Hot topics

Data Center Noise Pollution

How loud data centers actually run, why residents' lawsuits keep hinging on noise standard measuring tools don't detect, and what the real (if still thin) research says about the effect on nearby wildlife and ecosystems. Plainly, with sources.

Last updated September 29, 2026

Terms

dBA
An A-weighted decibel — the standard noise measurement, adjusted to how the human ear perceives pitch. It's what most municipal noise ordinances enforce, and it's also the source of the biggest dispute here: it systematically undercounts low-frequency sound.
Infrasound
Sound below about 20 Hz, at or under the threshold of conscious human hearing. It can still be physically felt as vibration or a low hum, and standard dBA noise meters are poorly suited to registering it at all.
Nuisance law
The legal theory behind most current data center noise lawsuits: that a facility's noise unreasonably interferes with neighbors' use of their own property, regardless of whether it violates a specific decibel limit.

How loud, and from what

The noise comes from cooling fans, chillers, and compressor arrays running continuously, plus backup diesel generators and — increasingly — on-site natural-gas turbines brought in to power AI facilities faster than the grid can connect them. Reported levels vary widely by source and distance: a persistent hum in the 40–59 dBA range is common near Virginia's data center corridor, some facilities have been measured near neighboring homes at up to 96 dBA continuously, and backup generators or turbines can reach about 105 dBA — comparable to a jet flying overhead. For comparison, the EPA's long-standing guideline for outdoor noise is a 24-hour average of 55 dBA, or 45 dBA indoors.

Not everyone agrees the problem is that severe. The Taxpayers Protection Alliance, an industry-aligned group, points to a 2024 Virginia legislative review comparing data center noise to "ambient library sounds or a normal conversation." That framing describes routine ambient hum at a distance; it doesn't address the specific complaints below, which are about noise measured much closer to occupied homes, or about frequencies standard meters don't register well at all.

The infrasound problem

The recurring legal and scientific problem is that data center cooling equipment — massive chillers, slow-turning fans, compressor arrays — generates long-wavelength, low-frequency noise that standard A-weighted decibel meters aren't built to measure well, even when residents can clearly feel it. Near AI and bitcoin facilities in Granbury, Texas and Southaven, Mississippi, residents living within about 100 yards of the equipment have reported vertigo, nausea, migraines, elevated blood pressure, and insomnia, alongside measured noise as high as 96 dBA. In Michigan's data center corridor, one resident described it as a stuck vacuum cleaner left running: "the filter is clogged up, so it's a high-pitch whining. And they just left it on and walked out."

The lawsuits

At least six noise-related lawsuits against data center operators were filed in 2026, mostly built on private and public nuisance and negligence claims rather than a specific decibel-limit violation — precisely because standard monitors struggle to document the low-frequency noise at the center of the complaints:

Microsoft (Fairwater)Mount Pleasant, WI — filed June 2026, low-frequency hum
xAISouthaven, MS — filed June 2026, gas-turbine noise/vibration/air
Unnamed data centerVineland, NJ — filed May 2026, private/public nuisance
Hyperscale digital-asset facilityDowagiac, MI — class action, noise
Data center operatorNorth Tonawanda, NY — nuisance claim
Data center operatorHood County, TX — nuisance claim

As of this writing, none of these cases has reached summary judgment or trial.

By the numbers

EPA's outdoor 24-hour noise guideline
55 dBA
Noise measured near some facilities, continuously
96 dBA
Noise-related lawsuits filed against operators in 2026
6+

EPA; EESI/Tom's Hardware reporting; Smart Cities Dive/WilmerHale legal tracking.

What it does to nearby ecosystems

The most concrete case study is in Nashville, where a proposed data center would sit near the Nashville Zoo's animal habitats, on land the zoo had hoped to use for a conservation center; nearly 400,000 people signed a petition opposing the project. University of Michigan associate professor Neil Carter, whose research covers how noise and light affect bird reproduction in species like the northern cardinal and barn swallow, has described data centers as potential "sensory danger zones" — places where noise and light exceed the thresholds at which wildlife's normal functioning and fitness break down.

The broader ecological research isn't data-center-specific, but it's well established and lines up uncomfortably with the decibel ranges above: terrestrial wildlife shows measurable responses starting around 40 dBA — roughly the low end of a data center's routine hum — and hearing damage risk begins around 85 dBA, within range of what's been measured near some facilities. Birds exposed to 55–61 dB of noise spend significantly more time on alert and less time feeding than birds in quieter conditions; one widely cited synthesis found that minke whales lose roughly 80% of their communication range — from over 100 kilometers down to under 20 — as background noise rises from 67.5 to 87.5 dB. Zoo-animal studies specifically found individual animals increasing vigilance and movement at higher sound levels, and elephants relocating away from continuous construction noise.

Nonpartisan, plainly

The dBA measurement gap is a real, widely documented technical problem, not just a resident complaint — multiple law firms and outlets covering this litigation independently describe the same issue: standard noise ordinances measure a frequency range that data center cooling equipment doesn't mainly operate in. Reported noise levels genuinely vary enormously by distance and source, from a background hum comparable to routine ambient sound at a distance, to measured levels near 96 dBA close to some facilities — both figures are real, they're just describing different distances and situations, and industry-aligned voices tend to cite the former while affected residents cite the latter. On wildlife specifically, we should be honest about the state of the evidence: there is essentially no large body of research measuring actual ecological damage at real data center sites yet. What exists is a well-established general science of how noise at these decibel levels affects animals, applied by researchers as a reasonable warning about data centers specifically — a prediction based on solid science, not yet a confirmed on-the-ground finding.

Talking points

These are the questions we think you should ask those who are running for office and will represent you.

  1. Should local noise ordinances be updated to measure low-frequency and infrasound noise, not just A-weighted decibels?
  2. Should a data center be required to fund independent, continuous noise monitoring at the property line for its entire operating life, not just at initial permitting?
  3. Should new data centers be required to complete a wildlife and noise-impact study before permitting, when sited near parks, zoos, or conservation land?
  4. Who should pay for soundproofing or relocation if a data center's noise is later found to exceed safe levels: the company, the local government that approved it, or the resident?

Read more

WONKRZ STORE

Upgrade Your Gear Today

Discover our latest collection of products and exclusive deals. This store supports various initiatives to keep people informed and engaged in our democracy regardless of political party.

Shop Now →