Tech Giants Understate Water Usage of AI Data Centers Amid Growing Environmental Concerns
The AI revolution is driving unprecedented electricity demand and massive water consumption, much of which is underreported by major tech companies. A Wall Street Journal investigation reveals that Microsoft, Google, and Amazon significantly understate their data centers' true water use by excluding the vast quantities consumed by power plants generating the electricity that runs these centers. Analysis by the Lawrence Berkeley National Laboratory shows that water used at power plants is on average 12 times greater than the water used directly for cooling data centers.
Meta stands out as the only major tech company transparently reporting water consumption at power plants powering its data centers, with 19 billion gallons used in 2024, over 20 times the water used for cooling. Google reported using 10.9 billion gallons in 2025, mostly for cooling, but research indicates power plants supplying this electricity consume three times more water than Google discloses. Amazon claims its data centers are seven times more water-efficient than industry averages and has achieved 75% of its goal to replenish one gallon of water for every gallon withdrawn, though these figures exclude power plant water use.
Microsoft plans to shift to closed-loop cooling systems by 2027, eliminating water use during normal operations, and Nvidia recently introduced a similar system that drastically reduces water and electricity consumption. However, most existing data centers still rely on evaporative cooling, which consumes large water volumes. Experts warn that upgrading these facilities is costly and complex.
The environmental impact is especially severe in drought-stricken regions like Phoenix, Arizona, where two-thirds of new U.S. data centers are located. Currently, data centers account for about 3% of Phoenix's annual water use, but this could exceed 20% by 2031, rivaling residential irrigation demand. Tech companies often choose locations with cheap land and electricity, frequently in water-stressed areas. Meanwhile, AI's energy demands prolong coal plant operations and drive new natural gas plants, both heavy water users.
Researchers highlight a lack of transparency around AI's true water footprint, complicating public understanding of its environmental costs. Sustainability expert Alex de Fries Gao told the Wall Street Journal, "The public usually sees only a very small part of the picture. This is just the tip of the iceberg."
