Opinion: AI’s growing appetite for power is a public health issue. Here’s a fix

By Sten H. Vermund and Patricia J. Kissinger

June 18, 2026

Vermund is dean of the University of South Florida College of Public Health and a professor of epidemiology. Kissinger is associate dean of the Tulane University Celia Scott Weatherhead School of Public Health and Tropical Medicine and a professor of epidemiology.

From accelerating drug discovery to improving diagnostics, AI’s potential in health care is enormous.

But AI is also creating a new and largely overlooked strain on something fundamental to health: the electricity and water systems that sustain modern life.

The computing power behind AI depends on vast data centers running continuously. These facilities require  extraordinary amounts of electricity and water  for cooling at a scale growing faster than many communities and power grids can accommodate.

As hundreds of new data centers are planned across the U.S., residents and policymakers are asking difficult questions. Will this surge in demand raise electricity prices? What will be the greenhouse gas consequences? Will it strain aging infrastructure? Will it create heat islands with adverse local consequences? Who will bear the health consequences if it does?

Reliable electricity is a public health necessity. Hospitals, clinics, emergency services, vaccines and medic medicines, and home medical equipment all depend on stable and affordable power. When energy systems are stressed, the consequences can include service disruptions, higher costs for vulnerable populations, and increased reliance on fossil fuels that worsen air quality and climate-related health risks. Power outages can mean the failure of air conditioning systems or life-saving home medical equipment. That’s particularly dangerous for older people, those who are disabled , and those with chronic illness .

Current approaches to managing these impacts are inadequate. Technology companies often rely on carbon offsets or annual renewable energy purchases to balance their emissions. These strategies do not ensure that clean energy is available when and where it is needed. Data centers may draw power from fossil-fuel–dependent grids while still claiming to be “net zero” on paper.

The implications extend well beyond accounting frameworks. When data centers rely on fossil-fuel-dominated electricity, they contribute to emissions of fine particulate matter and other pollutants that are linked definitively to cardiovascular disease, respiratory illness, and premature mortality.

These exposures are not evenly distributed. Communities of lower-income and/or with historically marginalized populations are more likely to be located near power plants and major transmission infrastructure. Lower-income communities thus bear a disproportionate share of the resulting health burden. Increasingly, our continued reliance on fossil-fuel-energy generation contributes to greenhouse gas emissions that intensify climate-related risks, including extreme heat , now a leading cause of weather-related mortality and a driver of adverse cardiovascular, renal, and maternal health outcomes.

Growing electricity demand also has implications for health system resilience. Strain on regional power grids can increase the risk of instability or outages, with direct consequences for patients at hospitals, clinics, and at home who depend on uninterrupted power for life-sustaining care. Data center expansion also increases demand for water used in cooling systems, exacerbating local water stress and indirectly affecting sanitation, heat mitigation, and overall community health.

There is a more effective and transparent solution. Every kilowatt-hour of electricity consumed by AI data centers should be matched by new, clean energy added to the same grid at the same time. This “hourly, location-based” matching is sometimes called 24/7 carbon-free energy; it can ensure that increased demand is paired directly with increased clean supply. Emerging research suggests that aligning large electricity demand with renewable generation, storage, and real-time grid management can simultaneously reduce costs, emissions, and system strain. Importantly, it can also reduce real-time exposure to air pollution in the communities where electricity is generated. In this sense, kilowatt-hour matching is not only a policy innovation but a targeted public health intervention.

Carbon offsets, by contrast, can be difficult to verify and are often disconnected from local conditions. Kilowatt-hour matching is measurable and enforceable, aligning corporate incentives with public health and environmental needs. Data center operators gain more predictable energy costs and improved reliability. Communities gain new clean energy infrastructure rather than additional exposure to pollution. Regulators gain a clearer, more accountable framework for oversight.

Maintaining resilient infrastructure, protecting vulnerable populations from rising energy costs, and preventing the health harms associa…