Imagine growing piles of hazardous electrical waste, water consumption six times greater than that of Denmark, and electricity usage ten times that of a single Google Search. These are just a few of the hidden environmental costs of Artificial Intelligence (AI) and the massive data centers that power it.
Over the past decade, AI has surged in popularity and use. According to the UN Environmental Program, there were roughly 500,000 data centers worldwide in 2012. Today, that number has skyrocketed to an estimated 8 million.
AI data centers are specialized facilities that house the computing infrastructure required to train, deploy, and deliver AI applications and services. Traditional data centers are often unequipped to handle AI’s immense energy and processing demands, leading to a wave of retrofitting and new construction across the globe.
In the United States alone, Goldman Sachs Research projects that AI systems will require 600 kilowatts of power by 2027. That is enough to supply electricity to 500 homes. Globally, AI is expected to represent 28% of the overall data center market by the same year. While this technological growth signals economic opportunity, it comes at the expense of small, marginalized communities.
To stay cool, these facilities consume millions of gallons of water yearly. In cities such as Phoenix and Atlanta, many centers are located near predominantly Black and Brown neighborhoods, often built there without much input from residents. These communities already face environmental burdens such as heat islands—urban areas that face significantly higher temperatures than surrounding rural areas—, aging infrastructure, and limited political influence to advocate for change.
In response to growing criticism, tech companies like Anthropic and Nvidia have begun adopting “Green AI” policies. This movement aims to reduce energy waste and promote sustainable development. Still, the transition toward environmental accountability remains slow.
The harm goes beyond carbon emissions. AI data centers produce constant mechanical noise. Its servers reach sound levels as high as 96 decibels, well above the 85-decibel limit that the Occupational Safety and Health Administration (OSHA) identifies as potentially damaging to human hearing.
Residents also report increased utility costs, restricted access to green spaces, and rising public health concerns, including asthma and other respiratory illnesses.
In Virginia’s Data Center Alley, backup gas-powered generators are causing an estimated 14,000 cases of asthma symptoms. If the generators were operating at just 10 percent of the allowable pollution level, the public health cost from these emissions would be between $220 million and $300 million per year.
The US Environmental Protection Agency (EPA) believes that by 2030, AI data centers will create a national public health cost exceeding $20 billion.
As AI becomes more integrated into daily life and takes a larger hold on various industries, its environmental footprint cannot be ignored. Without stronger regulation, community input, and investment in green infrastructure, AI data centers may leave vulnerable communities paying the price for the world’s digital gain.
For more information:
AI has an environmental problem. Here’s what the world can do about that.
How AI Is Transforming Data Centers and Ramping Up Power Demand | Goldman Sachs
What Is an AI Data Center? | IBM
Explained: Generative AI’s environmental impact | MIT News
US: Public health cost of AI data centers could exceed $20 billion
