America's AI Data Center Boom Is Raising New Questions About Electricity Bills, Water and Local Communities
The buildings behind chatbots are now a kitchen-table issue. Here is what the evidence says about power prices, water, grid investment and the people who live next door — and where the claims on both sides run ahead of the data.
Why this became a local fight
A few years ago, most Americans had never thought about a data center. In the run-up to the November 3, 2026 midterms, they became a campaign issue in Ohio, Texas, Pennsylvania, Michigan and Virginia, with candidates in both parties promising to slow or condition new projects.
The reason is scale. The U.S. hosts roughly 4,767 data center sites, by far the largest concentration in the world, according to Data Center Map figures cited by the BBC. Harvard reporting earlier this year counted more than 4,000 in operation and about 3,000 more planned or under construction. Annual U.S. construction spending on data centers rose from about $15 billion in 2022 to more than $35 billion in 2025 in constant dollars, Vox reported.
The newest "hyperscale" AI campuses are not server rooms. They are industrial facilities drawing hundreds of megawatts — a single campus can need as much power as a mid-sized city — and they are increasingly built in rural counties where land and transmission lines are available.
That is the backdrop for four questions residents keep asking: Will my power bill go up? Will we run short of water? Who pays for the new grid? And what is it like to live next to one?
Question 1: Are data centers raising electricity bills?
Short answer: in some regions, yes, through specific mechanisms — but the national picture is more mixed than political ads suggest.
The Lawrence Berkeley National Laboratory's 2024 report to Congress estimated data centers used about 4.4% of U.S. electricity in 2023 and projected 6.7% to 12% by 2028. Utilities have since filed forecasts that, taken together, imply some of the fastest load growth the country has seen in decades.
How that reaches a household bill depends on the market:
| Mechanism | How it can raise bills | Where it shows up |
|---|---|---|
| Capacity markets | Grid operators pay generators to be available at peak; tighter supply forecasts push the price up for everyone | PJM (13 states plus D.C.), where recent capacity auctions cleared at record highs |
| Transmission build-out | New high-voltage lines and substations are often recovered from all customers in a zone | Virginia, Ohio, Texas, Georgia |
| New generation | Utilities build gas plants or extend old ones, then recover costs in rates | Vertically integrated states in the Southeast and Midwest |
| Spreading fixed costs | If a data center pays its full share, it can lower everyone's per-unit cost | Where tariffs require it — the case industry cites |
FactCheck.org's October 29 review is a useful corrective. Tens of millions of dollars in ads have asserted that data centers caused rising bills; Energy Secretary Chris Wright has claimed the opposite, that data centers bring multiyear price freezes. FactCheck found neither claim holds up as a blanket statement. Electricity prices have also been driven by natural gas costs, wildfire and storm hardening, aging distribution lines and general inflation. The data center share varies by utility and is still being litigated in rate cases.
What is reasonably well supported: in PJM, independent market monitors have attributed a large share of recent capacity-price increases to projected data center demand, and those costs flow to residential customers. Elsewhere, the effect depends on whether regulators make large loads pay for what they trigger.
What to look for locally: whether your state utility commission has approved a "large load tariff" — rules requiring long contracts, minimum payments, or upfront collateral from data centers so other customers are not left paying for stranded infrastructure if a project leaves. Ohio, Virginia, Georgia and others have adopted or debated such rules.
Question 2: How much water do they really use?
Short answer: it depends heavily on cooling design and on where the electricity comes from. Local impact can be serious in dry regions; national comparisons are often exaggerated.
There are two kinds of water use:
- Direct (on-site) water, mostly evaporative cooling. A large evaporative-cooled campus can use millions of gallons a day in summer peaks. Closed-loop and air-cooled designs use far less on-site water but more electricity.
- Indirect water, consumed at the power plants that supply the electricity — often larger than direct use, and much less visible to residents.
Vox's May analysis argued the national water impact has been "greatly exaggerated" compared with agriculture and other industries. That is true at national scale. It is less comforting in a specific aquifer-dependent county in Arizona or Central Texas during drought.
The bigger problem is that the data is missing. KUT reported in August that Texas officials do not know exactly how much water and power planned data centers will use, and Governor Greg Abbott effectively paused new grid connections for data center projects until regulators better understood resource demands. Many local water agreements are covered by non-disclosure clauses, so residents often learn the numbers only after approval.
What to look for locally: the cooling technology in the site plan, whether the project uses potable or reclaimed water, and whether the water agreement is public.
Question 3: Who pays for the infrastructure?
The AI build-out requires new substations, transmission lines, gas pipelines and generation. Developers and hyperscalers increasingly fund their own power — signing contracts for nuclear restarts, gas turbines, solar-plus-storage and, in some cases, building on-site generation. The Stargate campus in Abilene, Texas, is an example of a project designed around dedicated power.
But not all costs are contained inside the fence. Local roads, emergency services, and grid upgrades beyond the site are typically shared. Tax incentives matter too: many states offer sales-tax exemptions on servers and equipment, which can be worth billions statewide. Communities gain property tax revenue — often substantial, which is why some counties actively court data centers — but relatively few permanent jobs once construction ends, commonly in the dozens to low hundreds per campus.
NRDC's October 30 guidance summarizes the position of many environmental and consumer groups: technology companies should pay for the infrastructure their facilities require, bring new clean energy to meet new demand, and give communities a real say.
Question 4: What is it like to live next door?
Residents' concerns are not only about bills. Common complaints include:
- Noise. Cooling fans and backup systems produce a constant low hum. In Northern Virginia, the world's densest data center market, noise is among the most frequent complaints.
- Heat. Arizona State University researchers measuring Phoenix-area facilities found rooftop cooling units running about 15°C above surrounding air temperature, according to reporting by DongA Science, raising urban-heat questions for nearby homes.
- Air quality. Diesel backup generators and on-site gas turbines emit pollutants. Permits for on-site gas generation have drawn particular scrutiny.
- Land and transmission lines. New high-voltage lines can cross farmland and private property, sometimes through eminent domain.
- Process. Many residents say they learned about projects late, after code names, NDAs and fast-track approvals.
Public opinion has shifted accordingly. Foreign Policy, citing polling, reported that about two-thirds of Americans oppose local data center construction, and Pew Research has documented broad concern about water and power use. Vox reported at least 20 projects shelved in the first three months of 2026 alone.
The case for data centers
A fair account should include the other side. Supporters point to:
- Tax base. In counties like Loudoun, Virginia, data centers provide a large share of local revenue, helping fund schools and keeping residential property tax rates lower.
- Grid modernization. Large, steady customers can help finance upgrades that benefit everyone — if tariffs are designed well.
- Clean energy demand. Hyperscalers are among the largest corporate buyers of renewable and nuclear power.
- National competitiveness. The AI race with China is frequently cited as a reason to build domestically.
Whether these benefits outweigh costs is a local question. A site with reclaimed-water cooling, a strong large-load tariff and a fair tax deal is a different proposition from one with secret water terms, generous abatements and diesel generators near homes.
A practical checklist for residents and local officials
- Power: Has the utility commission approved a large-load tariff? Who pays for new substations and lines?
- Water: Evaporative or closed-loop cooling? Potable or reclaimed water? Is the agreement public?
- Tax: What abatements were granted, for how long, and what are the projected net revenues after services?
- Jobs: How many permanent jobs, verified by whom?
- Noise and air: What decibel limits apply at the property line? What backup or on-site generation is permitted?
- Transparency: Is the end user named? Are there NDAs with local officials?
- Exit: What happens to infrastructure and costs if the project is cancelled or downsized?
WHAT WE DO NOT KNOW YET
- How much of AI demand is real. Utility forecasts include speculative projects; some developers file in multiple places for the same capacity. Actual load growth could come in well below projections.
- The true household bill impact. Rigorous, utility-by-utility attribution studies are still scarce, and rate cases are ongoing.
- Water data. There is no consistent national reporting requirement for data center water use.
- Efficiency gains. More efficient chips and cooling could reduce per-task demand, though history suggests efficiency often increases total use.
- Policy after the midterms. State moratoriums, tariff rules and federal permitting changes could shift sharply in 2027.
Sources
- BBC News, "Data centre backlash grows ahead of US midterms," October 25, 2026 — https://www.bbc.com/news/articles/cq2e9we7w2ejo
- FactCheck.org, "Data Centers and Electricity Prices: It's Not as Simple as Voters May Think," October 29, 2026 — https://www.factcheck.org/2026/10/data-centers-and-electricity-prices-its-not-as-simple-as-voters-may-think/
- Foreign Policy, "Trump's Data Center Boom Drains Energy, Water, and Land," October 12, 2026 — https://foreignpolicy.com/2026/10/12/data-center-energy-water-land-pollution/
- NRDC, "Data Centers, AI, and the Growing Demand for Energy," October 30, 2026 — https://www.nrdc.org/resources/data-centers-ai-and-growing-demand-energy-impacts-risks-and-solutions
- KUT (Austin NPR), "Data centers use lots of water and power, but Texas officials don't know exactly how much," August 5, 2026 — https://www.kut.org/energy-environment/2026-08-05/texas-data-centers-water-power-energy-use-environment
- Vox, "Data centers could actually be good for your hometown," May 18, 2026 — https://www.vox.com/politics/488754/data-centers-ban-electric-bill-water
- Harvard Gazette, "Why are communities pushing back against data centers?" April 9, 2026 — https://news.harvard.edu/gazette/story/2026/04/why-are-communities-pushing-back-against-data-centers/
- DongA Science, "AI data centers trigger neighborhood conflicts over heat, power and water," October 2026 — https://www.dongascience.com/en/news/80471
- Lawrence Berkeley National Laboratory, "2024 United States Data Center Energy Usage Report" — https://eta.lbl.gov/publications/2024-lbnl-data-center-energy-usage-report
Editorial note: This article synthesizes public reporting and research. Figures are attributed to their sources; where evidence is contested, we say so. TheNewsRupt has no commercial relationship with any company named.
Every claim above is sourced to a document, a named person, or a record we hold. Where we could not verify a claim, we say so. Read our standards and corrections policy →