Are AI Data Centres Making Their Neighbours Sick? An Emerging Public Health Concern

The artificial intelligence revolution is usually presented as an extraordinary technological achievement, promising greater productivity, scientific breakthroughs and a transformation of everyday life. Yet behind the apparently immaterial world of digital information stands an enormous physical infrastructure of computer servers, cooling equipment, electrical systems and power generators. These facilities consume vast quantities of electricity and water, generate heat and produce a variety of environmental disturbances. Increasingly, people living near large data centres are reporting something more troubling: they believe that the facilities are making them sick.

An article published by A Midwestern Doctor, entitled What Is Making People Near Data Centers Sick (and What Could Fix It), raises concerns about the possible health effects of living close to these rapidly expanding industrial complexes. Residents have reported headaches, dizziness, nausea, vertigo, sleep disturbances, anxiety and other symptoms that they associate with the operation of nearby facilities. While the evidence is still developing, the complaints raise legitimate questions about whether the technological industry has moved ahead of our understanding of its environmental and public health consequences.

One possible culprit is noise pollution, particularly the persistent low-frequency humming generated by industrial cooling systems, ventilation fans, transformers and power equipment. Unlike the familiar noise of traffic or construction, some of these sounds can continue throughout the day and night. Their low-frequency components may penetrate buildings and be difficult to eliminate through ordinary soundproofing.

This is important because noise does not have to be deafening to become a health problem. Persistent unwanted sound can interfere with sleep, increase stress and diminish quality of life. Low-frequency noise can be particularly intrusive when it contains a tonal hum or vibration. Infrasound, technically defined as sound below approximately 20 hertz, is another possible component, although claims that inaudible infrasound is directly causing particular illnesses require considerably more investigation.

Residents living near data centres have described an assortment of symptoms that appear, in their experience, to coincide with the operation of the facilities. Some report difficulties sleeping, while others describe headaches, ear pressure, dizziness or a general feeling of being unwell. Such experiences should not simply be dismissed because the mechanism is uncertain. Equally, the presence of symptoms does not establish that a neighbouring data centre caused them. Proper investigation requires environmental measurements, medical assessments and comparisons with populations who are not similarly exposed.

The issue has become sufficiently prominent to attract attention beyond alternative health publications. Reports in May 2026 described complaints from American communities about persistent humming and vibration associated with data centre operations. In September, residents of Slough in Britain, where numerous data centres are concentrated, described noise problems and concerns about their deteriorating health.

Australia is not immune from these controversies. In September 2026, ABC News reported complaints from residents living near a large NextDC facility in West Footscray, Melbourne. Their concerns included noise, fumes from backup diesel generators and the proximity of industrial infrastructure to residential homes. The facility was reported to be only 15 metres from some houses.

The Victorian government responded to the broader planning controversy by announcing new restrictions, including a proposed 150-metre separation between data centres and homes or schools. However, the restrictions would not apply retrospectively to already approved developments. Residents argued that the measures came too late for communities already living alongside the facilities.

Noise is not the only possible environmental concern. Many data centres rely upon diesel generators for emergency backup power, with periodic testing and maintenance requiring their operation. Diesel exhaust contains fine particulate matter and nitrogen oxides, pollutants whose adverse effects on respiratory and cardiovascular health are well established. Some facilities also use gas-fired power generation, creating additional questions about emissions.

The distinction between a potential hazard and demonstrated injury is important. Diesel exhaust is a recognised health hazard, but the actual risk to a neighbouring household depends upon emissions, operating hours, distance, weather conditions and measured exposure. It would be scientifically irresponsible to conclude that every data centre creates dangerous pollution levels. It would be equally irresponsible to assume that the absence of detailed local measurements establishes that there is no problem.

Water consumption presents another issue. Large data centres may require substantial quantities of water for cooling, depending upon their design and local climate. In regions already experiencing water shortages, such demand can place additional pressure upon community resources. Although water consumption does not directly explain reports of dizziness or headaches, it illustrates the broader environmental costs of supposedly clean and immaterial digital technology.

An important contribution to this debate appeared in October 2026 in the journal Environments. A systematic review examining environmental pressures and potential health pathways associated with data centres identified 36 relevant studies containing 112 quantitative estimates. The researchers found evidence of environmental pressures, including noise, air pollution and localised heating. However, none of the included studies directly established health outcomes among exposed populations through epidemiological or clinical investigation.

This finding deserves emphasis. It does not establish that data centres are harmless. Rather, it demonstrates how little direct research has been conducted into the health of people living near them. Environmental exposures have been investigated to some extent, but the crucial connection between those exposures and actual illness remains inadequately studied.

There is a familiar historical problem here. Industrial technologies frequently expand before their long-term environmental consequences are fully understood. Asbestos, leaded petrol and various industrial chemicals provide examples of hazards whose recognition lagged behind their widespread use. These examples do not prove that data centres represent a comparable danger, but they illustrate why waiting for conclusive evidence of injury before conducting proper investigations can be a serious mistake.

The situation also raises questions about the application of the precautionary principle. When communities report persistent disturbances from industrial facilities, the reasonable response is not to declare the technology dangerous without evidence, nor to dismiss residents as irrational. It is to investigate the complaints independently, using appropriate acoustic equipment, air pollution monitoring and health assessments.

In particular, conventional noise measurements may not adequately describe every aspect of a persistent low-frequency sound. Investigations should examine the frequency spectrum, night-time noise levels, tonal characteristics and vibration, rather than relying exclusively upon a single overall decibel reading. Residents' sleep disturbances and other reported symptoms could then be assessed alongside measured environmental conditions.

There are practical engineering solutions worth considering. Better acoustic insulation, quieter cooling technologies, appropriate separation distances, improved generator emissions controls and stricter planning requirements could reduce potential problems. Independent monitoring, with results made publicly available, would also help establish whether particular facilities are complying with environmental standards.

For older people, these concerns may be especially significant. Ageing is often accompanied by increased vulnerability to sleep disruption, cardiovascular stress and respiratory illness. Individuals with existing medical conditions may be less able to tolerate additional environmental burdens. This makes the location of large industrial facilities near established residential communities a matter deserving careful scrutiny.

There is also a broader political and economic question. The companies developing artificial intelligence infrastructure are frequently among the wealthiest corporations in the world. They stand to gain enormous commercial advantages from the expansion of data processing capacity. Yet if nearby residents experience noise, pollution, falling property values or deteriorating quality of life, those costs may be imposed upon people who receive little direct benefit from the developments.

This is a familiar problem of economic externalities. The profits are privatised while some of the environmental burdens are transferred to surrounding communities. Proper planning and regulation should require developers to account for these costs rather than leaving residents to fight expensive legal and administrative battles after facilities have been constructed.

The emerging controversy over data centre health effects therefore deserves attention, but also scientific restraint. The reports of illness are real as reports of people's experiences. Whether the full range of symptoms can be attributed to data centre operations remains an open question. Persistent noise and combustion-related pollution are plausible pathways of harm, but specific causal claims require evidence.

The sensible course is to investigate now rather than wait decades for the consequences of uncontrolled expansion to become clearer. The artificial intelligence industry should not be permitted to treat residential communities as an uncontrolled environmental experiment. If the technology is as beneficial as its promoters insist, its infrastructure should be designed and operated in ways that protect the people living nearby.

The digital revolution may be transforming the world, but it should not come at the expense of the ordinary person's right to a quiet home, clean air and a decent night's sleep.

https://www.midwesterndoctor.com/p/what-is-making-people-near-data-centers

https://www.abc.net.au/news/2026-09-23/senate-inquiry-probes-ai-data-centre-affects-on-communities/107185098

https://www.frontiersin.org/journals/climate/articles/10.3389/fclim.2026.1648912/full