Do Data Centers Cause Cancer? The Scientific Evidence

No published study has found that living or working near a data center directly causes cancer. The question keeps coming up because data centers combine several things people already worry about: electromagnetic fields, diesel exhaust, industrial chemicals, and round-the-clock operations. Each of those carries its own body of research, and the picture is more nuanced than a simple yes or no. What makes this question especially tricky is that no one has actually studied data centers as a unified cancer risk; researchers have studied the individual exposures a data center produces, often in very different contexts.

What a Data Center Actually Exposes You To

A data center is not a single hazard. It is a building full of servers, cooling systems, backup power generators, and networking equipment, running 24 hours a day. The potential health-relevant exposures fall into a few distinct categories: electromagnetic fields from the equipment and power infrastructure, diesel exhaust from backup generators, chemical compounds from the electronics themselves, persistent low-frequency noise, and the occupational effects of shift work for employees. A recent paper calling for more research on the topic identified these same hazards, noting that hyperscale data center operations involve “persistent low-frequency noise pollution, elevated ambient temperatures, and the extensive use of diesel backup generators that emit hazardous air pollutants.”¹ The authors emphasized the urgent need for epidemiological studies and environmental health assessments for communities near these facilities. That paper is essentially a research agenda, not a set of findings. The fact that it had to be written tells you how little direct evidence exists.

Electromagnetic Fields and Cancer Risk

The most common fear about data centers involves radiation. Servers, power distribution units, and the transformers feeding a large facility all produce electromagnetic fields. These fields fall squarely in the non-ionizing range, meaning they do not carry enough energy to break chemical bonds in DNA the way X-rays or gamma rays do. That distinction matters, because direct DNA damage is the most straightforward route to cancer.

The research on whether non-ionizing radiation can still promote cancer through indirect pathways has been going on for decades without reaching a firm conclusion. Laboratory studies have found that prolonged exposure to radiofrequency electromagnetic fields can cause oxidative stress, genomic instability, and disruption of cell-signaling pathways. A 2026 review characterized these effects as potentially contributing to “the promotion and progression stages of carcinogenesis” while noting that the evidence “is not sufficient to conclude that RF-EMF exposure is a direct-acting genotoxic carcinogen or cancer initiator.”² An earlier paper framed the mechanism more bluntly: while ionizing radiation directly damages DNA, non-ionizing radiation interferes with oxidative repair mechanisms, leading to oxidative stress and downstream cellular damage.³

Epidemiological studies on electromagnetic fields and cancer have produced a frustrating patchwork. Some studies find a positive association between residential or occupational exposure to extremely low frequency fields and certain cancers, while others find nothing. A review of this literature acknowledged the inconsistency, concluding that magnetic fields “might not be the direct cause of cancer but may contribute to produce reactive oxygen species and generate oxidative stress, which could trigger or enhance the expression of oncogenes.”⁴ That is a long way from saying data centers give people cancer. It is closer to saying the mechanism is biologically plausible under certain conditions, but proving it in humans at real-world exposure levels has been elusive.

The International Commission on Non-Ionising Radiation Protection (ICNIRP) sets exposure guidelines that most countries follow, and data centers operating within those limits produce field strengths well below the thresholds considered potentially harmful.⁵ For people living near a data center, the EMF exposure from the facility itself is almost certainly trivial compared to the fields from their own household electronics, Wi-Fi router, or the cell phone they hold against their head.

The Power Line Question

Large data centers draw enormous amounts of electricity, and their power supply infrastructure often includes high-voltage transmission lines. This creates a secondary concern: not the data center’s own emissions, but the power lines feeding it. Research on residential proximity to high-voltage lines and childhood cancer has been ongoing since the late 1970s, and the results remain inconclusive.

A large case-control study in England and Wales found that for each increase of 0.2 microtesla in magnetic field exposure from power lines, the estimated relative risk for childhood leukemia was 1.14, but the confidence interval was wide enough to include no effect at all.⁶ A Japanese study found that children living in districts where at least half the area fell within 300 meters of high-voltage lines had a higher incidence of hematological malignancies, but the association was not statistically significant.⁷ A smaller Iranian study did find a significant association between living near high-voltage lines and childhood acute lymphoblastic leukemia, but it involved just 22 cases and 100 controls.⁸

The overall picture is that if power line proximity carries any cancer risk, it is small and difficult to separate from confounding factors. And while a data center does attract heavy electrical infrastructure, you would face similar exposure living near any large industrial power consumer or substation. This is not a data-center-specific hazard.

Diesel Backup Generators

This is where the evidence gets more concrete, though still not in the way most people expect. Nearly every data center has rows of diesel generators standing ready to kick in during power outages. These generators are tested regularly, and during testing or actual use, they produce exhaust containing particulate matter, nitrogen oxides, and other combustion byproducts. A study of Texas data centers estimated that generator testing alone can emit roughly 12 metric tons of nitrogen oxides per facility per year, contributing to ozone problems in urban areas.⁹

The International Agency for Research on Cancer classified diesel engine exhaust as a Group 1 human carcinogen in 2012, meaning there is sufficient evidence that it causes cancer in people.¹⁰ A pooled analysis of 14 case-control studies confirmed the carcinogenicity of diesel exhaust, though it noted that exposure-response relationships across different lung cancer subtypes remained poorly understood.¹¹ A systematic review and meta-analysis found a statistically significant dose-response relationship between occupational diesel exhaust exposure and lung cancer risk.¹²

The crucial context here is exposure level. The studies establishing diesel exhaust as carcinogenic were done on miners, truck drivers, and railroad workers who breathed the stuff for hours a day over years or decades. A data center’s backup generators run intermittently, typically for short testing periods and occasional outages. The exposure experienced by a neighbor or even a facility worker is orders of magnitude lower than what a diesel miner inhaled. That does not make it zero, and communities downwind of large facilities with many generators have a legitimate concern about cumulative air quality, but extrapolating from occupational diesel exposure studies to the neighbor of a data center requires care.

Flame Retardants and Electronic Hardware

Data centers contain vast quantities of circuit boards, cables, and plastic housings, all of which are treated with flame retardants to meet fire safety codes. The most widely used brominated flame retardant is tetrabromobisphenol A (TBBPA), found in printed circuit boards and plastic casings. Some earlier flame retardants, like polybrominated diphenyl ethers (PBDEs), were phased out of production over concerns about persistence and bioaccumulation.¹³

Exposure to flame retardants is most concerning during electronics recycling and dismantling, when circuit boards are broken apart and heated. Measurements at electronics recycling sites found airborne concentrations of PBDEs, TBBPA, and related compounds at levels that researchers flagged as potentially hazardous to workers.¹⁴ For workers inside an operating data center, however, the electronics are intact and enclosed. A 2025 risk assessment that reviewed 55 studies on TBBPA exposure from electronic devices found that hazard quotients across all exposure scenarios, including high-exposure e-waste settings, remained well below 1, meaning a low likelihood of adverse health effects.¹⁵ Even workers in electronics dismantling facilities came in under the toxicological concern threshold.

So while flame retardants in electronics are a real occupational health issue for people who break apart old hardware for a living, they are not a meaningful cancer pathway for data center technicians walking through rows of sealed servers, much less for people living nearby.

Shift Work and Circadian Disruption

Data centers run around the clock, which means a significant portion of their workforce does shift work, including overnight shifts. This is a well-established health concern that has nothing to do with the data center itself and everything to do with what irregular hours do to the body. The International Agency for Research on Cancer has classified shift work involving circadian disruption as a probable human carcinogen (Group 2A), based on associations with breast, prostate, and colorectal cancers.¹⁶ The mechanism involves nighttime light exposure suppressing melatonin production and activating stress-related pathways that interfere with sleep-wake cycles.

This is worth knowing if you work night shifts at a data center, but it applies equally to nurses, factory workers, warehouse employees, security guards, and anyone else on a rotating or overnight schedule. It is not a data center problem; it is a modern labor problem. Still, given that many data center workers spend years on rotating shifts in brightly lit facilities, this may represent the most evidence-supported cancer-adjacent risk for the people who actually work inside them.

Noise as an Indirect Stressor

Large data centers generate persistent low-frequency noise from cooling fans, HVAC systems, and power infrastructure. This kind of hum is distinct from the louder, more variable noise of construction or traffic. Research has found that even moderate levels of low-frequency noise can have measurable physiological effects. A study on noise-sensitive individuals found that low-frequency noise significantly blunted the normal daily decline in cortisol, indicating elevated stress hormone levels throughout the workday.¹⁷ Chronic stress and elevated cortisol have been linked to immune suppression and other downstream health effects in broader medical literature, but the leap from “data center noise raises cortisol in sensitive individuals” to “data center noise causes cancer” is enormous and unsupported.

For neighbors of data centers, low-frequency noise is more likely to cause sleep disruption, annoyance, and stress-related symptoms than to pose a direct cancer risk. These quality-of-life effects are real and worth taking seriously on their own terms, but they belong in a different conversation than carcinogenesis.

Occupational EMF Exposure for Electrical Workers

Data center technicians work in proximity to substantial electrical infrastructure, and some studies on electrical workers in general have found modestly elevated cancer risks. An overview of occupational EMF exposure studies found that electrical workers showed “a moderate consistency” of elevated leukemia risk, with risk ratios typically between 1.2 and 2.0, and similar patterns for brain tumors.¹⁸ A Los Angeles County study of electrical workers found a weakly positive trend in leukemia risk with increasing magnetic field exposure.¹⁹

These findings come with important caveats. “Electrical worker” is a broad category that includes power line technicians, electricians, and substation operators, many of whom experience much higher field exposures than a data center technician walking server aisles. The elevated risk ratios are small, and the studies struggled to separate EMF exposure from other occupational hazards like solvent exposure or socioeconomic factors. A data center worker’s EMF exposure profile would be quite different from that of someone who spends their day on high-voltage transmission towers.

Why People Are Worried Anyway

Public concern about electromagnetic fields and cancer runs well ahead of the scientific evidence, and data centers have become a lightning rod for that anxiety as they proliferate near residential areas. A survey of health professional students found that about a third thought chronic EMF exposure increases cancer risk, while another third were unsure, and nearly half expressed general concern about radiation exposure despite having limited knowledge of EMF sources.²⁰ A scoping review of general practitioners found that high levels of concern about EMF health effects among both doctors and the public “stand in contrast to the rather limited scientific evidence for an association between EMF exposure and various health outcomes.”²¹

Part of what drives this disconnect is that data centers are new, large, and mysterious. Most people have never been inside one and have no frame of reference for what the building actually does or what it emits. They see a windowless concrete structure with industrial cooling, hear a persistent hum, and worry. That reaction is human and understandable, but the current evidence base does not support it as a cancer concern.

What Actually Deserves Attention

If you live near a data center, the exposure most worth monitoring is air quality from diesel generators, especially if you are in an area that already struggles with ozone or particulate matter pollution. Large facilities with dozens of generators running regular test cycles add a nontrivial amount of nitrogen oxides and particulate matter to local air. That air quality degradation affects respiratory health in ways that are well-documented, and in the long run, chronic exposure to fine particulate matter does carry an established cancer risk for lung cancer specifically. Communities have every right to push for emissions controls, cleaner backup power alternatives like battery storage, and transparent air monitoring.

For data center workers, the most evidence-backed concern is circadian disruption from shift work, not the servers or the wiring. If you are on a rotating night schedule, the health risks associated with that schedule are real, backed by substantial epidemiological evidence, and manageable to a degree with sleep hygiene, light exposure management, and scheduling practices that minimize the most harmful rotation patterns. The flame retardants in the equipment you walk past every day appear to pose minimal risk at normal operating conditions, and the EMF levels inside a well-designed facility fall within international safety guidelines.

Underground fuel storage is another area where data centers overlap with known environmental health risks. Data centers with on-site diesel storage tanks share the same leaking-tank concerns as gas stations and other fuel depots. Across the United States, there are hundreds of thousands of underground storage tanks holding petroleum or hazardous substances, and many of them leak, potentially contaminating soil and groundwater in surrounding communities.²² For data centers that store large volumes of diesel fuel, proper tank maintenance and monitoring are not a theoretical concern. Hydrocarbon contamination of drinking water is a genuine public health pathway, even if it has nothing to do with the servers inside the building.