Construction and Consequences: The Human Impacts of AI Data Centers

With the rapid expansion of generative AI, we are seeing more and more large-scale data processing centers emerge across the country, especially in rural areas. Governments, researchers, and communities alike have been forced to face a glaring reality: where there are new AI data centers, there are human lives directly impacted by their creation and operation.

Water

We rely on water for everything — drinking, cooking, cleaning, and growing food. Data centers are among the thirstiest industrial users of water, requiring enormous quantities to cool the computing hardware operating inside them. In 2021, just one of Google's data centers in Oregon consumed 355 million gallons of water. In 2023, all of Meta's data centers worldwide used approximately 1.4 billion gallons — around 672 million of those gallons drawn from local water sources.

Unlike some industrial water uses, the extraction process for data center cooling is largely permanent. These facilities deplete millions of gallons from communities' local water supply every year, and with the industry's rapid expansion, that consumption will only grow. Some residents living near new data centers report concerns that these centers are draining local wells and aquifers, leaving nearby homes with reduced water pressure or water quality issues. For communities in drought-prone regions, the competition between industrial and residential water needs poses a serious long-term challenge.

Energy

AI data centers require extraordinary amounts of energy. Global data center power usage was estimated at around 200 terawatt-hours per year in 2016, and is projected to rise to nearly 3,000 terawatt-hours per year by 2030. This surge in electricity demand is straining local energy grids, particularly in rural areas where infrastructure was not designed to handle industrial-scale power loads.

The majority of data centers' energy still relies on fossil fuels and conventional power plants. This increased pressure raises the risk of more frequent and more costly outages for surrounding communities. Research from Harvard suggests that under-the-radar agreements between utility providers and large tech companies may be contributing to increased electricity rates for everyday residential customers — with ordinary families unknowingly subsidising the energy consumption of trillion-dollar corporations.

Public Health

Since AI data centers rely heavily on fossil-fuel power, they increase local air pollution — particularly in communities that already face environmental burdens. These facilities also run diesel backup generators in case of grid failure. Those generators can release 200 to 600 times more nitrogen oxides (NOx) than a natural gas plant while producing the same energy output. NOx exposure causes irritation of the eyes, throat, and nose, respiratory infections, and in severe cases, death.

The Institute of Electrical and Electronics Engineers (IEEE) estimated that data center air pollution caused approximately $6 billion in public health damages in 2023 alone. Because these facilities are often sited in rural areas with large lower-income populations, the health burden falls disproportionately on communities with the fewest resources to address it. This pattern reflects a broader environmental justice issue — the people who benefit least from AI technology are often the ones bearing the highest costs of producing it.

Economy

There is an economic upside to data center construction: the building phase creates thousands of temporary jobs, and permanent facilities do hire local staff. However, the long-term employment picture is often smaller than communities expect. Some large data centers operate with as few as 50 permanent employees after construction concludes. Even facilities that commit to hundreds of local hires represent a relatively modest economic return compared to the land use, infrastructure demands, and health costs they impose.

Community members have also raised concerns about opportunity cost — the land occupied by data centers could often have been used for agriculture, light manufacturing, or other development that produces more sustained local employment without the same environmental footprint.

What Now?

Artificial intelligence is not going away. We can expect dozens more data centers to break ground in the coming years. But behind AI's remarkable capabilities, there are real communities, real water supplies, and real lungs affected by the infrastructure powering it.

Companies building these centers should commit to renewable energy sources, adopt water-recycling systems, and improve the efficiency of their computing hardware. Local governments must ensure that infrastructure costs do not land on ordinary ratepayers, and that environmental impact assessments are genuinely independent. With the right policies and genuine corporate accountability, AI infrastructure can be built in ways that support — rather than exploit — the communities that host it.