WATER POLITICS

Water-stressed regions of the U.S., including in Arizona and Texas, are expecting increased manufacturing of data centers, chip fabrication facilities, and other AI-related infrastructure that need massive amounts of water to operate, according to a report called “The Future of Water Resilience in the U.S.” by JPMorgan Chase (JPM+1.47%) and the ERM Sustainability Institute.

Data centers, which are essential to AI model training, require water to cool server rooms — often from drinking water sources, the report said. While a growing and moving U.S. population is the main driver of a decrease in fresh water supply in parts of the U.S., according to the report, “20% of the water used by data centers today is drawn from already stressed watersheds, presenting risks to the technology industry and the surrounding communities and environment.”

A mid-sized data center consumes, on average, about 300,000 gallons of water each day, according to the report. But larger data centers can use up to 5 million gallons of water each day — or the same amount as a town with 10,000 to 50,000 residents, the report said, adding that in 2023, U.S.-based data centers used over 75 billion gallons of water. Additionally, data centers require chips, which need vast amounts of water to manufacture. Chips also discharge “highly toxic wastewater that is saturated with chemicals and heavy metals,” according to the report.

Climate change and water management issues already create challenges for the water supply, but “AI and data centers are increasing the scale of the challenge,” Rama Variankaval, global head of corporate advisory at JPMorgan and a leading contributor to the report, said in a statement shared with Quartz. But it’s “also putting a spotlight on the issue.”

Mishandling the impact on water-stress “could cause real disruption to global supply chains,” the report said, especially in the AI industry, which count data centers and chips as crucial businesses. Despite the negative impact to climate goals, experts previously told Quartz that data and other AI infrastructure will be the winners of AI’s next phase, as companies seek to power their growing AI offerings.

Water investment is needed in the U.S., the report said, adding that water “scarcity and unpredictability” could slow GDP growth by up to 6% in areas already impacted by water stress. According to the report, water infrastructure in the U.S. has an annual shortfall of $91 billion in public spending.

And growing data center demand is happening amid changing precipitation patterns that are increasing rainfall and snowfall in some areas, while leading to shortages in others. While data center sites are determined with cheap, clean energy access in mind, “incremental water stress on these locations is expected to impact site location for future data centers,” the report said.

Investments into flood control infrastructure and water and wastewater treatment infrastructure are needed, the report said, while emerging technologies such as decentralized and circular waste systems have growing market opportunity.

“The increased scale of the challenge, the availability of a set of solutions (each applicable in different situations), and the public attention to the issue are bringing in various stakeholders, including investors, to focus on the water space,” Variankaval said.