Former Environmental Protection Agency officials have issued a stark warning that the expansion of artificial‑intelligence data centres will drive a surge in air‑pollution‑related health costs, projecting an additional £20 billion in annual medical expenses by 2028. The group, which includes senior figures who once oversaw air‑quality regulations, says the primary culprit is the reliance on gas‑fired power plants to supply the massive electricity loads required by AI facilities across the United States.
The Surge in AI Data Centre Energy Demand
The rapid growth of AI services has created an unprecedented demand for computing power. According to recent industry analyses, the sector’s electricity consumption is on track to increase by more than 30 % over the next five years, outpacing even the most optimistic forecasts for renewable‑energy adoption. Data‑centre operators argue that this rise is essential to support breakthrough technologies in fields ranging from autonomous vehicles to natural‑language processing. However, the additional load is being met largely through existing natural‑gas infrastructure, which already accounts for a significant share of national emissions.
Health Risks Linked to Gas‑Fired Power Generation
Gas‑fired plants emit a cocktail of pollutants, including nitrogen oxides, particulate matter, and volatile organic compounds. These substances have been consistently linked to respiratory illnesses, cardiovascular disease, and premature death. The former EPA officials’ modelling suggests that the incremental emissions from powering new AI facilities will translate into roughly 2,500 additional premature deaths each year by 2028, alongside a sharp rise in asthma attacks and other acute health events. The financial impact is quantified in the £20 billion figure, which reflects projected increases in hospital admissions, long‑term care, and lost productivity.

The coalition’s report, released late last week, cites a series of peer‑reviewed studies that demonstrate a direct correlation between proximity to gas‑fired generation and elevated health‑care utilisation. One study, conducted in the Midwest, found that communities within a 10‑mile radius of a gas plant experienced a 12 % higher rate of emergency‑department visits for respiratory issues compared with similar locales powered by cleaner sources. The former officials argue that the economic calculus must shift to reflect these hidden costs, urging policymakers to consider them when approving new data‑centre projects.
Washington’s Policy Response and Industry Counter‑points
In response to the warnings, a bipartisan group of senators has introduced legislation that would require data‑centre developers to disclose their energy sources and to contribute to a mitigation fund for affected communities. The proposal, known as the “Clean AI Infrastructure Act”, would also incentivise the construction of renewable‑energy facilities specifically to serve AI operations. Proponents of the bill argue that it aligns with broader climate‑change goals while protecting public health.
Industry representatives, however, caution against overly prescriptive measures. The Information Technology Industry Council argues that mandating renewable‑energy build‑outs could inflate project costs, potentially slowing innovation and job creation. They point to voluntary agreements already in place between major cloud providers and utility companies, which they claim are delivering measurable emissions reductions. The debate is intensifying as Congress moves toward finalising the annual spending package, with the fate of the proposed fund likely to be decided in the Senate’s environment committee.
Why it Matters
The emerging clash between AI expansion and environmental health underscores a broader challenge for the United States: how to harness cutting‑edge technology without compromising the well‑being of its citizens. The projected £20 billion health‑care burden illustrates that the true cost of AI progress extends far beyond capital expenditures, touching the everyday lives of millions who live near gas‑fired plants. As Washington grapples with legislative solutions, the outcome will shape not only the nation’s climate trajectory but also set a precedent for how future technological revolutions are managed. The stakes are high, and the decisions made now will echo through both the data‑centre corridors of the 2020s and the healthier communities of the 2030s.
