There is a clear imbalance in the research literature about data centers and noise pollution. Electricity use (and associated emissions) and water consumption of data centers have been the subject of many peer-reviewed papers, reviews, life-cycle assessments, and quantitative modeling studies for years. These cover power demand growth (especially from AI), carbon footprints, water-use intensity (direct cooling + indirect grid water), efficiency metrics like PUE/WUE, regional variations, and projections under different cooling technologies or grid mixes.
By contrast, peer-reviewed work specifically on community noise impacts from data centers (the continuous low-frequency hum from cooling systems, transformers, etc., and effects on nearby residents) is sparse. What exists is mostly:
- Engineering/acoustic modeling papers and conference proceedings (noise source characterization, propagation modeling, mitigation strategies for cooling equipment and generators).
- Occupational studies measuring high internal noise levels (often >85 dBA) for workers inside facilities.
- Exploratory health assessments, grey literature, government reviews (e.g., Virginia’s JLARC), and media/community reports of complaints about sleep disturbance, annoyance, and low-frequency effects.
- Analogies drawn from the broader environmental noise literature (sleep disruption, stress, cardiovascular risks from chronic noise).
Direct epidemiological studies linking data-center noise exposure to measured health outcomes in surrounding communities remain limited or largely absent. Multiple recent papers and reviews explicitly note this research gap: direct evidence on health impacts of data centers (including noise) is sparse, most work relies on indirect analogy or modeling, and empirical community-level studies are urgently needed as facilities proliferate.
Public debate, local complaints, lawsuits, and media coverage about the “constant low-frequency hum” have indeed intensified with the AI data-center boom, often outpacing the formal peer-reviewed science on health effects. Noise levels at property lines are frequently reported in the 40–60+ dB range (sometimes higher during generator testing), which can be noticeable especially at night, but rigorous quantification of population-level health consequences specific to these sources is still catching up.
Contact your local Member of Congress or other politicians and ask that they address this problem.
Here are some scientific papers addressing this issue.
Global data center expansion and human health: A call for empirical research
Understanding the impact of data center noise pollution
Noise Exposure Assessment in the Occupational Environment of Air-Cooled Data Centers