A recent Tom’s Guide analysis examined one of the most common criticisms of artificial intelligence: that AI data centers are consuming so much water that they are depleting local supplies. The conclusion is more nuanced than many headlines suggest. While AI data centers do use substantial amounts of water and electricity, the evidence shows that their impact varies greatly depending on location, cooling technology, water sources, and local environmental conditions.
The strongest concern is not national water consumption but local concentration. Studies show that many new AI data centers are being built in regions already facing drought or water stress. A recent analysis found that roughly two-thirds of planned U.S. AI data centers are located in areas that experienced drought conditions within the past year, raising concerns that even a relatively small share of total water use can have significant effects on nearby communities.
At the same time, experts caution against oversimplifying the issue. Data centers account for a relatively small percentage of overall water withdrawals compared with sectors such as agriculture, power generation, and municipal consumption. In addition, many operators are investing in recycled wastewater, closed-loop cooling systems, and water-efficient technologies to reduce their reliance on fresh drinking water supplies. Some newer facilities are even designed to operate with little or no water for cooling under certain conditions.
Another important point is that the water footprint of AI extends beyond the data center itself. Research indicates that a large portion of AI-related water consumption comes indirectly through electricity generation and semiconductor manufacturing rather than server cooling alone. As AI adoption grows, these upstream demands may become just as important as the water used directly by computing facilities.
The article ultimately argues that the question is not whether AI data centers use water—they clearly do—but whether communities, regulators, and technology companies can manage that demand responsibly. The evidence suggests that fears of AI universally “draining towns dry” are often exaggerated, but concerns about specific projects in water-scarce regions are legitimate and increasingly shaping public debate over the future of AI infrastructure.
