Quick Buzz Feed
Large data centers for artificial intelligence (AI) are increasing demand for water, energy and infrastructure. Learn how public health can help balance innovation, sustainability and community health.
The rapid growth and widespread use of artificial intelligence (AI) are transforming many sectors but also carry significant environmental implications. The infrastructure required for AI leads to increased demand for water, higher energy consumption, and expanded grid infrastructure. These demands necessitate careful management to prevent environmental damage and harm to communities.
U.S. data centers consumed an estimated 183 terawatt-hours (TWh) of electricity in 2024, accounting for over 4 percent of the nation's total energy usage. This figure is projected to surge by 133 percent to 426 TWh by 2030. States like Virginia, Texas, and Oregon have recorded the highest data center-attributable emissions. Training typical AI research pipelines involves thousands of models, leading to cumulative carbon dioxide (CO2) equivalent emissions in the tens of thousands of pounds over several months. Training a single AI model alone can emit more than 626,000 pounds of COâ‚‚ equivalent, which is nearly five times the lifetime emissions (including manufacturing) of an average American car.
Data centers' direct water consumption was estimated at 17 billion gallons in 2023, with projections indicating a potential doubling or quadrupling of this use by 2028. By 2021, one in five U.S. data centers was already situated in areas experiencing water stress. Water use in data centers occurs both directly for cooling and indirectly through electricity consumption. Cooling processes alone can account for up to 40 percent of a data center's total electricity consumption, with some facilities drawing over half their water from potable sources. The continuous increase in data center capacity means a corresponding rise in water demand as computing infrastructure expands.
In the United States, electricity generation for data centers is predominantly fossil fuel-based (56 percent), followed by renewable energy (22 percent) and nuclear power (21 percent). The rapid expansion of data centers is causing delays in coal plant closures and hindering national, state, and local efforts to transition to clean energy. This is because existing renewable energy sources are currently insufficient to meet the combined demands of hyperscale data centers and existing users.
Emerging technologies like Artificial Intelligence often require larger, more energy-intensive data centers to satisfy their significant computational demands. Searches powered by generative AI, for instance, utilize four to five times more energy than traditional web searches. Generative AI tasks, such as text generation, summarization, image captioning, and image creation, typically exhibit higher energy and carbon intensity compared to discriminative (predictive) use cases, which involve using existing data in machine learning models to identify patterns, anticipate user behaviors, and forecast events.
Policy responses to the environmental footprint of emerging technology and AI vary, often navigating the balance between state-level economic development interests and local community health and environmental concerns. States typically focus on energy reporting, protecting ratepayers, and environmental assessments. Local governments, driven by immediate resource impacts and community opposition, tend to implement more direct actions concerning land use, zoning, and permitting. Examples of local actions include Loudoun County, Virginia, which ended by-right zoning for data centers and now requires public hearings for new applications (2025). Kansas City, Missouri, classified data centers as industrial, mandating council approval and impact studies on water and electricity rates (2026). Marana, Arizona, has prohibited potable water use for cooling and requires disclosure of water sources (2024).
Public health practitioners are strategically positioned to mediate the inherent tension between economic development and community health in the context of emerging technologies. The three core functions of public health—policy development, assessment, and assurance—provide a robust framework for this work. In policy development, public health can advocate for health impact assessments during permitting, push for transparency requirements, establish community notification standards, and contribute expertise to state and local rulemaking. For assessment, practitioners can track and analyze cumulative environmental exposures, monitor air quality near diesel generators, evaluate water availability in stressed regions, and assess electricity cost burdens on low-income households, while also advocating for mandatory data collection from operators. In assurance, public health professionals can monitor long-term health outcomes in affected communities, hold operators and regulators accountable to environmental standards, and ensure that vulnerable populations have meaningful access to decision-making processes. A foundational recommendation emphasizes that communities must be equipped with the necessary information, access, and standing to participate in decisions about AI infrastructure that will have long-term impacts on their health. Public health can therefore be a crucial partner in guiding the ethical and sustainable rollout of AI and emerging technologies in our increasingly digital society.
You can download the full report to delve deeper into the detailed findings and analysis.
This report is built upon a growing body of research dedicated to the responsible integration of artificial intelligence (AI) in public health. Special acknowledgment is given to the foundational work from a collaboration involving the Kansas Health Institute, Health Resources in Action, and the Wichita State University Community Engagement Institute. Their publication, 'Developing Artificial Intelligence (AI) Policies for Public Health Organizations: A Template and Guidance,' significantly contributed to critical thinking regarding the rapid expansion of AI infrastructure in public health policy contexts. Jasmin Kamruddin completed work on this project during her internship at the Kansas Health Institute.
The Kansas Health Institute is a nonpartisan, nonprofit educational organization established in 1995 and based in Topeka. It supports effective policymaking through nonpartisan research, education, and engagement, believing that evidence-based information, objective analysis, and civil dialogue are essential for policy leaders to champion a healthier Kansas.