This image is provided by Geoffrey Moffett via Unsplash. 

TL;DR: The rapid growth of Artificial Intelligence (AI) data centers is increasing energy demand, water consumption, and environmental pressures across the United States and the world. While these facilities support advanced AI applications, their reliance on fossil-fuel-powered electricity can contribute to data center air pollution and greenhouse gas emissions. Understanding data center environmental impacts helps balance technological innovation with community health, sustainability, and emissions reduction efforts.

The growing demand for AI

Artificial Intelligence is increasingly being used in all walks of life, from medical devices and self-driving cars to internet searches. In 2024, 78% of business organizations reported using AI, a number which rose to 88% in 2025. Also in 2025, the global corporate investment in AI totalled 581.69 billion US dollars.

Across the United States, data centers are being built at a drastic rate to accommodate exponential growth in AI adoption. This unprecedented rapid expansion is fueled by billions of dollars in tax incentives and fast-track permitting. The number of large data centers operated by hyperscale providers reached 1,136 by the end of 2024, having doubled over the previous five years, with the United States accounting for over half of the total worldwide capacity, measured by MW of critical IT load. 

Synergy Research Group predicts that it will take less than four years for the total hyperscale data center capacity to double once again, with 130 to 140 additional hyperscale data centers coming online yearly. 

This Synergy Research Group graphic illustrates the growth of data centers and where the most data center capacity is located. 

What are AI data centers? 

AI data centers are the facilities that house the specific information technology infrastructure needed to train, deploy, and deliver artificial intelligence applications and services. These data centers contain advanced architectures for computing, networking, and storage. 

AI data centers share similarities with traditional data centers, but AI data centers have much more computing power to accommodate high-intensity artificial intelligence workloads. They use a large number of high-performance graphics processing units (GPUs), which require certain associated infrastructure like advanced storage, networking, energy, and cooling capabilities. They also often take up more physical space than traditional data centers due to these requirements. 

Hyperscale data centers are massive. They are engineered for large-scale workloads, such as generative AI. These data centers have at least 5,000 servers, likely miles of connection equipment, and can take up millions of square feet. 

This graphic, provided by the World Resources Institute and the Polsky Center for the Global Energy Transition, based on data from datacentermap.com, shows data centers that are operational, planned, land-banked, or under construction across the United States as of early 2026.

The consequences of data centers

Data centers might help support the growing demand for AI, but they also come with unintended consequences for public health and the environment. Just one modern AI data center can use as much electricity as 100,000 households, with the largest ones currently under construction possibly consuming up to 20 times more energy than that. 

Water use is another significant contributor to data center environmental impacts. Data centers rely on water to cool their facilities. A medium-sized data center consumes roughly 300,000 gallons of water per day, which is about the same amount as 1,000 households. Larger data centers can use up to 5 million gallons per day, or roughly 1.8 billion annually. All the United States data centers together consume many billions of gallons of water every year. This water consumption is especially harmful for the data centers located in areas already facing water scarcity and drought. 

Unfortunately, data centers often exist in areas that already face above-average health burdens from air and water pollution. Communities of color may disproportionately experience the effects of data center air pollution. This image is provided by Anne Nygård via Unsplash. 

The AI data center community impact extends beyond energy consumption and water use, affecting quality of life in lesser-known ways. Data centers are responsible for a significant amount of noise pollution, which can disrupt the lives of locals. They also take up large expanses of land, sometimes well over a square mile, which can accelerate the loss of productive farmland. 

AI data center air pollution

The high energy demands of data centers frequently lead to the increased use of dirty energy, with roughly 56% of the electricity used to power data centers in the United States coming from fossil fuels. Many AI data centers are located in carbon-intensive areas, using energy generated from fossil fuel power plants.

This World Resources Institute and Polsky Center for the Global Energy Transition graphic, made with data from datacentermap.com, illustrates the US states with the most data centers. Northern Virginia is the largest data center market in the world. Unfortunately, Virginia’s power grid relies heavily on fossil fuel power, contributing to data center air pollution. 

Other AI data centers are even building or planning to build on-site fossil fuel-driven power plants to keep up with energy demands and alleviate strain on the local energy grid. These new power plants, often fueled by natural gas, might help keep up with the growing power demand, but they contribute to climate change and harm local communities. A study looking at 2,132 data centers in the US operating between September 2023 and August 2024 found that the centers generated more than 105 million tons of CO2e. 

Data centers also have diesel generators for emergency backup power, which produce fine particulate matter (PM2.5), carbon monoxide (CO),  nitrogen oxides (NOx), and other air pollutants. While these generators are theoretically used sparingly, they can emit 200 to 600 times more NOx than natural gas plants, as well as more fine particulates and other toxic emissions. One analysis found that data center backup generators in Virginia could already cause 14,000 asthma-related cases every year, along with other health impacts, resulting in a total public health burden of $220-300 million annually. 

This image, taken from a Data Center Map, illustrates some of the data center hot spots centered around the area known as Data Center Alley in Northern Virginia. One data center in this area, which uses on-site gas turbines for power, was estimated to result in between $53 million and $99 million in yearly health damages from associated air pollution. The facility was also estimated to cause 3.4 to 6.5 additional premature deaths per year across the impacted region. 

What can be done about AI data centers’ environmental impacts?

Fortunately, there are some steps that AI data centers can take to mitigate their negative impacts. For instance, data centers can take advantage of more water-efficient cooling methods such as closed-loop cooling systems or immersion cooling. They can even use acoustic barriers to help reduce noise pollution. Deliberate permitting practices can encourage AI data centers to make sustainable choices. 

AI data center air pollution can be mitigated by the use of more renewable energy, either by building new centers in areas with more clean energy or generating their own renewable energy on-site. They can also take steps to make their facilities more energy efficient and use battery energy storage systems instead of diesel generators. 

Air quality monitoring helps regulators enforce air pollution standards while providing communities with valuable information about potential health risks from local pollution sources like AI data centers. Partner with Clarity to implement an air quality monitoring system to promote clean air and healthy communities.