Every AI query evaporates real water. Here are the verified numbers on who's using the most and why it keeps growing
June 2026 · 18 min read
Image Credit: Leonardo AI
Every time you type a prompt into ChatGPT, something you probably never think about happens thousands of miles away. A server heats up. A cooling system kicks in. And water, real, physical water drawn from local supplies, evaporates into the air.
AI is thirsty. And the numbers are starting to get uncomfortable.
What is an AI Data Center?
An AI data center is a large facility packed with servers, GPUs, and networking equipment designed to run AI workloads. Training a model like GPT-4, serving ChatGPT queries, or running Google's Gemini assistant all happen inside these buildings.
They are not ordinary offices with a few servers in the basement. A modern hyperscale AI data center can cover hundreds of thousands of square feet and draw as much electricity as a small city. The companies that operate them, Amazon Web Services, Microsoft, Google, Meta, and Equinix, are spending at a genuinely historic scale. Between January and August 2024, Microsoft, Meta, Google, and Amazon collectively spent $125 billion on AI data centers. Amazon, Alphabet, Meta, and Microsoft are expected to spend up to $700 billion in capital expenditures in 2026 to fuel their AI buildouts.
That level of infrastructure requires an enormous amount of power. Power generates heat. Heat requires cooling. Cooling requires water. That is the chain.
Why Do Data Centers Need Water?
The servers inside a data center generate heat constantly. Left unchecked, that heat would destroy the hardware. So data centers use cooling systems to keep temperatures in a safe range.
The dominant method is evaporative cooling. In this setup, hot air from the server floor passes over or through water. The water absorbs the heat and evaporates, carrying the heat away. Approximately 80 percent of the water drawn into an evaporative cooling system is lost to evaporation. The rest returns to local water systems, sometimes at higher temperatures and with chemical residues from the cooling process.
That discharged water is not harmless. Cooling water that returns to the supply carries a higher concentration of dissolved solids, including calcium, chloride, and silica. These can affect the taste of drinking water, lower crop yields, and be toxic to aquatic life.
AI workloads make this worse than traditional cloud computing. Power densities for advanced AI racks are scaling fast. In 2023, the average was between 25 kW and 40 kW per rack. By 2025, that figure is expected to exceed 200 kW, with densities of 1 MW per rack projected in the near future. More heat per rack means more cooling demand per square foot.
How Much Water Does a Data Center Use?
The numbers depend on the size of the facility, the cooling method, and the local climate. The scale is not small.
A typical 100 MW AI data center consumes 1.5 to 3.0 million cubic meters of water per year for evaporative cooling. To put that in context, the average American uses about 80 to 100 gallons per day. A single 100 MW facility can match the residential water use of a city of 50,000 people.
Data centers are already one of the top 10 water-consuming industries in the United States. Current growth rates suggest that by 2030, AI data centers could drain between 731 and 1,125 million cubic meters of water annually, equivalent to the household water usage of 6 to 10 million Americans.
Nationally, the trajectory is steep. Direct US data center water use sits around 17 billion gallons annually. The Lawrence Berkeley National Laboratory's 2024 federal report projects that figure will reach 38 to 73 billion gallons by 2028, driven primarily by AI training.
That figure only covers water used directly at the facility. The same Lawrence Berkeley report estimated that an additional 211 billion gallons were consumed indirectly through the electricity required to power those same data centers in 2023. The indirect figure is approximately 12 times larger than the direct one.
How Does AI Use Water Compared to Traditional Computing?
Traditional data centers that run web servers or store files also use water for cooling. What changed is the density and intensity of AI workloads.
A standard Google Search query uses roughly 0.3 watt-hours of energy. Per-query energy estimates for ChatGPT-class queries range from approximately 0.3 to 3 watt-hours, roughly 3 to 10 times more than a Google search, depending on model size and serving infrastructure.
More energy means more heat. More heat means more cooling. More cooling means more water. The math is direct.
Image generation is estimated at around 23 mL of water per generated image, based on energy consumption data from the UC Riverside and UT Arlington research team. AI workloads running today are not just slightly more demanding than what came before. They represent a structural shift in how much thermal load a single building has to handle.
How Much Water Does ChatGPT Use Per Query?
This is where things get genuinely confusing, because the honest answer is: it depends on what you are measuring.
The most widely cited figure comes from a research paper by Pengfei Li and colleagues at UC Riverside and UT Arlington, titled Making AI Less Thirsty. The figure for generating a 100-word email with ChatGPT is approximately 519 milliliters of water, close to a standard bottle.
Image Credit: Leonardo AI
OpenAI CEO Sam Altman pushed back publicly. Altman stated the average query uses just 0.3 mL, about one-fifteenth of a teaspoon. His figure accounts only for direct operational water at the data center level, not the water used in electricity generation or chip manufacturing.
The gap between 0.3 mL and 519 mL is not a measurement error. It is a scope decision. The 519 mL estimate includes indirect water use, the water consumed by power plants generating electricity for the data center. The 0.3 mL figure only counts water physically flowing through pipes at the facility itself.
Both numbers are defensible within their own definitions. The disagreement is about which costs should count.
In 2025, Google disclosed that each query to its Gemini assistant consumes 0.26 mL of water on-site. That is a direct, facility-level figure similar to Altman's claim for ChatGPT. It does not include the power plant water.
For most users, the practical takeaway is this: a single query is negligible. ChatGPT has over 100 million active users. Multiply any per-query number by billions of daily queries, and you get the real story.
The ChatGPT Data Center: What OpenAI Actually Runs On
OpenAI does not operate its own data centers. ChatGPT runs on Microsoft's Azure cloud infrastructure. The water and power footprint of ChatGPT is embedded inside Microsoft's broader data center operations.
A training run for GPT-4 in West Des Moines, Iowa, consumed 11.5 million gallons of water in July 2022 and 13.4 million gallons in August 2022. Across its five facilities in that city, Microsoft used 68.5 million gallons of water in 2024, making it the region's largest user.
That is the cost of training a single large model. Inference, actually running ChatGPT queries, adds to that continuously.
According to Google's 2025 Environmental Report, its data center water consumption grew from 4.3 billion gallons in 2021 to 7.7 billion gallons in 2024, an increase of nearly 80 percent in three years. Microsoft consumed approximately 1.69 billion gallons in FY2024, up from 1.26 billion gallons in FY2022, a rise of around 34 percent, even as its per-unit efficiency improved.
These are not companies being reckless. They are companies growing at a pace that their sustainability pledges have not caught up with yet.
Which AI Data Center Companies Use the Most Water?
The hyperscalers, companies that own and operate massive data center fleets, are the biggest consumers. Here is where the publicly available data lands.
Sources: Google 2025 Environmental Report (data centers, 2024); Microsoft 2025 Environmental Sustainability Report (FY2024); Equinix 2025 Sustainability Report (2024); Meta 2024 Sustainability Report (2023, latest disclosed DC total); Amazon does not disclose aggregate water consumption. All figures represent water consumed, not withdrawn.
Google's data centers consumed 7.7 billion gallons of water in 2024, up from 6.1 billion in 2023 and 4.3 billion in 2021. A single facility in Council Bluffs, Iowa, used 1 billion gallons in 2024 alone, more than any other Google site. In total, Google withdrew 7.8 billion gallons globally in 2024 across data centers and offices combined, consuming 78 percent of that. The company's data center water consumption has grown more than 70 percent since 2021. In 2024, Google replenished 4.5 billion gallons through water stewardship projects, bringing its freshwater replenishment rate from 18 percent in 2023 to 64 percent, with a stated goal of 120 percent by 2030.
Microsoft
Microsoft consumed approximately 1.69 billion gallons (6.4 million cubic meters) of water across its global operations in FY2024, according to its 2025 Environmental Sustainability Report. The company has pledged to become water positive by 2030. In August 2024, it began deploying a closed-loop, chip-level cooling system that eliminates evaporative water use. Once filled at construction, the system recirculates coolant continuously, avoiding an estimated 125 million liters of water consumption per facility each year. Pilot sites are running in Phoenix, Arizona, and Mt. Pleasant, Wisconsin, with all new data center designs from August 2024 onwards following the zero-water approach. Microsoft also reduced its WUE from 0.49 L/kWh in 2021 to 0.30 L/kWh in FY2024, a 39 percent improvement, though total consumption continued to rise as new capacity came online.
Equinix
Equinix, which operated 268 data centers worldwide in 2024, reported withdrawing 1.4 billion gallons and consuming 1.2 billion gallons, or 85 percent of what it withdrew.
Meta
Meta consumed 813 million gallons of water globally in 2023, its most recent fully reported figure, with 95 percent of that attributable to data centers. A single facility in Newton County, Georgia, consumes around 500,000 gallons per day, roughly 10 percent of the county's supply. Meta's newest facility in Richland Parish, Louisiana, its largest worldwide, is registered to draw up to 8.4 billion gallons per year at maximum permitted capacity, though Meta states actual annual use will run between 500 and 600 million gallons. Meta's data centers run on 100 percent renewable energy. In 2024, the company's water restoration projects returned more than 1.59 billion gallons to high and medium-water-stress regions.
The Water You Cannot See: Embodied Water in AI Chip Manufacturing
Almost every article on this topic stops at operational water use. Nobody traces the water consumed before a server ever boots.
Every AI training run depends on thousands of high-end GPUs, and those chips carry a water debt long before they reach the data center. TSMC, the world's dominant semiconductor manufacturer, consumes roughly 150,000 metric tonnes of water per day across its Taiwan operations. An average chip fab uses approximately 10 million gallons of ultrapure water daily, equivalent to the daily household consumption of 33,000 Americans, according to the World Economic Forum.
The water intensity scales with chip sophistication. When TSMC moved to 16 nm process nodes in 2015, water usage per unit increased by over 35 percent, according to S&P Global Ratings. More advanced nodes require more ultrapure water per wafer because the fabrication process involves more steps, each of which requires rinsing. It takes roughly 1,400 to 1,600 gallons of municipal water to produce 1,000 gallons of ultrapure water to meet chipmaking demands.
TSMC Arizona's first fab in Phoenix already uses 4.75 million gallons per day. When all three planned Phoenix fabs are operational, the combined daily demand is projected at 17.2 million gallons, before the company's Industrial Reclamation Water Plant, expected online in 2028, reduces that through recycling.
The 2021 drought in Taiwan illustrated what this dependency looks like under stress. Water shortages forced Taiwan's water authority to reduce agricultural supplies to prioritise chip production. The semiconductor supply constraints that followed affected electronics prices globally, including GPU availability for AI labs. From 2015 to 2019 alone, TSMC's total water consumption surged by 70 percent as chip complexity increased.
No AI company currently reports the embodied water in the chips they buy. The water numbers in every sustainability report represent the operational floor, not the true lifecycle cost. A full accounting that included chip manufacturing and facility construction would be substantially higher for every company on this list.
When Water Efficiency Metrics Lie: The WUE Problem
WUE, Water Usage Effectiveness, is the industry's primary tool for measuring how efficiently a data center uses water. Developed by The Green Grid and reported in liters per kilowatt-hour, a lower number signals better efficiency. Zero is the theoretical ideal, achievable only in fully air-cooled facilities.
The metric sounds authoritative. But there are structural problems with how it gets reported that most coverage ignores entirely.
WUE only counts water at the facility boundary. A data center that purchases chilled water from a third-party municipal cooling utility does not count that water in its WUE, even though the same water is being consumed to serve that facility's cooling needs. The scope of the measurement can be drawn differently by different operators, and there is no uniform regulatory standard enforcing a consistent definition.
Companies also report annual average WUE, which masks seasonal spikes. A facility in Phoenix might run a WUE of 0.5 L/kWh in January and 4.0 L/kWh in August. The annual average looks acceptable. The local aquifer in summer does not. Monthly or quarterly reporting provides a far more accurate picture of real-world water stress on local communities, but annual disclosure remains the industry norm.
Some colocation operators lease space to hyperscalers and do not report water consumption in their own disclosures. The hyperscaler reports server-level efficiency. The colo reports building-level performance. Nobody publishes the combined figure. Sustainability reports from Amazon do not disclose total water consumption at all. Microsoft does not break down water use by individual facility. Only Google provides individual data center figures, and even those lack context on facility size and cooling technology used.
Claims of "zero-water" or "water-free" cooling also deserve scrutiny. Most such claims refer to eliminating evaporation at the facility level. The heat absorbed by closed-loop liquid cooling systems still has to go somewhere, often to dry coolers or rooftop heat exchangers that may use water depending on ambient temperature. Describing a facility as zero-water because the cooling tower is indoors rather than outside does not change the physics involved.
| Boundary Definition | What Gets Counted | What Gets Excluded |
|---|---|---|
| Facility-level WUE (industry standard) | On-site cooling tower water, humidification | Third-party chilled water, grid electricity, water, and chip manufacturing |
| Campus-level WUE | All water across a campus, including shared cooling plants | Upstream electricity generation, embodied water in hardware |
| Full lifecycle WUE | Direct, indirect (grid), and embodied (hardware) water | Rarely reported; no industry standard exists |
Source: The Green Grid WUE framework; Lawrence Berkeley National Laboratory 2024 Data Center Energy Usage Report
An impressive WUE score can coexist with serious local water stress, depending entirely on how the measurement boundary is drawn.
Why Data Centers Keep Getting Built in Water-Stressed Regions
Coverage of this issue focuses almost entirely on what AI companies consume. The more revealing question is why they keep building in dry places when wetter places exist.
Water availability sits below several other variables in most site selection decisions. Land cost, fiber proximity, permitting speed, labour market depth, and tax incentives consistently rank higher. Water-rich regions in the upper Midwest and Great Lakes area were slower to offer competitive incentive packages, so they lost the builds to Arizona, Texas, and Virginia.
Arizona's incentive structure has been particularly aggressive. The state enacted data center sales and use tax exemptions in 2013 and extended them through 2033. Companies qualifying for the exemption save roughly 9 cents on every dollar spent on equipment in the Phoenix area. The incentive cost Arizona $38.5 million in foregone tax revenue in fiscal year 2025 alone, up from $1.4 million in 2020. By fiscal year 2027, that figure is projected to exceed $60 million annually, according to the Grand Canyon Institute.
The regulatory window dynamic compounds the problem. Once a company has a permit and a shovel in the ground, water restriction changes rarely force a relocation. Developers who moved fast in Phoenix, Goodyear, and Mesa before municipal water caps took effect are effectively grandfathered. Arizona Governor Katie Hobbs called for repeal of the tax exemption in her 2026 State of the State address, acknowledging she voted for it as a state senator in 2013. Unwinding it requires a two-thirds legislative supermajority.
There is no federal mandate requiring water impact assessments before data center construction. Environmental review requirements vary by state and are often waived for economic development projects above a job threshold. The financial incentives are real. The water costs are diffuse, slow-moving, and land on communities that had no vote in the site selection decision.
The industry is not building in water-stressed zones despite the risk. It is building there because the financial incentives outweigh the risk in the short term, and the regulatory framework, as currently structured in most US states, allows it.
Image Credit: Leonardo AI
What AI Companies Actually Mean When They Say Water Positive
Several of the largest AI infrastructure operators have made headline commitments to water positivity. Microsoft has pledged to become water positive by 2030. Google has committed to replenishing more water than its data centers consume by the same date. Amazon made a similar pledge at re: Invent 2024. These commitments deserve scrutiny before they are treated as achievements.
Myth: Water positive means the company physically puts more water back than it takes
Most water positivity commitments are met through purchasing water replenishment credits, not through on-site physical restoration. Water consumed in Mesa, Arizona, is "offset" by funding a watershed project in a different state or country. The local aquifer in Mesa receives no benefit. A Kairos Fellowship investigative report found that Google is using annual savings from water stewardship projects elsewhere to offset actual consumption figures in its sustainability reporting.
Myth: Running on 100 percent renewable energy means water-neutral operations
Many renewable energy sources are water-intensive. Nuclear, geothermal, and some solar thermal plants consume significant water. Hydropower loses water through reservoir evaporation. Running on "100% renewable" says nothing about the water consumed to generate that power. The link between energy source and water intensity is not as clean as the marketing suggests.
Myth: Liquid cooling solves the water problem
Liquid cooling reduces on-site evaporative water use, sometimes to near zero. But closed-loop systems still require makeup water for losses over time. The heat the fluid absorbs still has to be rejected somewhere, often to outdoor dry coolers that may use water at high ambient temperatures. Immersion cooling uses dielectric fluids that carry their own environmental handling requirements at the end of life. The transition from evaporative cooling to liquid cooling is genuinely meaningful progress. Treating it as a complete solution overstates what the engineering currently delivers.
Myth: Efficiency gains will offset growth
The data so far does not support this. Google's WUE improved between 2021 and 2024. Water consumption increased 88 percent in the same period. Microsoft reduced its WUE by 39 percent between 2021 and 2024 while its overall water consumption rose. Efficiency gains are real. Capacity additions are outpacing them by a substantial margin.
| Claim | What it actually means | What it excludes |
|---|---|---|
| Water positive by 2030 | Plans to purchase replenishment credits equal to consumption | Local watershed impact: indirect and embodied water |
| 100% renewable energy | Energy sourcing is matched to renewable certificates | Water consumed at renewable generation sites |
| Zero-water cooling | No evaporative water loss at the facility level | Heat rejection systems, makeup water, fluid manufacturing |
| WUE of 0.30 L/kWh | Annual average facility-level efficiency | Seasonal peaks, third-party cooling water, and upstream water |
Sources: Microsoft 2024 Sustainability Report; Google Environment Report; Kairos Fellowship data center water analysis; The Green Grid WUE framework
Thermal Density and the Cooling Threshold AI Has Already Crossed
This section is technical. If you already understand the basics of data center cooling, this is where current AI infrastructure gets genuinely different from anything built before.
The relationship between rack density and cooling water demand is not linear. Going from 10 kW to 40 kW per rack does not quadruple water use. It more than quadruples it, because cooling efficiency drops as heat load increases and ambient temperatures rise. The physics of heat transfer gets less forgiving at higher densities, not proportionally harder.
At around 100 kW per rack, air cooling reaches a thermodynamic limit regardless of airflow volume. The air simply cannot carry enough heat away fast enough. This is not a design preference or a cost tradeoff. It is a physical ceiling. Facilities built or planned for air cooling before AI workloads arrived are now structurally underequipped. Many data centers that were considered modern in 2019 cannot safely host current-generation GPU clusters without retrofitting.
The specific challenge in mixed-use colocation facilities is worse. A colo running a combination of traditional cloud workloads at 10 to 20 kW per rack and AI GPU clusters at 100 to 200 kW per rack cannot cool uniformly. Hot spots develop. Water-based spot cooling has to be retrofitted around existing raised floor infrastructure, conduit runs, and structural columns, often at high cost and with performance compromises. The building was not designed for the thermal geography that AI creates.
NVIDIA's GB200 NVL72 rack, which houses 72 GB200 GPUs in a single cabinet, operates at densities that effectively make liquid cooling a baseline requirement, not an upgrade option. The transition from optional to required happened faster than most facility planners expected. Several colocation operators signed long-term leases with hyperscale customers at density commitments that they are now technically unable to meet without capital investment they did not budget.
There is also a temperature quality problem that gets almost no coverage. Waste heat reuse, one of the industry's most promising solutions, depends on producing output water that is hot enough to be useful for district heating or industrial processes. Facilities optimised purely for low WUE at moderate rack densities often produce warm water at 35 to 40 degrees Celsius. District heating networks typically need water at 60 to 80 degrees Celsius to be useful. At higher rack densities and with direct liquid cooling, output water temperatures can reach 60 degrees or above, making it genuinely reusable. The data centers that could contribute to waste heat recovery are exactly the ones running the densest, most water-intensive AI workloads.
| Rack density | Viable cooling method | Approximate water demand | Notes |
|---|---|---|---|
| Up to 10 kW | Air cooling | Low to moderate | Standard cloud workloads, storage, and web servers |
| 10 to 40 kW | Air cooling with supplemental water | Moderate | Most pre-AI hyperscale builds |
| 40 to 100 kW | Rear-door heat exchangers, in-row cooling | High | Early AI training clusters; current inference workloads |
| 100 kW and above | Direct liquid cooling is required | Low if closed-loop; high if open evaporative | Current frontier AI training; Nvidia NVL72, GB200 rack form factors |
Source: Arup 2025 Foresight Report; ASHRAE thermal guidelines; Nvidia GB200 NVL72 product specifications
Are Data Centers Bad for the Environment?
The honest answer: it depends on how you weigh the costs against the benefits, and where the facility is located.
On the carbon side, the International Energy Agency estimates that data centers produced approximately 182 million tons of CO2 in 2024, representing about 1 percent of global energy-related emissions. A 2024 study of 2,132 US data centers found their average carbon intensity was 48 percent higher than the national average across all economic sectors. That is because many facilities are in regions where the electricity grid still runs heavily on coal and natural gas.
At the current growth rate for AI, approximately 24 to 44 million metric tons of carbon dioxide will enter the atmosphere, equivalent to adding 5 to 10 million cars to US roadways, according to a 2025 report by Cornell University researchers.
On the water side, location matters enormously. A data center in water-rich Iceland or Norway drawing from hydropower carries a very different footprint than one in drought-prone Arizona or New Mexico.
In Loudoun County, Virginia, the densest data center cluster in the world, facilities used 1.6 billion gallons in 2023, roughly 10 percent of all water consumed in the county, after growing 250 percent in just four years.
The Netherlands permanently banned hyperscale data centers above 70 MW effective January 2024. Singapore's Green Data Centre Roadmap set a target WUE of 2.0 m3/MWh and requires a PUE of 1.25 or better for new builds. In the US, over 190 data center bills were introduced across state legislatures in 2025, and Arizona municipalities are imposing water caps that have already forced developers to commit to zero-water cooling designs.
The environmental picture is also not uniform across communities. Data centers are often located in regions with cheaper land and lower regulatory barriers. These areas tend to have lower-income populations who bear the brunt of the resulting air and water impacts. Unlike carbon emissions, the health effects caused by a data center drawing from one region's aquifer cannot be offset by cleaner water elsewhere.
What Is Being Done to Reduce Data Center Water Usage?
The industry knows the problem. Several solutions are already in deployment.
Liquid cooling and immersion cooling
Liquid and direct-to-chip cooling systems can reduce water use by up to 90 percent compared to traditional evaporative methods. Instead of using water to cool air that then cools servers, these systems bring liquid directly into contact with the chips. Closed-loop systems lose almost no water during operation. Microsoft's Fairwater design, piloting in Phoenix and Mt. Pleasant since 2024, is the most publicly documented example of this transition at scale.
Geographic placement
Geographic optimisation, locating facilities in naturally cooler or water-abundant regions, is becoming a key design principle for sustainable AI infrastructure, according to Arup's 2025 Foresight Report on Water-Conscious Data Centers. Finland, Sweden, and Iceland have attracted significant data center investment precisely because ambient temperatures reduce cooling demand for most of the year.
Waste heat reuse
European researchers published findings in March 2026 showing that data center waste heat can be redirected to power water purification and carbon capture, potentially making facilities water-positive in a meaningful physical sense. That would mean a data center that produces more usable water than it consumes, a complete reversal of the current model. The viability depends on producing output water hot enough for industrial use, which current high-density AI clusters are beginning to achieve.
Scheduling and timing
Companies, including Google and Microsoft, are already training AI models at night or in cooler regions to reduce cooling demand and evaporation losses. Cooler ambient temperatures reduce how hard the cooling system has to work. A 10-degree Fahrenheit drop in ambient temperature can reduce cooling water consumption by 15 to 25 percent at a facility using evaporative systems.
Industry collaboration
Microsoft, Google, Amazon, and Meta have signed on to the Data Center Innovation Initiative, led by nonprofit investor Elemental Impact, aimed at funding up to 10 startups developing technologies around cooling, energy storage, and low-carbon building materials. These solutions exist. The constraint is deployment speed relative to the pace of AI expansion.
AI Data Center Power: The Electricity Side of the Problem
Water and electricity are linked in AI infrastructure. More power consumed means more heat generated, which means more water is needed for cooling.
National electricity consumption by data centers is projected to grow from roughly 4 percent to 12 percent of total US electricity use by 2030. In raw terms, US data center electricity consumption is expected to rise from 183 terawatt-hours in 2024 to 426 TWh by 2030.
The surge in demand has led to historic price increases in regional energy markets. The July 2024 capacity auction in the PJM regional market cleared at roughly $270 per MW-day, an 800 percent increase from the previous year's price of $30. By July 2025, prices reached the market cap of nearly $330 per MW-day.
In 2025, Microsoft, Google, Amazon, and Meta are projected to spend a combined $320 billion on AI infrastructure, more than double the $151 billion spent in 2023. The power grid in many parts of the US was not designed for this kind of concentrated, high-density demand. Data center clusters in Virginia, Texas, and Arizona are already straining local grid capacity. Water scarcity in those same regions compounds the problem.
The Global Picture: Water Scarcity and AI on a Collision Course
The regions where AI data center growth is fastest are often the regions with the least water to spare.
Arizona has over 370 golf courses in a desert climate, so data centers are hardly the first industry to make questionable water choices there. The scale of AI growth is different. Local municipalities have already started imposing caps. By 2030, global data center water consumption is projected to exceed 1.2 trillion liters annually, surpassing the total annual water use of London's 9 million residents.
The Li and Ren research paper projects that global AI demand will require somewhere between 4.2 and 6.6 billion cubic meters of water withdrawal annually by 2027. For reference, the entire country of Australia uses roughly 74 billion cubic meters per year across all sectors.
The IEA's Energy and AI report from April 2025 and analysis from MSCI covering 680 data center assets worldwide both flag water scarcity as a growing physical risk for the sector, not just an environmental concern, but an operational one. If the water is not there, the cooling does not work. If the cooling does not work, the servers do not run.
What the Numbers Actually Tell You
AI data centers are real infrastructure with real physical footprints. They consume water, generate heat, draw electricity, and discharge warmer, chemically altered water back into local supplies. That is how the cooling physics works right now.
The companies building this infrastructure, AWS, Google, Microsoft, Meta, Equinix, Digital Realty, and others, are not ignoring the issue. Immersion cooling, closed-loop systems, waste heat recovery, and geographic optimisation are all real investments. The question is whether the pace of deployment matches the pace of expansion.
The data says it does not, at least not yet. Water consumption is rising faster than efficiency improvements are cutting it. That may change. The engineering is improving, the regulatory pressure is real, and the economic incentive to reduce cooling costs is significant. The gap between where the industry is headed and where it needs to be is large enough that it deserves honest attention.
Every major AI company publishes annual sustainability reports. If you want to hold them accountable, those reports are the right place to start. The numbers are in there. Most of the time, they are not flattering.
The water conversation around AI has a transparency problem. Most major AI companies report only the water they directly evaporate on-site, excluding the water their chip suppliers consume, the water used to generate the electricity that powers their facilities, and the seasonal peaks that make annual averages look more acceptable than the underlying reality. When OpenAI and Google report per-query water figures in the fractions of a milliliter, those numbers are defensible only within a deliberately narrow boundary. The UC Riverside research team that produced the 519 mL figure for a 100-word ChatGPT response was not being alarmist. They were counting more of the supply chain.
The location arbitrage story is equally underreported. The same companies publicly committed to water stewardship are building in Phoenix, Mesa, and Northern Virginia because financial incentives and permitting speed outweigh water risk in their internal site selection models. Arizona extended its data center tax exemption through 2033 in 2021, and the incentive structure is now so entrenched that even the governor who supported it is having difficulty unwinding it. The communities bearing the water stress had no meaningful input into those decisions.
Two facts can be true simultaneously. Microsoft's closed-loop cooling announcement in 2024 is genuine progress. So is the reality that Microsoft's total water consumption rose roughly 34 percent between FY2022 and FY2024, reaching 1.69 billion gallons, and that its West Des Moines facilities used 68.5 million gallons in 2024 alone. Efficiency gains and consumption growth are running in parallel, and right now, consumption is winning.
The most useful thing a reader can do with this information is treat corporate sustainability commitments the way you would treat a financial forecast: look at the methodology before accepting the headline number. Water positive by 2030 means very different things depending on whether the offset is local and physical or a credit purchased in a different watershed. Both are reported the same way in most sustainability disclosures.