Tech companies are pouring massive investments into building data centers lately, significantly boosting the U.S. economy, and the AI race shows no signs of slowing down. Last week, Meta’s Mark Zuckerberg told President Donald Trump that the company plans to invest $600 billion in U.S. infrastructure—including data centers—by 2028. OpenAI has already pledged a staggering $1.4 trillion.
A new in-depth analysis has examined the environmental impact of data centers across the U.S. to better understand what the country might face as this expansion continues in the coming years and where data centers should ideally be built to minimize their environmental consequences. The study, published in the journal Nature Communications on Monday, utilizes a range of data, such as demand for AI chips and information on state electricity and water availability, to assess potential environmental implications of future data centers through 2030. It models several scenarios regarding the effects these facilities could have on the U.S. and the planet while warning that tech companies’ commitments to achieving net-zero emissions may not align with the extensive energy and water requirements of these massive operations.
Fengqi You, a professor of energy systems engineering at Cornell and one of the study’s authors, notes that the research, which began three years ago, comes at “a perfect time to understand how AI is impacting climate systems and water usage.” He adds that the AI industry “is growing much faster than we expected,” particularly with the Trump administration’s intense focus on the sector. “This whole thing is just gaining so much momentum right now.”
Not all data centers have the same environmental impact; their water and carbon footprints vary based on location. Some states boast energy grids that rely more on renewable sources or are making significant strides toward cleaner energy, helping to reduce carbon emissions for data centers connected to these grids. States with less water scarcity are also better equipped to support the large quantities of water needed for cooling—this process is a major part of data center energy consumption. The analysis identifies the best locations for future data centers in the U.S. as states that balance these two factors: Texas, Montana, Nebraska, and South Dakota are deemed “optimal candidates for AI server installations.”
Historically, much of the data center expansion in the U.S. has concentrated in places like Virginia—the country’s data center hub—and Northern California. Being close to Washington, D.C., and Silicon Valley has been crucial for these companies, as has the abundant fiber connectivity and skilled workforce in those areas. Virginia, in particular, has provided significant tax incentives for data centers over the years, a strategy that other states are now adopting to attract development. According to Data Center Map, which tracks data center growth, there are over 4,000 data centers in the U.S., with more than 650 located in Virginia—the highest in the nation—with California following with over 320, ranking third.
