Data centers keep our digital lives running — from streaming movies to storing personal and business data to powering the chatbots we use every day. But the story isn’t all rosy. As these facilities multiply and grow more energy-intensive, they place increasing pressure on the environment, strain local resources, and affect the well-being of the communities where they’re built.
Data centers are the largely unseen infrastructure that keeps our digital lives running. At their core, these facilities store data, run software, and move information across the world so that businesses, governments, and everyday users can stay connected.
There are several kinds of data centers, but they all serve the same basic purpose: housing the hardware that makes online services possible. Most fall into four categories — enterprise, colocation, hyper-scale, and edge — each designed to meet different operational and scale needs. Increasingly, a growing share of this infrastructure is made up of AI data centers. These facilities are built to supply the massive computing power needed to train and operate artificial intelligence systems — everything from machine learning tools to large language models.
As AI workloads grow, so do the demands placed on the data centers behind them. While 174 countries around the world have data centers, the US now houses nearly 40% of them. There are currently 4,000 to 5,000 data centers in the US (depending on the source), with the highest concentration in the state of Virginia. Loudoun County, Virginia, houses approximately 200 data centers — with an additional 117 in development.
According to researchers, “There are multiple reasons for this concentration - from the cheap electricity and high-speed fiber (including undersea cables) that allow fast data transfers, to historical reasons such as early interconnection points and local tax incentives.”
US technology companies like Amazon, Google, and Meta utilize the largest type of data center: hyper-scale data centers, which can be as big as about 20 football fields. As market pressures increase in frontier AI, they need more data centers to keep pace with local and global competitors.
While data center infrastructure is often out of sight, the experience is very different for those living next to these large facilities. Data centers can place substantial strain on local land and communities — from economic pressures to environmental and public health impacts, including noise and light pollution and the conversion of farmland and forests into industrial sites.
AI companies often make big promises to the communities that will house their data centers, like major economic benefits. Unfortunately, few of those incentives trickle down to the local residents.
For one, they promise local job creation — hundreds of new jobs — from their data centers. Yet, the industry’s lack of transparency means no state reports the actual number of jobs created. What’s more, the construction-related roles are often filled by workers from outside the town or state.
To make matters worse, AI companies often receive massive tax breaks from local authorities, which can last for multiple decades. In 2025, Texas and Virginia each spent approximately $1 billion to subsidize data centers. These subsidies can eat into the region’s finances, directly reducing vital spending in areas like education and infrastructure. Oregon's Crook County, which houses numerous data centers including Meta and Apple facilities, “gave away $29 million in 2024 in local revenue to property tax exemptions to corporations—orders of magnitude away from Meta’s donation of roughly $2 million to the county school district over the past decade,” AI Now revealed.
Essentially, public money supports data center construction, with little to no transparency on these massive contracts from officials. In fact, as researchers note: “The data center industry has embraced secrecy: hiding behind non-disclosure agreements, project code names, and subsidiary names, especially Limited Liability Companies (LLCs).”
While AI has beneficial outcomes in many fields, the footprint of this industry is felt deeply at both local and global levels.
Since data centers generate extensive heat, the infrastructure must be constantly cooled. But the cooling process for a single small data center uses more than 26 million liters of water each year. For a hyper-scale data center, that number jumps to 20 million liters per day — as much water as a town of 50,000 people would consume.
Looking deeper into the problem, Business Insider reports that “40% of the nation's planned and existing data centers are in areas characterized as experiencing ‘extremely high’ or ‘high’ water scarcity.”
According to the International Energy Agency, roughly 60% of the energy used to power data centers is generated by fossil fuels, such as natural gas and coal. Researchers add that “in the United States alone, data centers are estimated to generate more than 105 million tons of CO2e, representing 2% of US emissions in 2023.” Their backup diesel generators also contribute to pollution through the nitrogen oxides they emit, which can cause eye, throat, and nose irritation — and in extreme cases, respiratory infection and even death.
Unfortunately, the effects of these pollutants on the public are expected to increase over time. Researchers state that US data centers will push up public health costs to “more than $20 billion in 2028.” They warn that the impact will likely be felt most in disadvantaged communities “where the per-household health burden could be 200x more than that in less-impacted communities.”
Data centers consume massive quantities of electricity to run their servers and keep their infrastructure cool. In the US, data centers use approximately 4.5% of the electricity supply — roughly the same power that would serve 17 million households. Looking ahead, by 2030, data centers may use as much as 12% of US electricity consumption — three times more than current usage.
In Virginia, already home to 35% of all hyper-scale data centers worldwide, these facilities put immense strain on the local grid and push up the cost of electricity for residents.
Unfortunately, data centers can’t rely entirely on renewable energy because their electricity needs can swing sharply from moment to moment. Google’s recent acquisition of Intersect Power may signal a shift toward sourcing “the cleanest possible energy,” as Intersect describes it, which could ease pressure on local grids. Still, the large energy parks or clean energy infrastructure required for these projects can span thousands of acres, raising land-use concerns.
The AI market’s rapid expansion and record investments expose the economy to unprecedented risks. Spending on data center infrastructure is predicted to reach over $1 trillion by 2029, up from $455 billion in 2024, researchers suggest. Microsoft, Amazon, Alphabet, and Meta have already invested roughly $344 billion in 2025, making up 1.1% of the US gross domestic product.
The AI Now Institute’s ‘North Star Data Center Policy Toolkit’ explains what’s at stake, “if there is indeed an AI bubble and that bubble bursts, this dynamic could jeopardize the stability of our pensions, retirement savings, and insurance policies; it could also potentially produce stranded assets.”
This growing dependence on a highly subsidized industry, characterized by speculative returns and aggressive borrowing, has many analysts concerned that an AI boom slowdown could trigger a massive recession.
While data centers power much of today’s technology, their impacts are becoming increasingly clear. Rapid AI development is intensifying demands on energy grids, water supplies, and the communities that host these facilities. The encouraging shift is that more people are beginning to recognize these trade-offs and speak up. As journalist Karen Hao reminds us, “We need the public to be active participants in calling out these companies.” Ultimately, shaping a future where AI serves the common good means treating its development not as a race to win, but as a shared responsibility — one that benefits when communities stay informed and engaged.
Lindsay Langenhoven
Content Writer