The Data Center Debate

Arguing for

The case for data centers

Eight arguments supporters make, each with the evidence behind it, the best response from the other side, and the reply. Excerpted from the handbook, which is written as a guide for arguing a side, so the voice is 'you'.

This page presents one side as its own advocates would. The other side's response is included under every point, and the opposite page does the same in reverse. Numbers in superscript link to the source list, which says who paid for each document.

1 These buildings get built no matter what. The only question is where.

Last updated October 12, 2026 by the daily research pass. Tracks: AI race and national security, Laws, bills and regulation.

The claim in one line

Blocking one in your county does not stop it existing. It moves it, along with the tax revenue and the jobs, somewhere else.

What it means

The demand for computing comes from customers worldwide, not from your zoning board. Turning down a project does not reduce how much gets built. It reduces how much of it happens under your rules and on your tax rolls.

The evidence

The federal government has adopted this position formally. The national AI plan treats this infrastructure as central to both the economy and national security, with more than ninety actions to speed construction.2 A companion executive order instructs agencies to exempt qualifying projects from the standard federal environmental review, put them on an accelerated permitting track, and open federal land for them.3

The AI race version of this point, with the mechanics spelled out

How advanced an AI system can be is set by how much computing power trained it, and computing power is not an idea, it is racks of chips in a physical building drawing city-scale electricity. Whoever has more of those buildings trains stronger models faster and can put AI to work across more of its economy and military. Epoch AI, which measures global computing, finds about three-quarters of the world's GPU cluster performance sits in the United States, with China second at about 15 percent, measured by where the hardware physically sits.47 A separate facility tracker identifies 43.6 gigawatts across 84 US sites against 1.1 gigawatts across seven in China, though it sees far less of China than of the US. Even the pro-buildout advocacy groups frame it starkly: of the four pillars of AI strength, energy, compute, talent, and adoption, compute is the one pillar America still clearly leads.30 U.S. export controls keep the best chips out of China's hands, but a chip only becomes capability once it is installed in a powered building. The data center is where America's chip advantage gets converted into something usable. That is why security analysts argue frontier facilities need to be on U.S. soil rather than built overseas.30

The honest caveat to carry with it

the race framing assumes AI capability keeps scaling with compute, which it has so far, and it concedes the other pillar: China's electric grid is growing far faster than America's, which turns every argument about permitting and power into an argument about the race itself.

How this side says it

"The only thing a county actually controls is whether this gets built inside its borders, under its rules, paying its taxes. Not whether it gets built."

A caution for this side

This argument can come off as a threat. Frame it as a choice about control, not as take it or lose it.

What the other side says back

"That is what gets said in every race to the bottom. It is an argument for a floor of standards, not for having none. And if this infrastructure is as valuable and strategic as you claim, it can afford to be built properly."

The reply

Agree with the standards, out loud, immediately. Your argument was never "no rules." It is about who gets to write them: the only rules that will ever apply to this building are the rules of the place that hosts it. A county that says no gets no revenue and writes no rules. Say: "You want it built properly. So do I. The community that hosts it is the only one that gets to define properly. Walking away is how you end up with someone else's definition."

What has changed since the handbook

  • October 7, 2026 Helps the case for Google's Oct. 6 deal with Constellation funds 890 MW of nuclear uprates across 11 units in three states under a 20 year contract, the largest hyperscaler nuclear commitment to date. It is cited by supporters as evidence that AI driven demand can finance new firm clean capacity rather than only strand ratepayers with gas and coal.
  • October 6, 2026 Both sides Alabama Senate Bill 270 took effect October 1, requiring the Public Service Commission to find that any 150 MW-plus data center contract makes the facility pay all incremental utility costs and produces benefits for existing customers, while still allowing heavily redacted contracts and no mandatory public hearing.
  • October 3, 2026 Helps the case against The US Senate failed on October 1 to invoke cloture on the House passed Ratepayer Protection Act, 57 to 43, leaving the push to require state regulators to consider making large loads pay for their own infrastructure stalled before the midterms. Opponents said the bill made consumer protections optional; supporters said households will keep absorbing data center driven grid costs without it.
  • October 2, 2026 Both sides The federal push to legislate ratepayer protection against data center cost shifting stalled on September 30 when the Senate failed to advance the Ratepayer Protection Act in a 57 to 43 vote. The outcome keeps the issue almost entirely in state hands for now, with Democrats arguing the House passed bill was too weak and some Republicans preferring a stronger alternative.
  • October 1, 2026 Helps the case against The Senate's 57 to 43 procedural vote fell short of the 60 needed to advance the Ratepayer Protection Act, which passed the House 417 to 3 and would have required utilities to design rates so data centers of 100 MW or more cover full generation, transmission, and distribution costs. Senator Heinrich said he blocked it as too weak and that stronger language is being negotiated into separate permitting legislation.
  • October 1, 2026 Helps the case for Amazon's 20 year, 690 MW power purchase agreement with Constellation at Calvert Cliffs, which enables about $3 billion of Maryland investment and roughly 190 MW of new emission free uprate capacity by 2032, is the latest in a run of hyperscaler nuclear deals that supporters cite as evidence domestic AI compute can be matched with firm, low carbon generation.
  • September 23, 2026 Helps the case against California became the first state to enact a comprehensive package of data center laws covering ratepayer cost allocation, energy reporting, water review and exemptions, while Texas expanded its permitting pause to include environmental permits at TCEQ. Both actions were taken by governors in different parties on the same weekend.
  • September 22, 2026 Helps the case against Texas broadened its data center moratorium on September 21 to include environmental permits, marking the fourth executive action in four months and expanding the state's role beyond grid interconnection into water and air permits.
Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 2own account, 3own account, 30paid for, 47

2 The tax money is real, it is already proven, and it is lowering people's bills right now.

Last updated October 12, 2026 by the daily research pass. Tracks: Local property tax revenue.

The claim in one line

This is not a projection. One Virginia county has cut its property tax rate ten years in a row because data centers pay a huge share of the budget.

What it means

A homeowner's property tax bill depends on the total the county needs, divided among everyone who pays. Add an enormous new taxpayer and everyone else's share goes down. That is not theory. It shows up on the bill.

The evidence

In Loudoun County, Virginia, data centers produced roughly $1.2 billion in property tax revenue in fiscal year 2026, about 39 percent of the county budget and nearly half of all property tax collected. Over that same decade the county cut its property tax rate every single year, from $1.145 per $100 of assessed value in 2016 down to $0.805 in 2026. The average homeowner's bill went down year over year even though home values went up.4

How this side says it

"One industry is covering a share of the cost of schools, deputies, and roads that would otherwise land on families. That is not a forecast. That is ten straight years of tax cuts you can look up."

What the other side says back

Three attacks, none of them on the numbers. "Loudoun is the single most favorable example on earth, so it proves nothing about anywhere else. Thirty-nine percent of a county budget depending on one industry is not a success story, it is a household living on one paycheck. And Loudoun's own residents are among the loudest opponents in the country, so the tax discount is not even buying consent."25

The reply

Concede that Loudoun is the extreme case, then point out you do not need the extreme: a fraction of that revenue still changes what a county can afford. On concentration, that is an argument for managing the ratio, not refusing the taxpayer. Counties manage exactly this with reserve funds and caps, the way a family banks a bonus instead of building a lifestyle on it. On the residents: their complaints are about siting next to neighborhoods, not about the tax rate, and notice that in ten years of public hearings nobody has petitioned to give the money back.

What has changed since the handbook

No dated note yet. The daily research pass writes one here when new evidence materially strengthens, weakens or changes this point.

Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 4own account, 25funder unknown

3 The jobs argument, told honestly. And the construction workers are not a footnote.

Last updated October 12, 2026 by the daily research pass. Tracks: Jobs.

The claim in one line

Do not claim thousands of permanent jobs. Claim what the research actually found, plus the very large and very real construction workforce.

What it means

If you overstate the permanent jobs, you will be corrected using a real study and you will lose the round. So use the study yourself, first, and get credit for honesty. Then pivot to construction, where the numbers are genuinely large and where the people making the argument are unions, not companies.

The evidence: permanent jobs

Two Brookings economists compared roughly 1,500 built facilities against 52 projects that were announced and then canceled, which is a clever way to compare similar places instead of comparing boomtowns to nowhere. Counties that got their first major data center saw computing employment rise 56 percent and telecom employment rise 43 percent over a decade, which works out to roughly 100 to 200 permanent jobs. Home values rose 2 to 5 percent.6

The evidence: construction

Local 26 of the electricians' union, covering Washington, DC, Maryland, and Virginia, reports its members worked 28 million hours last year, with data centers accounting for at least half of that. (Two figures often attached to this, that the total is double a decade earlier and that the local took 600 new apprentices in one year, are not in the article cited here, so leave them out until you can source them.) The national construction trades federation has partnered with Microsoft and OpenAI on training programs.7

The downstream jobs question, handled honestly, because it will come up

"How many jobs does one data center create beyond its own walls?" There are three layers, and they are not equally solid. Layer one, inside the building: 50 to 200 permanent, measured.6 Layer two, the supply chain: the industry's own economic study, by PwC for the Data Center Coalition, estimates each direct job supports 6.5 more across construction, utilities, logistics, and services, a 7.5 multiplier, putting the industry at 4.7 million supported jobs nationally.31 Use it with its label on: industry-funded, produced by economic modeling rather than measurement, counting gross activity rather than net new jobs. The measured county-level result, Brookings, is far smaller, which tells you most of that multiplier lands somewhere other than the host community. Layer three, the jobs created by the computing itself, the AI companies, products, and productivity running on it: no honest number exists per building, and you should say so before your opponent does. The compute is national and interchangeable; nobody can attribute a San Francisco startup's payroll to one building in Ohio. So make layer three an analogy instead of a number: nobody ever measured the electric grid by counting the jobs inside the power plant. Power plants employ few people too. The last century's economy ran on them anyway. If AI matters, this is its grid, and "how many people work in the building" is the wrong yardstick for a piece of infrastructure.

And know the counter to that analogy, because it is a good one

if the benefit is national and diffuse, the cost is local and concentrated. The host community gets the noise, the water draw, and the power bills, while the downstream jobs land in tech hubs somewhere else. That is not an argument against the grid. It is an argument that the host community deserves to be paid like a host, which loops back to taxes, standards, and binding terms.

How this side says it

"An apprenticeship is a credential that outlasts any one job site. And notice who is making this argument. That is organized labor fighting for its own members, not a company press release."

What the other side says back

Your own study, used against you. "One hundred to two hundred permanent jobs, next to Ohio giving up $1.6 billion in one year. The same study found wages did not rise at all, and home prices up an extra 2 to 5 percent over the decade is a cost for renters and first-time buyers. And Brookings found the subsidies flow to the wrong buildings: only 2 percent of investment in the big company-owned facilities, but 62 percent in the low-job rental ones."6 20 21

The reply

Give them the subsidy and keep the building: "Cap the exemptions. Kill them if you want. I will support that." Their own evidence says the big facilities barely need the subsidy, which means the industry stands on its own without it.21 You just took their best argument off the table at no cost to yours. Then return to what nobody disputes: 28 million union hours, 600 apprentices in a single year, credentials that outlast any one site.7 The jobs argument they are attacking is not the one you made.

What has changed since the handbook

  • October 12, 2026 Helps the case for Fresh Indiana numbers report about 18,000 trades union members on data center projects in 2026 and 12,000 new union members added in two years, with Google expecting roughly 5,000 construction workers daily across three sites and Meta's Lebanon project 4,000 at peak. These are construction jobs that last for the project, not permanent operations jobs.
  • October 8, 2026 Helps the case against A new Illinois Economic Policy Institute report puts firm modeled numbers behind a longstanding skeptics' claim. It projects $57 billion in planned Illinois data center investment through 2035 would create about 121,000 total jobs across the economy but only 2,800 direct permanent positions, a ratio of roughly 43 modeled jobs to one permanent one.
  • September 30, 2026 Helps the case for Industry and labor rolled out coordinated jobs messaging: Blackstone, OpenAI, QTS and SoftBank launched a coalition with five building-trades unions targeting seven states, and Google's data center chief used a Fortune interview to point at hundreds of thousands of unfilled trades roles. The argument becomes harder to characterize as anti-worker where the IBEW and pipefitters are publicly on the developers' side.
  • September 28, 2026 Both sides The confirmed Alabama Delta Forge 2 announcement offers a fresh order of magnitude data point: about 1,000 peak construction jobs and about 100 permanent onsite jobs for a $3.2 billion facility, with the developer's CEO publicly acknowledging that hyperscale data centers do not create large permanent workforces relative to their power use. That both supports the construction jobs case and reinforces the critique on permanent job counts.
Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 6, 7, 20, 21, 31paid for

4 The cost-sharing problem is real, it has a known fix, and the fix is being installed.

Last updated October 13, 2026 by the daily research pass. Tracks: Who pays for the grid buildout, Laws, bills and regulation, Who holds the risk.

The claim in one line

Whether a big customer pays its own way is a question of how the referee writes the rules. Those rules are being rewritten right now, fast.

What it means

Back to the group dinner. If splitting the check evenly is unfair, you do not cancel dinner. You change how the check gets split. Regulators do this by creating a separate customer group for gigantic users, with its own rules, so their costs stay with them.

The evidence: part one

An analysis by the consulting firm E3 found no measurable evidence to date that other customers have been subsidizing data centers, found some studies where large customers actually paid more than they cost and pushed other people's rates down, and counted at least 38 new rules for giant customers created between 2018 and 2026, thirty of them in just 2025 and 2026.8 It also noted that the states with the fastest growing demand, Texas and Virginia, had among the smallest price increases, while California and New York had the largest price increases with demand falling.8

The evidence: part two

Virginia is the concrete case. In November 2025 the state regulator created a separate customer class for anyone using more than 25 megawatts. They must sign fourteen year contracts and pay for at least 85 percent of the wires capacity and 60 percent of the generating capacity they reserved, whether they use it or not, starting January 1, 2027.9

How this side says it

"You reserved the table for fourteen years, you pay for the table, even if you never show up and even if the building never gets finished. That is what Virginia just required."

Disclose this first

The E3 study was paid for by the industry's trade group. Say so before your opponent does. Then add that E3 is the same firm Virginia's legislature hired for its own independent study, and that E3 openly wrote that the risk "is not absent."

What the other side says back

"Your study was bought by the industry's lobbying group and reviewed before release, and even it admits the evidence is thin and the risk is not absent. Virginia's fix does not start until January 1, 2027, years into the buildout. Creating a fix is an admission the problem was real. And consumer groups say the regulator ducked the hardest question, who pays for the transmission lines."8 9

The reply

You already disclosed the funding, so that punch lands on air. Add that E3 is the same firm the Virginia legislature independently hired for its own study. On "admission": of course the problem was real, that is how all regulation works, rules follow problems. What matters is speed, and thirty of the 38 new rules landed in just the last two years.8 Then flip the transmission question into an offer: "Name the rule you want added and I will support it. If no rule would satisfy you, then rules were never your objection."

What has changed since the handbook

  • October 13, 2026 Helps the case for The Indiana regulator's October 7 approval of the AES Google deal in Monrovia is a worked example of a but-for allocation: Google is contractually responsible for the roughly 1.3 billion dollars in solar, storage and transmission needed to serve its site. If the structure holds, it supports industry arguments that large customers can be charged for the costs they cause without automatic ratepayer exposure.
  • October 11, 2026 Helps the case against The U.S. Department of Energy made its first FERC filing in five years to push PJM to reallocate the costs of its 6.8 GW capacity backstop so new data centers pay rather than existing households. Even an administration generally friendly to data center buildout now agrees the current cost allocation would shift costs onto ratepayers.
  • October 10, 2026 Helps the case against Nvidia-backed Firmus pulling a $5 billion IPO on October 9 after investors balked at the valuation and share release terms is the first clear sign in this debate that public-market investors are pricing in bubble risk around AI data centers. It does not alter who pays if a US project fails, but it shows the operator-side risk is now visible, not hypothetical.
  • October 10, 2026 Helps the case against The Arizona attorney general's October 8 request that regulators investigate whether APS's Power Supply Adjuster allocated about $587 million of large-user costs to other customers from 2023 to 2025 moves what was an advocacy analysis into a regulatory question. APS disputes the figure and no commission finding has been made.
  • October 9, 2026 Helps the case against Duke Energy and all major hyperscalers filed a proposed large load tariff settlement in North Carolina on October 7 to recover costs from customers using 50 megawatts at least 80 percent of the time. The deal still needs commission approval, and consumer and environmental groups say it leaves cost gaps.
  • October 8, 2026 Helps the case for Pennsylvania's HB2828 moved unanimously out of the House Consumer Protection, Technology and Utilities Committee on October 6. It would make data center developers apply for electric service and pay infrastructure fees before seeking zoning, strengthening the case that ratepayer protections can be codified rather than left to utility tariffs.
  • October 6, 2026 Helps the case against The Minnesota attorney general's October 1 PUC filing on the Xcel-Google Pine Island deal is the first state-level official estimate putting a specific nine-to-ten-figure cost-shift risk on a single data center ESA, warning of about $1 billion in exposure for other customers if the deal is approved as filed.
  • October 6, 2026 Helps the case for Oracle's October 2 proposal to subscribe to 125-250 MW from Point Beach and absorb what it describes as $300 million in rising fuel costs for We Energies customers is a concrete, if PSC-contingent, example of a hyperscaler agreeing to carry generation costs that would otherwise sit with ratepayers.
Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 8paid for, 9own account

5 The grid has far more room than people think, if the load will bend.

Last updated October 10, 2026 by the daily research pass. Tracks: Flexible load.

The claim in one line

We are not asking the grid to grow into this. We are asking data centers to step back during the handful of hours a year when the grid is strained.

What it means

The highway built for the worst traffic jam of the year is empty most of the time. A customer who agrees to stay off the road during rush hour can use all that empty capacity the rest of the year, and nobody has to build new lanes. New lanes are what costs everyone money.

The evidence

Researchers at Duke University found that roughly 100 gigawatts of new demand, an enormous amount, could be added across the grid regions serving about 95 percent of the country using the equipment already in the ground, as long as those customers accept short, occasional cutbacks.11 This is not just a paper anymore. A utility research institute is running live tests at nine sites across the U.S. and Europe with 49 participants, including Google, Meta, Microsoft, Oracle, NVIDIA, utilities, and grid operators.12

How this side says it

"We are not asking anyone to build a bigger grid. We are asking one customer to sit out a few hours a year, and in exchange we get room for a hundred gigawatts."

A caution for this side

today this is voluntary. If they challenge you to make it mandatory, say yes. Defending voluntary is a losing position and conceding costs you nothing.

What the other side says back

"Your hundred gigawatts only exists if data centers actually ease off at peak hours, and today that is voluntary. Most contracts do not require it. A promise to maybe step back is not grid capacity."11

The reply

One word first: "Yes." Then: "Make it mandatory. A binding condition of connecting to the grid. I will say it before you finish asking." Defending the voluntary status quo is the only way to lose this exchange, and conceding turns their attack into a shared policy you both now support. Then show it is already moving from paper to practice: live tests at nine sites across the U.S. and Europe with 49 participants, including Google, Meta, Microsoft, and the grid operators themselves.12

What has changed since the handbook

  • September 18, 2026 Helps the case for Nvidia, Google and Emerald AI launched the AI Energy Management Alliance on Sept 17, with Google saying it already has more than 1 GW of demand response capacity and Nvidia backing a 100 MW power-flexible AI facility in Manassas, Virginia. It is the largest industry commitment to date around scheduling AI workloads or discharging batteries when the grid is tight.
Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 11, 12own account

6 This industry has actually gotten more efficient before, and the scary forecasts have been wrong before.

Last updated October 12, 2026 by the daily research pass. Tracks: Efficiency and forecasts, Water use.

The claim in one line

Ten years ago experts predicted data center energy use would double, triple, or quadruple. It did not. It stayed roughly flat while the amount of computing exploded.

What it means

Efficiency improvements, plus moving computers out of thousands of wasteful small server rooms into a few well-designed giant ones, kept energy use nearly flat for years. Worth remembering that the confident alarming forecast has a track record here, and the track record is bad.

The evidence

A peer-reviewed paper in the journal Science showed that widely repeated predictions of data center energy doubling or quadrupling were simply wrong, because efficiency gains held global use near one percent of world electricity while computing output grew enormously.13 On water, Microsoft reports it put back more freshwater than it took in during 2025, five years ahead of its own target, cut data center water use roughly 90 percent compared to its earliest buildings, and says its newest facility designs greatly reduce cooling water.14 The stronger versions of this claim, no water at all and a matching Google pledge by 2030, are not in that source. And on the buildout itself, an analyst at ITIF notes that announced projects consistently overstate what gets built: deals fell more than 40 percent in late 2025, and only about a third of announced projects are actually under construction.15

How this side says it

"The technology that was evaporating water to cool itself is being replaced by closed-loop systems that use none. The problem they are describing is one the industry is already engineering out."

A caution: this is the weaker point

the Science paper covers roughly 2010 to 2018, and those savings are largely spent. Use it as a reason for humility about forecasts, never as a claim that energy use is not rising now. It is rising, and the same authors are the ones documenting it.

What the other side says back

"That paper covers 2010 to 2018. The savings came from moving computers out of thousands of wasteful server closets into a few efficient giant buildings, and you can only do that once. It has been done. And the same authors now publish the federal projections showing use climbing from 192 TWh toward 578 to 782 by 2030."1 13

The reply

Concede the dates before they say them, and use the history for exactly one thing: humility about confident scary forecasts, because the last one was wrong for a reason. Engineers were working the problem, and they still are. Then move to the present tense, where you have fresher evidence: announced projects wildly overstate what gets built, deals fell more than 40 percent in late 2025, only about a third of announcements are actually under construction15, and the newest cooling designs sharply reduce water use.14 Say: "The forecast assumes the industry stands still. It has never once stood still."

What has changed since the handbook

  • October 12, 2026 Helps the case against PJM's own market monitor now says 42 percent of the data center load it had forecast for 2026 to 2029 is already delayed or canceled, strengthening the long standing argument that large load forecasts tend to overshoot. The data does not mean demand will stop growing, only that the specific pipeline PJM planned around has shrunk by about 4 gigawatts.
  • October 7, 2026 Helps the case against Virginia DEQ documented seven sediment runoff incidents at a Google Chesterfield data center construction site between December 2025 and August 2026, affecting roughly 7,735 linear feet of stream channel and about 1.4 acres of wetlands, with some incidents not reported within required windows. The notice adds a construction phase example to the usual focus on operating water use.
  • September 25, 2026 Helps the case against The Texas attorney general opened an investigation into whether hundreds of data centers filed the water use reports state law requires, giving weight to critics who say water disclosure has been largely voluntary in practice.
Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 1, 13funder unknown, 14own account, 15paid for

7 The noise and the looks are real complaints with engineering answers, and the answers are being written into law.

Last updated October 13, 2026 by the daily research pass. Tracks: Noise, appearance and rural character, Siting, zoning and standards.

The claim in one line

A data center is a quiet neighbor compared with almost any other industrial use. Where it has been a loud one, the fix was fans and setbacks, not a ban, and communities that wrote real standards got results.

What it means

Do not wave away noise and appearance. They are the complaints that fill the room, and dismissing them is how a project loses a vote it should have won. Instead, treat them as the siting problems they are. A building with a few dozen employees generates almost no traffic, no smoke stacks, no shift changes and no trucks. What it generates is fan noise and a big flat wall, and both have known remedies.

The evidence: part one, the fix works

The worst-known case is Great Oak, next to an Amazon data center south of Manassas, Virginia, where residents measured up to 65 decibels at night.87 Amazon replaced all 424 rooftop exhaust tubes with taller ones so the air vented upward instead of toward the homes, and the noise fell by about 10 decibels, half as loud, to around 50 decibels by both the residents' and the company's measurements.88 That is what a mitigation requirement produces when it is enforced.

The evidence: part two, the rules are arriving

Chandler, Arizona has required, since January 2023, a pre-construction sound study shared with residents, mitigation so operating noise does not exceed the baseline, limits on when backup generators can be tested, neighborhood meetings and an on-site liaison.91 Loudoun County, which hosts more data centers than anywhere on earth, ended by-right data center approvals on March 18, 2025, so every new one now needs a special exception with conditions, and in September 2025 it began writing use-specific standards covering compatibility, aesthetics and infrastructure.94 Prince William County has proposed a noise limit of 52 decibels by day and 47 at night, about the sound of moderate rainfall.89

The evidence: part three, the property values

The one careful study of counties that got a data center found home values rose 2 to 5 percent more than in near-identical counties over the following decade.6 Whatever the buildings do to a view, the market did not price them as a blight.

How this side says it

"Nobody should have to live next to a hum. That is a design standard, and the counties writing them are getting quieter buildings. It is not a reason to send the tax base to the next county over."

What the other side says back

"Every fix you name came after the building did. Great Oak waited two years and still fails the county's new limit, the low-frequency roar is not covered by any rule, and the generators are exempt. Loudoun says it will not have a low-frequency standard until 2027. Chandler took nine years. And a county-wide home value average says nothing about the houses on the fence line."89 90 91

The reply

Agree that late is the problem, then make it the offer: "Write Chandler's rules into the approval before the first shovel: a baseline sound study, a mitigation guarantee, setbacks, screening, generator testing windows. I will support every one of them." A developer who accepts those conditions has answered the complaint; a county that rejects the project anyway was never arguing about noise. On the fence line, do not argue the average, argue the remedy: setbacks and berms are exactly what special exception conditions are for, and Loudoun's move to conditional approvals makes them enforceable.94

What has changed since the handbook

  • October 11, 2026 Helps the case against In the same week, developers advanced a $66 billion plan in Wyoming's Fremont County, which has no zoning code, and a $15 billion plan in Hudspeth County, Texas, where the county judge said she had very little advance information. The two cases strengthen the argument that the biggest projects are migrating to jurisdictions with the weakest local review.
  • October 2, 2026 Both sides Two local votes on October 1 showed the split in how counties are handling moratoriums. Forsyth County, North Carolina rejected a pause after its attorney warned a Chatham County developer had successfully challenged a similar moratorium, while Mercer County, Kentucky unanimously killed draft zoning that would have allowed hyperscale sites near a historic landmark.
  • September 30, 2026 Both sides Saginaw's new ordinance, adopted 8-0 and effective October 9, illustrates a middle-path approach: data centers are allowed but only in industrial districts, at least 500 feet from homes, with a 55-decibel noise cap, mandatory water and energy disclosures and two neighborhood meetings before approval.
  • September 29, 2026 Helps the case against A new Washington Post satellite-image analysis reports that only about 40 percent of large US data centers were built on previously developed land, with about a third on former farmland, 12 percent on former forest and 11 percent on shrubland. This strengthens the case for tighter siting rules on greenfield conversion.
  • September 27, 2026 Helps the case against The Maryland executive order requires any data center at 25 megawatts or larger that seeks any state action to be scored against five public criteria and posted on a monthly Data Center Dashboard, setting a floor for state review that mirrors calls in Virginia and California to end approvals with no state visibility.
  • September 24, 2026 Helps the case against Prince William County, one of the largest existing data center markets in the country, voted 8 to 0 on September 23 to end by-right data center approvals and shrink its overlay district from about 9,700 to roughly 3,641 acres. Future projects outside the reduced overlay now require a special use permit.
  • September 23, 2026 Helps the case against Texas broadened its data center pause to include all TCEQ environmental permits, meaning water, air and land use approvals now sit on hold statewide until ERCOT and Texas Water Development Board audits are complete. That extends the moratorium beyond grid interconnection into the full permitting stack.
  • September 19, 2026 Helps the case against Virginia's executive order proposes ending by-right approval and requiring local government sign-off for any data center using more than 25 megawatts, and directs an accelerated noise rule and water-use standards. Loudoun County advanced a 12-month pause on new data center and substation applications on September 16.
Latest evidence tagged to this argument

Tagged automatically by the daily update from the updates feed. New evidence does not rewrite the argument above; read both and judge whether the point still stands.

Sources cited in this point: 6, 87, 88, 89, 90own account, 91own account, 94

8 When an investment like this lands in a parish like that one, "hold out for a better offer" is not a real answer.

Last updated September 24, 2026 (handbook revision; no newer evidence yet).

The claim in one line

The biggest data center in the country is going up in one of the poorest parishes in Louisiana. The construction money, the contracts and the tax base are already arriving. Telling that community to wait for something cleaner is easy to say from somewhere that has options.

What it means

Rural counties do not get a menu. They get an offer every decade or two, and most of those offers are a warehouse or a prison. When the offer is tens of billions of dollars of private capital, the honest question is not whether it is perfect. It is what the alternative on the table actually is, and usually there is not one.

The evidence

Richland Parish has about 19,800 residents, a poverty rate of 24.1 percent against 12.5 percent nationally, and a median household income of $53,544, roughly two thirds of the national figure.67 Meta's project there was announced at $10 billion and raised to more than $50 billion by July 2026.60 The state reports more than $1.6 billion in contracts already awarded to Louisiana businesses, 7,500 construction workers at peak and about a thousand permanent operating jobs.60 Meta reports that construction driven sales tax funded parish teacher bonuses of up to $50,000, and has committed $215 million to bill assistance and efficiency programs.61

How big is it, really

A thousand permanent jobs in a parish of 19,800 is a different proposition than a thousand jobs in Northern Virginia. Say it carefully, because the easy version of this line is wrong: a thousand jobs against 19,800 residents is about one job for every twenty people who live there, which is not an employment rate. Population counts children and retirees, so the share of the actual workforce is larger, and if you call it an employment figure someone will correct you. It is a scale comparison, and at that scale it is genuinely unusual. It also only works where the denominator is small. Do not carry it into a suburban county and expect it to survive.

How this side says it

"Name the other fifty billion dollar offer this parish turned down. I will wait."

Why this lands

It moves the fight off national averages, where your side tends to lose, and onto a specific place where the numbers are lopsided in your favor. It also forces the other side to argue against a poor community's own choice, which is uncomfortable ground.

What the other side says back

"Those are the company's numbers, the utility's numbers and the numbers from the state agency that recruited the project.60 61 62 The plants built to serve it run 30 to 40 years against a contract of 15 to 20, and intervenors put roughly $3.4 billion of plant cost still unpaid when that contract ends.63 When the commission's own judge ordered the jobs math disclosed,65 the commissioners overruled him 3 to 1.76 A good deal does not need to be sealed."

The reply

Concede the disclosure point, because it is correct and defending secrecy makes you look like you have something to hide. Say: "They should publish it, and I will say so on the record." Then keep the ground you actually hold: the construction contracts and the tax receipts are already landing and are verifiable locally, whatever happens to the generation contract in 2045. Argue for the protection rather than against the project: the leftover plant cost belongs on the customer that caused it, and that belongs in the contract in writing.

Sources cited in this point: 60own account, 61own account, 62own account, 63own account, 65, 67, 76

Answering the other side's opening points

When they are on offense with their own six points, your answers are already written: each of their points in Part 3 ends with a "what they will say back" block, and that block is you. Those are your lines, sourced and ready. Read Part 3 the way they will read Part 4 and you have the whole exchange from both sides. Two of their openers to drill, because they open with them constantly:

Their Harvard opener (ratepayers are subsidizing Big Tech): your answer is under their Point 1. The short version: it is a serious legal argument, not a measurement, and the study that measured found none.8 16 Then concede the secrecy immediately and support public contracts. It costs nothing and takes away their applause line.

Their watchdog opener ($6.3 billion in one auction): your answer is under their Point 2. The short version: the dinner distinction, delivered slowly. Who caused the cost is not disputed. How the check gets split is what the 38 new rules and Virginia's customer class govern.8 9 Never dispute the number itself.

Questions this side asks the other

  1. "You cite Virginia's study. That study found 74,000 jobs and $9.1 billion a year. Do you accept the whole finding or only the half you quoted?"22
  2. "You say regular customers are subsidizing data centers. Name the study that measured it. Harvard read filings and made a process argument. Which study put a number on it?"16 8
  3. "Texas and Virginia had the biggest demand growth and among the smallest price increases. California and New York had the biggest price increases with demand falling. What is your explanation?"8
  4. "If every tax exemption were eliminated tomorrow, would you support these projects? If not, the tax break was never really your objection."
  5. "Your own opposition tracker is funded by a firm that works for AI companies. You trust their numbers. Why not their framing?"25

Where this side is weak, in its own words

Every honest case has soft spots. These are the places this side's own handbook warns not to overclaim.

  • Never claim large permanent employment. The honest ceiling is roughly 100 to 200 jobs per county.6 Overclaim and you lose the whole round on credibility.
  • Do not dispute the $6.3 billion figure. It comes from the independent market watchdog, not from activists.17 Reframe it using the dinner distinction. Denying it makes you look like you are hiding something.
  • Your best cost study was paid for by the industry's lobbying group and reviewed by them before release.8 Disclose it first, every time. Then note that E3 is the same firm Virginia's legislature hired independently, and that E3 itself wrote that the risk "is not absent."
  • Your flexibility argument is conditional and you should concede that immediately.11 Most contracts do not require it today. Offer to make it mandatory rather than defending the status quo.
  • Virginia's protections do not start until January 1, 2027.9 Concede the delay. Arguing otherwise is a losing fact fight.
  • The efficiency paper is about 2010 to 2018.13 Those gains were largely spent before AI arrived, and its own authors now publish the projections showing steep growth.1 Use it for forecasting humility only.
  • Do not dismiss noise as a nuisance complaint. Great Oak measured 65 decibels at night, the fix took two years, and the low-frequency hum is still not regulated.87 89 Offer the standards before you are asked.
  • Do not dismiss the Memphis case.24 Hundreds of residents spoke and not one supported it, in a place already failing federal air standards. The right answer is that this is exactly what siting standards are for, not that it did not happen.
Now read the case against Get the full handbook