Artificial intelligence is not just hungry for data. It is thirsty, too, and Jacksonville is squarely in the splash zone.
A new analysis from the United Nations University says data centers already burn through electricity on the scale of a nation and could nearly double that demand by 2030. That global surge is not abstract for Northeast Florida, where utilities, planners and neighbors are staring down proposals for sprawling data campuses and asking who will pay the tab in power and water.
What the U.N. found
According to the latest report from the United Nations University, global data centers used about 448 terawatt-hours of electricity in 2025 and could need roughly 945 terawatt-hours by 2030. The study ties that to a water footprint of around 9.3 trillion liters.
The big energy hog is not training the models but running them. The report warns that inference, the everyday use of AI models to answer questions and generate content, could account for 80 to 90 percent of that energy use. To put it in household terms, a single complex AI video can swallow as much electricity as a 10-watt LED bulb running for about 42 hours and may require several liters of water along the way.
The report pairs those numbers with policy recommendations aimed at governments, utilities and tech firms that are trying to keep the AI boom from colliding head-on with local water and climate goals.
Experts and simple fixes
Experts quoted in national coverage argue that the easiest climate fix is behavioral: simply lean on AI less and avoid being nudged into generative tools by default settings. “The cleanest form of AI use is no use,” UNU director Kaveh Madani told The Associated Press, a story that local outlets such as News4JAX have picked up.
The same coverage highlights low-effort tweaks: switch to search engines like Ecosia or DuckDuckGo, tack “-ai” onto Google searches to dodge auto-generated answers, and choose lower-resolution defaults when you create images and videos. In other words, do not fire up a full AI engine for something a plain old search result or basic graphic can handle.
Why Florida should care
Florida lawmakers and county officials are already moving to get ahead of the curve. The state’s 2026 data center bill creates distinct consumptive-use permits and other rules for large facilities, with key provisions scheduled to take effect July 1, 2026, as detailed by the Florida Senate.
Next door to Jacksonville, Nassau County has hit pause with a temporary moratorium while officials study environmental and infrastructure impacts, according to Nassau County. The urgency is real locally because Duval County is already a known data center cluster, and the region’s utilities and planners are sorting out who funds new grid and water capacity, a debate reflected in materials from the Florida Data Centers industry group.
How to shrink your AI footprint
On the user side, the recommended playbook is simple: skip AI for quick lookups, default to lower-resolution outputs whenever you can, and favor search tools that do not spin up heavy generative models by default.
The UNU team cautions that efficiency gains in hardware and software can easily be wiped out by “rebound effects” if people respond by using AI more. That is why product defaults and demand-side guardrails, like token limits and conservative resolution settings, matter just as much as greener chips and servers.
Communities also have leverage. Research highlighted by regional experts shows that concentrated data center growth can translate into major water withdrawals. One recent estimate found that clustered data centers in Virginia used roughly 2.1 billion gallons of water in 2023, according to the Center for Secure Water. Showing up at hearings and pushing for transparent water and reclaimed-water plans can shape how and where new campuses are permitted.
For Jacksonville, the message is blunt: AI’s environmental price tag is no longer a distant policy seminar topic. With state rules on the way and county-level debates already in motion, residents, utilities and officials have a crucial window to demand clarity, reclaimed-water strategies and serious impact studies before the next hyperscale project starts pouring concrete.
