Tech companies do has invested so much money in building data centers in recent months that it is actively driving the US economy, and the artificial intelligence race shows no signs of slowing down. Meta CEO Mark Zuckerberg told President Donald Trump last week that the company would spend $600 billion on US infrastructure – including data centers – by 2028, while OpenAI has already pledged $1.4 trillion.
An extensive fresh analysis examines the environmental footprint of U.S. data centers to find out what exactly the country may face as this expansion continues over the next few years, and where the U.S. should build data centers to avoid the most harmful environmental impacts.
The testpublished Monday in the journal Nature Communications, uses a variety of data, including demand for AI chips and information on state electricity and water shortages, to forecast the potential environmental impacts of future data centers by the end of the decade. The study models a number of different possible scenarios for the impact of data centers on the United States and the planet, and warns that tech companies’ net-zero emissions promises are unlikely to meet the energy and water needs of the massive facilities they are building.
Fengqi You, a professor of energy systems engineering at Cornell and one of the study’s authors, says the study, which began three years ago, comes at “an ideal time to understand how artificial intelligence is impacting climate systems and water use and consumption.”
The artificial intelligence industry is “growing much faster than we expected,” he adds, especially given the Trump administration’s laser focus on the industry. “This whole thing is picking up a lot of momentum right now.”
Not all data centers are environmentally equivalent: much of their water and carbon footprint depends on where they are located. Some U.S. states may have grids powered more by renewable energy or are making good progress in bringing more neat energy to the grid; this significantly reduces carbon emissions from data centers drawing power from these networks. Similarly, states with lower water scarcity are better suited to providing the vast amounts of water needed to nippy data centers. (Cooling also accounts for a vast portion of energy consumption in data centers). The best locations for US data centers over the next few years are states that strike a balance between these two input factors: Texas, Montana, Nebraska and South Dakota, according to the analysis, are “optimal candidates for AI server installations.”
Historically, most U.S. data centers have been concentrated in places like Virginia, the U.S. data center hub, and Northern California. Proximity to Washington, D.C. and Silicon Valley was vital to data center companies, as was the regions’ dense fiber connectivity and skilled workforce. Virginia has also offered significant tax breaks for data centers for years, one of the techniques other states are turning to to attract development. According to Data center mapan industry tool that tracks data center development, of the more than 4,000 data centers in the U.S., more than 650 are in Virginia – the most in the country – and California has more than 320, ranking third.