Residents living near AI data centers in the US are experiencing dizziness, nausea, sleep disruption, and other symptoms linked to a continuous low-frequency hum emanating from data center cooling systems and backup turbines, raising new concerns about noise pollution and health impacts of AI infrastructure.
- Infrasound from AI data centers causes headaches, nausea, and sleep issues.
- Cooling systems and diesel/natural gas turbines contribute to noise levels near 100dB.
- Community opposition grows over health effects and environmental concerns.
What happened
People living near several AI data centers across the US have reported a range of symptoms including dizziness, nausea, vertigo, insomnia, and anxiety that are attributed to continuous exposure to low-frequency sound waves, or infrasound. These sounds, which are often felt rather than heard, originate from the cooling equipment and backup power generators essential for the operation of these centers. Noise levels in some locations approach 100 decibels around the clock, complicating residents' quality of life.
The problem is exacerbated by the vast heat output of GPU-powered AI data centers, which require significant energy input for cooling systems that can consume up to 40% of the total electricity of the facility. Additionally, diesel and natural gas-powered turbines used as backup or primary power sources generate loud, persistent noise similar to jet engines, sometimes audible miles away. These combined factors create a persistent hum that has proven difficult to mitigate or block, leading to community backlash and halted proposals for new data centers.
Why it matters
The emergence of infrasound-induced health issues highlights an underappreciated environmental impact of rapidly expanding AI infrastructure. While noise pollution from industrial sources is not new, the inaudible yet physically disruptive nature of infrasound means that traditional noise regulations and monitoring methods may be insufficient to protect public health. The symptoms reported by residents align with scientific findings that exposure to high levels of infrasound can affect human cardiac function and neurological well-being in short timeframes.
This situation also adds another layer to ongoing debates about the sustainability and ethical considerations of AI development. Communities facing continuous exposure to debilitating noise must now contend with both the environmental footprint and direct health consequences of AI data centers. The documented link between infrasound and illnesses, along with parallels to phenomena like Havana Syndrome, underscores the urgency for regulators, operators, and policymakers to address noise pollution in their environmental and public health frameworks.
What to watch next
Monitoring and regulation of noise levels, particularly infrasound frequencies below human hearing, will become a key area to observe as local governments respond to increased community complaints and safety concerns. New standards for noise auditing and limits on operational practices for cooling and power generation equipment may be developed. Grassroots activism and local opposition are already influencing data center planning and could impact future siting decisions for AI infrastructure.
Industry responses could include innovations in quieter cooling technologies, shifts to cleaner and less noisy power sources, or proactive community engagement to mitigate health impacts. Additionally, medical and environmental research into long-term effects of infrasound exposure will shape future regulatory and operational approaches, potentially establishing clear guidelines to balance AI expansion with resident well-being.