As the artificial intelligence boom accelerates, a less visible consequence is drawing increasing attention, which is water consumption. The quick expansion of the data centers, stemmed from soaring demand for AI services since the launch of ChatGPT in 2022, is driving significant growth in water use across the technology industry.
Data centers require water primarily for cooling servers and related equipment. Researchers estimate that U.S. data-center water demand could grow nearly five times between 2022 and 2028 as technology companies race to build new facilities and support increasingly powerful AI systems.
Among the companies facing scrutiny is Microsoft. In 2020, Microsoft President Brad Smith pledged that the company would become water positive by 2030. The commitment was presented as a part of Microsoft's broader environmental agenda and means the company intends to replenish more water than it consumes across its direct operations by the end of the decade.
However, internal projections reported by The New York Times indicated that Microsoft's water consumption is expected to rise substantially with the AI infrastructure expansion. According to the report, company forecasts initially suggested water use could specifically triple by 2030, though revised figures later reduced that estimate to approximately 150 percent growth. Microsoft reportedly adjusted some of those projections after being contacted by journalists but did not publicly explain the changes.
Water Positive Does Not Mean Using Less Water
The apparent contradiction between a water-positive goal and growing water consumption highlights an important distinction in environmental accounting.
A water-positive commitment does not necessarily require a company to reduce its total water use. Instead, Microsoft's target is based on replenishing more water than it consumes across its operations. In theory, a company can increase water use while still achieving a water-positive outcome if it funds conservation projects, watershed restoration efforts, or other replenishment programs that return an equivalent or greater volume of water to the environment.
However, critics argue that this accounting framework may not fully address local impacts. Water replenished in one region does not necessarily ease shortages in another. As a result, the central debate is not simply whether Microsoft can offset its consumption, but whether rapidly growing water demand creates additional stress in communities that are already struggling with limited supplies.
Local Communities Feel the Pressure
The effects of data-center water use are often concentrated in specific locations rather than spread evenly across national water systems.
Communities such as Phoenix, Arizona, where drought conditions have intensified concerns about water availability, have become focal points in discussions about technology infrastructure. Similar concerns have emerged in Jakarta, Indonesia, a city facing severe water-insecurity challenges, and in Pune, India, where residents have protested water shortages and questioned large industrial water consumption.
While data centers account for only a small fraction of overall national water use, experts note that water management is fundamentally local. A community's reservoirs, rivers, and groundwater supplies are affected by nearby demand, not by national averages.
Industry estimates suggest that a typical data center can use as much water daily as roughly 1,000 households. Larger facilities may consume amounts comparable to the needs of an entire town of 50,000 residents.
Direct and Indirect Water Use
Another important distinction concerns how water consumption is measured.
Direct water use refers primarily to the water consumed at a data-center site, especially for cooling systems. These figures are typically reported in corporate sustainability disclosures.
Indirect water use is broader. It includes the water consumed elsewhere to generate the electricity that powers the facility. In regions where power plants rely heavily on water-intensive cooling systems, this indirect footprint can be substantial.
Critics contend that official corporate figures often emphasize direct operational water use while overlooking the water consumed throughout the electricity supply chain. As AI workloads become more energy-intensive, indirect water demand may become an increasingly important part of the environmental discussion.
Technology May Reduce Future Demand
The industry is not solely focused on expansion. Companies are also investing in technologies designed to reduce freshwater consumption.
Newer cooling approaches, including closed-loop systems, advanced air cooling, and immersion cooling technologies, can significantly reduce the amount of freshwater required to operate data centers. Many technology firms see these innovations as essential if AI infrastructure is tcontinuing to grow while meeting the environmental commitments.
Nevertheless the challenge continues to be significant as AI adoption drives a construction boom in data centers, companies must balance economic growth and technological innovation with the realities of local water scarcity.
Whether global environmental pledges and water replenishment programs can address the immediate impacts adequately would be significant for this case.
){kind=small}
){kind=small}
){kind=small}
){kind=small}