describing a routine, unopposed residential-to-commercial rezoning — not a contested one. The only source we found addressing opposed cases directly puts them at $40,000 to $75,000, and $75,000 to $150,000 or more with serious opposition — also a marketing blog. We can price the hours with confidence. We cannot reliably price the number of hours, and neither, as far as we can determine, can anyone else. Even when counsel works pro bono, that donated time represents five figures of legal value someone is absorbing.

Digital advertising and petition promotion. Boosting social and search ads for a local issue costs a few hundred to several thousand dollars for meaningful reach. Average cost per thousand impressions tracked at $8.19 on Meta and $7.61 on YouTube as of late 2025, though no credible benchmark exists specifically for local advocacy advertising, and anyone quoting one should be asked where it came from. Hosting a petition on Change.org is free. Its paid promotion is priced dynamically at checkout rather than published, but the platform’s own help page offers worked examples — $2 buying about 18 views, $30 buying about 545 — which imply roughly $55 per thousand views at the larger contribution and $50 to $111 per thousand across the examples given. That is a figure we derived from their arithmetic, not one Change.org publishes. And it buys views, not signatures. Reaching hundreds of thousands of signatures requires either organic amplification or substantial spend.

Other costs. Crowdfunding is not free money: GoFundMe takes 2.9 percent plus 30 cents per donation on personal fundraisers, 2.2 percent plus 30 cents for verified 501(c)(3) charity fundraisers, and an additional 5 percent on recurring gifts. Rallies require permits, sound, and materials. Coordinating any of it consumes staff or consultant time.

The public side pays too, and increasingly says so out loud. Shawnee County, Kansas, raised its conditional-use permit fee from $700 to $15,000 on July 2, 2026 — a change that covers energy storage and solar conversion systems as well as data centers, not data centers alone. County Counselor Rich Eckert put the rationale plainly: the resolution lets the county hire whatever expert it needs and bill the applicant, so that “we’re not going to have the citizens of Shawnee County pay for these experts.” In Virginia, Prince William County supervisors amended their rules of procedure on March 19, 2024 to adjourn regular meetings after midnight, following hearings that had run past 5 a.m. The rule is suspendable by majority vote, and the county attorney advised that the Board would likely suspend it for land-use hearings anyway, because Virginia law requires re-advertising a continued public hearing — which blunts its effect on exactly the data-center hearings that prompted it.

How the Money Shows Up in Real Fights

Colorado Springs. Opponents of Project Taurus — a Raeden data center planned for a former Intel chip plant on Garden of the Gods Road — filed six appeals of the city’s administrative approval, of which five met filing requirements and were heard. Each carried the $176 fee. The group behind them raised more than $4,000 through GoFundMe for filing fees, legal advice, and possible media outreach, alongside a standing PayPal legal defense fund. The city planning department reported receiving more than 1,000 comments from citizens and organizations. The opponents’ own website reports 5,349 petition signatures and states that the appeal was prepared with pro bono counsel — both figures are self-reported by the group and we found no independent confirmation of either.

The hearing itself is the clearest illustration of what process costs. It began at 9 a.m. on July 23 and ran nearly fifteen hours, with the vote coming shortly before midnight. Commissioners denied all five appeals 6-2, attaching four conditions including a 50-megawatt cap and ten years of utility reporting. The documented cash outlay by opponents was modest — roughly $5,000 — but only because counsel was donated. Priced at the market rates above, the legal effort alone represents a five-figure subsidy, before counting volunteer hours. The city, meanwhile, absorbed a marathon special hearing and months of staff review for a project that was already permitted under existing industrial zoning and required no council vote at all. Appellants have until August 3 to take the fight to City Council, at another $176 per appeal.

Prince William County, Virginia. Here the money is not estimated. It is disclosed. Virginia Connects, an advocacy group created by the Data Center Coalition and launched in fall 2024 with backing from Amazon Web Services, CloudHQ, CyrusOne, Equinix, STACK Infrastructure, and Vantage Data Centers, blanketed county residents with mailers and text messages. It has since escalated: at least $708,000 in television advertising since June 9, 2026, of which roughly $243,000 is booked for October and November, plus at least $19,000 on Facebook. Its North Carolina counterpart spent about $367,000 on 747 ad spots in a single month. The Data Center Coalition’s political action committee has raised $640,100 since 2024.

Opponents paid too. Every property owner in the Oak Valley subdivision agreed to pay an extra $1,000 in homeowners association fees to fund the lawsuit that ultimately voided the Digital Gateway rezonings. The Coalition to Protect Prince William County reported $55,672 in revenue and $52,020 in expenses on its most recent Form 990-EZ, with no compensation paid to anyone.

And taxpayers paid most of all: $1.72 million in county legal fees over nearly thirty months, on top of a 27-hour approval hearing in December 2023. The sequence at the end is worth getting right, because it is often compressed: a circuit court voided the rezonings in August 2025 over improper public notice, the Court of Appeals upheld that ruling on March 31, 2026, the Board of Supervisors voted unanimously to withdraw from further appeal on April 14, and the project finally died on July 2, 2026, when a QTS affiliate withdrew its petitions to the Virginia Supreme Court. The county walked away in April. The developer walked away in July.

Nashville. The petition against a DC BLOX data center beside the Nashville Zoo was launched by the zoo itself and has passed 548,000 signatures, driven substantially by celebrity amplification from Brad Paisley, Jack White, and Sheryl Crow. Cash outlays for the petition itself were minimal, but the zoo retained both a land-use attorney and an environmental-rights lawyer, and the public process was enormous: a record-setting Metro Council meeting on July 7 issued more than 200 speaker tickets and did not finish public comment until roughly 1 a.m. The mayor signed Executive Order 59 on June 15, directing five agencies to study electrical, water, transportation, air quality, and public health effects within two months. On July 21 the council unanimously passed a moratorium running through December 1, 2026 plus a suite of restrictive ordinances.

The project is delayed, not dead. DC BLOX closed on the 24-acre site for $23 million and claims vested rights, and Metro’s own zoning administrator wrote that the new council rules are “not applicable to this project” under Tennessee’s vested rights statutes — a conclusion council members dispute and which is headed for the Board of Zoning Appeals and probably a courtroom. The city’s remaining lever is eminent domain, on a parcel the county assesses at $37,400,600. Stopping one building could cost Nashville taxpayers eight figures for land the developer bought for $23 million.

Northern Utah. Opposition to the Kevin O’Leary-backed Stratos campus in Box Elder County ran through formal water-rights protests, and the cash costs were higher than they look. Utah charges $15 to file a protest. Roughly 3,800 residents filed against the project’s first application and nearly 700 against a second — roughly $67,000 in filing fees, a figure we calculate from the protest counts rather than one any outlet has published. Both applications were then withdrawn, which voided every protest. A state water rights spokeswoman confirmed the fees do not carry over: opponents will pay again when the applications are refiled.

Add an open letter to the governor signed by more than 7,500 people, two Capitol rallies drawing 400 to 500 and then more than 600, and a county poll of 513 registered voters commissioned by opponents through Change Research, and the fight consumed real money and enormous unpaid labor.

It is worth being precise about what happened next, because the sequence is easy to get backwards. Utah House Bill 60 narrowed the grounds on which the state engineer may weigh a protest, confining public-welfare analysis largely to water quality and scarcity and stripping out broader environmental, economic, and social effects. But HB 60 passed both chambers in February and was signed on March 23, 2026 — two days before the first Stratos water application was even filed. The law was not retaliation for the protest wave. What it was, as Utah News Dispatch reported, is a statute that took effect on May 6, days after the developer withdrew — meaning any refiled application will be judged under the narrower standard. Same practical result for the protesters. Very different story about why.

Spending Thousands to Lose Millions

There is a second ledger, and it is the one local governments feel longest: what a community gives up when the opposition wins. These figures deserve careful labeling, because nearly all of them originate with parties who wanted the projects built.

Prince William County spent $1.72 million in legal fees before the Digital Gateway died — and with it went a projected $400.5 million a year in tax revenue from $24.7 billion in investment. That projection came from county finance staff, who described it as a figure reached at the end of twenty years, not immediately; the developer separately claimed $500 million. Tucson’s unanimous rejection of Project Blue walked away from roughly $250 million in ten-year tax revenue and 180 permanent jobs averaging $64,000 — figures that come from an Applied Economics study commissioned by the Chamber of Southern Arizona, not from an independent government analysis. And Tucson’s loss was not regional: the project relocated to a 290-acre parcel in unincorporated Pima County, leaving the city with the fight’s costs and none of the revenue. In Colorado Springs, the developer’s own materials claim more than $1 million a year in property tax and 60 to 100 permanent jobs, while the economic study it commissioned put the figures at $1.2 million and 95 jobs; the local school district told the Gazette it had conducted no independent financial review of any of it.

The benchmark all of these communities are declining is Loudoun County, Virginia, where data centers paid $894.5 million in local taxes in fiscal 2025 — 36.8 percent of local tax funding, rising to 42 percent in fiscal 2026. Note the denominator: the county elsewhere describes data centers as generating 38 percent of general fund revenue and “almost half” of property tax revenue, and those are three different measures of three different things.

The ratio most often quoted in support of all this deserves a closer look than it usually gets. The claim that data centers pay roughly $13 in taxes for every $1 in county services traces to a Federal Reserve Bank of Richmond article, which is real — but the Fed is relaying, not finding. The underlying analysis is by Mangum Economics on behalf of the Northern Virginia Technology Council, a trade association, using 2020 data. Its cost side is a narrow estimate of general-fund budgetary cost, excluding grid buildout, transmission, water infrastructure, and ratepayer cost-shifting — which is to say, excluding most of what the current fight is actually about. The ratio also moves sharply between editions: 15.1-to-1 for Loudoun on 2018 data, 13.15-to-1 on 2020 data, and 26-to-1 on 2022 data, with the 13-to-1 figure now describing Prince William rather than Loudoun.

The wage figure needs a similar correction. The frequently cited $134,308 average is real and comes from independent federal data — the BLS Quarterly Census of Employment and Wages — but it measures NAICS 518210, a cloud-and-software payroll code covering hosting, streaming support, and application services. In Virginia, 54 percent of those jobs sit in Fairfax County office parks and only 10 percent in Loudoun’s Data Center Alley. No BLS series isolates data center operations wages anywhere in the country. The closest independent proxies for the people who actually run these buildings are considerably lower: network and systems administrators average $116,130 in Virginia, user support specialists $70,650, HVAC mechanics $61,190.

The Other Ledger

A watchdog that reported only the revenue side of this would be doing public relations. The independent evidence against the fiscal case is substantial, and in several places it is stronger than the case itself.

Virginia’s Joint Legislative Audit and Review Commission — the legislature’s own nonpartisan oversight body — found in December 2024 that the state’s data center sales tax exemption “does not pay for itself,” returning about 48 cents per dollar spent. It also found that the standard IMPLAN modeling used in promoter impact studies is biased upward, overstating employee compensation and inflating the operations employment multiplier by roughly 20 percent, and that most of the economic benefit comes from construction rather than operations: about 50 full-time jobs per 250,000-square-foot facility against roughly 1,500 workers at construction peak.

Georgia’s state auditor commissioned an evaluation that found 70 percent of data center projects would have located in Georgia without the subsidy, and a net fiscal impact of negative $432.6 million in 2025. Official forecasts of what these exemptions cost have failed in the same direction in every state we examined: Ohio projected $136 million and lost $1.6 billion; Texas projected roughly $180 million per biennium against $3.2 billion for 2027-28; Virginia’s own agencies report figures for fiscal 2025 ranging from $1.6 billion to $1.94 billion depending on which body is counting. Ohio State’s Gabriel Lade, reviewing the pattern, notes that payments in lieu of taxes typically recover only 15 to 30 cents per dollar forgone.

On electricity, the strongest evidence comes not from advocates but from the independent market monitor FERC requires for the PJM grid. Monitoring Analytics calculated that including data center load raised capacity auction costs by $6.5 billion in the 2027/2028 auction and $23.1 billion across the last three auctions combined — costs borne by other customers. Capacity market costs are not the same as retail bills and reach households only partly and with a lag, but a report by Synapse Energy Economics for the District of Columbia Office of the People’s Counsel, a statutory consumer advocate, translated the capacity surge into roughly $10 a month for a typical residential customer. Harvard’s Electricity Law Initiative, reviewing about fifty regulatory proceedings, documented a $325 million utility discount to a single large customer alongside internal planning to shift that cost to other ratepayers.

In fairness to the other side, an EPRI working paper finds data center growth actually lowered residential rates about 6 percent over 2019 to 2024. EPRI is utility-industry funded, the paper is not peer reviewed, and its own authors caution that future supply constraints could reverse the effect — its window ends precisely where the market monitor’s evidence begins.

The Larger Picture

These are no longer low-budget neighborhood disputes. A single contested project can generate anywhere from ten thousand dollars to well over a million in combined spending by opponents, industry groups, and local governments; Prince William County alone accounts for $1.72 million in public legal fees plus six figures of industry advertising.

Multiply that across the communities now facing proposals. Data Center Watch, a project of the AI research firm 10a Labs, counted $156 billion in projects blocked or delayed during 2025, and 75 projects worth roughly $130 billion in the first quarter of 2026 alone. Its data, first reported in detail by NBC News, puts active opposition groups at 833 across 49 states as of March 2026, up from 396 at the end of 2025. Good Jobs First counted 54 local moratoria already passed as of February 2026 — a snapshot that has grown considerably since, and one worth re-checking against current records before anyone repeats it.

Public opinion has moved with it. Two separate polls, using different populations and different instruments, arrived at the same topline: Gallup found 71 percent of American adults opposed to a data center in their local area in polling fielded March 2 to 18, and a Heatmap News poll conducted by Embold Research among 4,118 registered voters, fielded May 15 to 28, found 71 percent opposed — up from 42 percent the previous September. The intensity differs between them, and the coincidence of the toplines should not be mistaken for one finding confirmed twice. Data Center Watch’s earlier survey of elected officials taking public positions against large projects found a 55-45 split between Republicans and Democrats, a figure now more than a year old but the only one on the record.

Meanwhile the industry’s advocacy apparatus has scaled to match. The Data Center Coalition now operates near-identical state advocacy nonprofits in seven states — Oregon, Indiana, Ohio, North Carolina, Pennsylvania, Texas, and Virginia. The Coalition reported about $4.6 million in operating revenue in 2024. Funding for the individual state campaigns is not publicly disclosed. That last sentence is the one this organization exists to write, and it applies with equal force to the opposition groups, whose funding is frequently no more transparent.

Residents donate to crowdfunding campaigns and give up unpaid evenings. Industry groups fund professional mail, text, and television programs through state affiliates. Cities absorb extended hearings, staff review, and litigation — and then, if the opposition prevails, whatever revenue the project would have produced, minus whatever costs it would have imposed, both of which are contested numbers produced largely by interested parties.

The tools work. Texts, mailers, ads, and legal filings mobilize people and force process, on both sides. They are also expensive, and the expense lands on the same communities twice. As more towns confront these proposals, the financial toll of fighting data centers — what the battles cost, and what the outcomes forfeit — is becoming a significant and largely unexamined part of the story. The honest summary is not that opposition is too expensive, nor that it is worth every penny. It is that almost nobody involved is publishing an accounting anyone can check.

# # #

---

---

Note on sources and fairness. This is a news summary compiled by Citizens United for Transparency from public reporting, official government records, court filings, and organizations’ own published statements, each linked above. It does not allege that any specific person or organization has engaged in unlawful conduct. Allegations described in lawsuits, letters, or audits are allegations only, and all parties are presumed innocent unless and until proven otherwise. Where material is reposted or excerpted from another organization, it is identified as that organization’s statement and reproduction does not constitute endorsement. Corrections are welcome and will be posted.

---

Media & tips: Contact us — we welcome confidential submissions from the public, researchers, and officials.

← Back to all news updates