← All FAQs

In depth · Noise & Infrasound

Data Center Noise and Infrasound: Frequently Asked Questions

Social-media posts have raised concerns about noise, low-frequency sound, and infrasound from AI data centers. Some of those concerns are legitimate, but several widely shared claims combine real examples with missing context or conclusions that current research does not support.

This FAQ explains what residents should know and what should be evaluated for any proposed facility near Linn Valley.

Can data centers create noise?

Yes. Potential sources include:

  • Cooling fans, chillers, pumps and air-handling equipment
  • Electrical transformers and substations
  • Emergency backup generators during testing or outages
  • On-site gas turbines, if a project includes its own power generation
  • Construction and vehicle traffic

Cooling equipment may operate continuously. Generators generally operate only during scheduled testing, power interruptions or emergencies unless a project is specifically designed to generate its own primary power.

The amount of sound reaching a home depends on the equipment, operating schedule, distance, terrain, buildings, sound barriers, weather and mitigation measures.

Do data centers expose nearby homes to 96 decibels around the clock?

A blanket claim of "96 decibels, 24 hours a day" is misleading.

Noise levels near individual machines or inside server and equipment areas can be very high. Those measurements do not establish the sound level at a property line or residence. Sound normally decreases with distance and can be further reduced through equipment enclosures, barriers, building design and operational controls.

For perspective, the National Institute for Occupational Safety and Health recommends limiting workplace exposure to an average of 85 dBA over eight hours. A sustained residential exposure of 96 dBA would be unmistakably loud and potentially hazardous, not "noise you can't hear."

Any useful noise figure must identify:

  • Where the measurement was taken
  • Its distance from the equipment
  • Whether it was measured in dBA, dBC or another scale
  • Whether it was an instantaneous maximum or a time average
  • Which equipment was operating
  • Whether the measurement represents a home, property line, equipment enclosure or workplace

Without that information, a decibel number cannot reliably describe community exposure.

What is low-frequency noise?

Low-frequency noise generally refers to sound concentrated toward the lower end of the audible range. It may be experienced as a hum, rumble, pulsation or vibration and can sometimes travel through structures more readily than higher-frequency sound.

A single overall dBA reading may not fully describe a noise problem dominated by low frequencies. A thorough study may also use dBC or unweighted measurements and analyze individual frequency bands.

What is infrasound?

Infrasound is sound below approximately 20 hertz, near or below the conventional lower boundary of human hearing.

"Inaudible" does not mean "unmeasurable." Appropriate sound-level instruments and frequency-analysis equipment can measure infrasound. Standard A-weighted meters are intentionally less sensitive to low frequencies, so investigators may need different settings and instruments when low-frequency sound is a concern.

Whether people can hear or feel infrasound depends on both its frequency and its sound-pressure level. There is no single decibel line separating all audible sound from all inaudible sound.

Can infrasound cause headaches, nausea, insomnia or anxiety?

Residents near various industrial noise sources have reported headaches, sleep disruption, nausea, dizziness, anxiety and other symptoms. Those reports should be taken seriously.

However, reports of symptoms do not by themselves prove that data-center infrasound caused them. A systematic review found some associations between low-frequency noise and annoyance, sleep-related problems, concentration difficulties and headaches, but concluded that the evidence, particularly for chronic medical conditions, was limited and affected by methodological weaknesses.

Research specifically connecting measured residential exposure from data centers to particular illnesses remains limited. It is therefore inaccurate to state as an established fact that ordinary data-center infrasound is making nearby residents sick.

This does not mean noise is harmless. Excessive environmental noise can disturb sleep, cause annoyance and affect health. It means exposure must be properly measured and evaluated before causation can be determined.

Can a decibel meter miss low-frequency sound?

A meter can understate low-frequency sound if it uses only A-weighting or lacks the necessary frequency range. That is a limitation of the measurement method, not proof of an undetectable hazard.

A competent investigation can use:

  • A-, C- and Z-weighted measurements
  • One-third-octave or narrow-band frequency analysis
  • Daytime and nighttime monitoring
  • Measurements inside and outside affected homes
  • Background measurements before equipment operates
  • Records showing which equipment was running
  • Monitoring under different weather conditions

Statements that "official meters read normal because infrasound cannot be detected" are incorrect.

Can backup generators reach 105 decibels?

Some large generators can produce sound levels around or above 100 dBA close to the equipment. That does not mean a home receives the same level.

The relevant questions are the sound level at the property boundary and nearby residences, how long the generators operate, whether they are enclosed, and what silencers or barriers are installed. Generator testing schedules and simultaneous operation should also be considered.

Are gas turbines really "jet engines bolted to the ground"?

Some large data-center projects have purchased aeroderivative gas turbines. These machines use technology derived from aircraft engines, so the comparison has a factual basis.

However, "jet engines bolted to the ground" is rhetorical language. Stationary generating turbines are packaged power-generation systems that may include exhaust controls, acoustic enclosures and other equipment.

One prominent Texas project reportedly acquired 29 turbines. That is a specific, unusually large project, not evidence that every AI data center uses 30 turbines. Whether a Linn Valley-area proposal includes gas turbines must be determined from its actual plans and permits.

Is it true that 47% of Americans oppose a data center near their home?

Yes, with context. A Redfin-commissioned survey conducted by Ipsos in November 2025 found that 47% of respondents opposed construction of an AI data center near their home, while 38% supported it.

Other surveys have found even greater opposition. A more recent Redfin-commissioned survey (May 2026) found opposition had risen to 53%, with 34% in support, and a Gallup poll likewise found more Americans oppose a data center in their area than support one. Polling measures public opinion; it does not establish noise exposure or health effects.

Do some communities impose data-center moratoriums?

Yes. Some local governments have delayed or restricted projects while reviewing zoning, power, water, noise, environmental and infrastructure issues. Other communities have approved projects with operating conditions and mitigation requirements.

A moratorium is a policy decision. It is not scientific proof that every data center creates a health hazard.

What should a credible noise study include?

For a proposed facility near Linn Valley, residents should expect an assessment based on the actual site design, not generic numbers copied from another project.

A strong assessment should address:

  • Existing daytime and nighttime background sound
  • Normal operations at full expected capacity
  • Cooling and electrical equipment
  • Generator testing and emergency operation
  • Any proposed on-site turbines or power generation
  • Low-frequency and tonal sound
  • Property-line and residential receptors
  • Simultaneous operation of multiple noise sources
  • Seasonal vegetation and weather conditions
  • Construction noise
  • Proposed mitigation and verification testing
  • A process for receiving and investigating complaints

Post-construction monitoring can determine whether predictions were accurate and whether additional mitigation is necessary.

What protections can reduce community noise?

Depending on the project, protections may include:

  • Greater distance from homes
  • Acoustic walls, berms and equipment enclosures
  • Quieter cooling technology
  • Low-noise fans and variable-speed equipment
  • Exhaust silencers
  • Limits on generator-testing hours
  • Restrictions on simultaneous generator testing
  • Property-line and nighttime noise limits
  • Special standards for tonal and low-frequency noise
  • Independent compliance monitoring
  • Public reporting of test results
  • Clear complaint-response and enforcement procedures

Several of these are already written into Linn Valley's draft ordinance, including noise limits at the nearest home, setbacks from homes, and a grid-connection requirement that limits generators to backup and testing. See the City's full response on noise.

The bottom line

Data-center noise is a legitimate planning issue. Cooling systems, generators, transformers and on-site power equipment can affect nearby communities if a facility is poorly located, designed or operated.

At the same time, viral graphics claiming that all AI data centers expose nearby homes to continuous 96 to 105 dB, undetectable infrasound and proven illness go beyond the available evidence.

The responsible approach is to require transparent project information, site-specific acoustic analysis, appropriate low-frequency measurements, enforceable limits and verification after operations begin.

Last reviewed: July 29, 2026

This page provides general educational information. Conclusions about any particular development should be based on that project's plans, permits, equipment and site-specific studies.

← Back to all FAQs