Behind every AI chatbot query and autonomous agent action sits a data center, and the biggest ones consume staggering amounts of energy.

Stargate ’s data center in Abilene, Texas, for example, is designed to draw 1.2 gigawatts of power at full capacity, close to the amount used in a year by one million US homes. And it is just one of many, with hundreds of billions of dollars being poured into the infrastructure needed to power the AI boom.

But the price of that expansion doesn’t stop with the technology companies building and using them.

Higher electricity bills, pressure on water supplies, generous tax breaks, pollution and health concerns can shift significant costs onto consumers, taxpayers, local communities and the environment. These are the hidden costs of the AI data center boom, and they are becoming increasingly difficult to ignore.

As more data centers come online, they add enormous new demand to already stretched electricity grids. Increasingly, some of those costs are finding their way onto consumers’ energy bills. A 2025 Bloomberg analysis found that wholesale electricity prices at some locations near significant data center activity were as much as 267 percent higher than five years earlier. And according to Consumer Reports , 78% of Americans are concerned household energy bills will rise because of data centers.

Researchers are increasingly examining the health impact of data centers on nearby communities. One report by Caltech and the University of California, Riverside projects that data-center-related air pollution could contribute to 1,300 deaths per year in the US by 2030. Researchers and communities have also raised concerns about asthma, cardiovascular disease and the impact of persistent noise, with nearby residents reporting headaches, nausea, sleep disturbances and hypertension.

Data center operators have benefited from substantial tax incentives as governments compete for investment. But those incentives also carry a cost for public finances. In Ohio, the value of lost tax revenue attributed to data center tax breaks rose from $555 million in 2024 to $1.6 billion in 2025, prompting the state governor to pause the program, while several other states have reported billion-dollar-plus losses. Governments offer these incentives in the expectation that infrastructure investment will generate jobs and economic growth. The trade-off is foregone tax revenue that might otherwise fund public services and infrastructure.

Last year, data centers used over a trillion gallons of water to prevent the powerful, energy-hungry chips they house from overheating. This will only grow in the future, which is a problem in a world where half of the population faces severe water scarcity . When you add in water used during the generation of electricity and the manufacture of components needed for data centers to run, the scale of the problem becomes even more worrying. This isn’t a cost that’s easy to quantify, but the knock-on effects in local communities can be severe.

The Political And Regulatory Response

The biggest corporations in the world have bet big on AI. But opposition to new data center developments is growing, and in some cases it is succeeding in delaying, changing or stopping projects.

That pressure is increasingly translating into government action. A recent Cullen International analysis of data center policies around the world found strikingly different approaches, ranging from incentives and accelerated permitting to tougher rules on energy and water use, and even restrictions on new developments.

For data center operators and Big Tech, this is far more than a PR problem. AI’s continued expansion depends on access to enormous quantities of electricity, water, land and infrastructure, which means companies increasingly need the support of governments and the communities where these facilities are built.

AI needs data centers, and demand for them is unlikely to disappear. But as their footprint grows, so will scrutiny over who benefits and who bears the costs. If Big Tech wants to keep building at the pace the AI boom demands, addressing those hidden costs may become every bit as important as building the infrastructure itself.