In June 2026, SpaceX revised its prospectus ahead of the IPO.
The revisions don’t touch on rocket tech, satellite internet, or Mars colonization plans. The new risk warning points to something more down-to-earth—water. The document outlines that water shortages, droughts, local water competition, or regulatory restrictions could prevent the company from securing enough cooling water, which could slow down data center expansions or even force the adoption of pricier alternative cooling solutions. Power, processors, and water are all listed together in the document, representing the core resource constraints for AI computational expansion.
This is the first time SpaceX has systematically emphasized water resource risks in public documents. A company known for its Starship and Falcon rockets is reminding potential shareholders to pay attention to the stability of their water supply.
The original wording of that risk warning was: 'water scarcity, drought conditions, competition for local water resources, or regulatory restrictions on water use could limit our ability to obtain sufficient water for cooling... delay or limit expansion... or require us to implement alternative cooling techniques that may be more costly.' The language is bland, reflecting the restrained tone of standard legal documents. But the mere appearance in an IPO filing is a signal.
SpaceX's AI business is xAI. According to a TechCrunch report from May, xAI had an operational loss of $6.4 billion in 2025, with annual revenue of $3.2 billion, and capital expenditures continuing to soar. This rate of burn corresponds to a mad rush of data centers, servers, and computing clusters. As tens of billions of dollars in hardware and infrastructure investment become the norm each quarter, any fluctuation in the supply of physical resources is no longer just an operational cost item to be optimized. It becomes a risk that must be accounted for to investors.
Water has shifted from operational costs into the risk disclosure framework, and this displacement is worth noting.
Operational costs are something companies can control; they can save money or change technical solutions. But risks are different. Risks are external variables, things that companies may not necessarily control. Drought is a weather issue, tightening local government water use permits is a policy issue, and community opposition is a political issue. These problems are not easily solvable by throwing money at them.
TechCrunch pointed out in its report that this revision reflects the AI industry’s reliance on natural resources is drawing new attention from regulators and investors. Analyzing this judgment requires answering a more fundamental question: How much water do AI data centers actually use?
17 billion gallons is just the portion for direct cooling.
Lawrence Berkeley National Laboratory provided a set of estimated data. In 2023, the water consumption for cooling in U.S. data centers was approximately 17 billion gallons, which is about 64 billion liters. This is just direct cooling. Data centers rely on electricity, and the power generation process itself consumes a lot of water too. The cooling for thermal and nuclear power, along with the evaporation from hydropower, adds indirect water use, with one estimate showing it could reach as high as 211 billion gallons.
17 billion gallons of direct water usage and 211 billion gallons of indirect water usage. The latter figure is over 12 times the former. When discussing the water footprint of AI, the direct cooling data seen is merely the tip of the iceberg.
This estimate also provides a trend forecast: by 2028, the direct cooling water usage of U.S. data centers may double to quadruple. The wide range in numbers is due to variables such as the pace of AI computing power expansion, the choice of cooling technology, and the distribution of new data center sites. Doubling is the most conservative scenario, while quadrupling is an aggressive expansion scenario. Regardless of which, the direction is sharply upward.
These numbers themselves are abstract. When applied to specific companies, the sense of scale becomes clearer.
Google disclosed in its sustainability report that it consumed 6.4 billion gallons of water in 2023, with 95% used for data centers. This means Google's data centers alone drank about 6 billion gallons that year. One site, the data center in Council Bluffs, Iowa, alone consumed 1 billion gallons of drinking water in 2024.
Meta's figures are slightly smaller but still considerable. In 2023, Meta consumed 813 million gallons of water globally, 95% of which was also from data centers.
Putting these numbers together, the water consumption of a single data center from Google is roughly equivalent to over one-third of the entire U.S. data centers' direct cooling water usage estimated by Lawrence Berkeley Lab. The site in Council Bluffs, Iowa, can consume enough water annually to support a mid-sized city.
Where is all this water going?
Most large data centers use evaporative cooling technology. The principle is not complicated: water comes into contact with hot air in cooling towers, evaporating and taking away heat, becoming water vapor released into the atmosphere. This process is called 'consumptive use of water.' The water is used up and does not return to rivers, lakes, or underground aquifers. This differs from residential water use, where shower and dishwater can be treated and returned. The cooling towers of data centers discharge steam. Once consumed, it’s literally consumed.
(Nature)'s journal npj Clean Water published a study in 2021 providing a technical scale: a typical 1-megawatt IT load data center, using traditional evaporative cooling technology, consumes about 25.5 million liters of water annually. A 1-megawatt IT load corresponds to the computing power of several hundred servers. And large data centers often operate at tens or even hundreds of megawatts. When scaled up, a 50-megawatt data center cluster can easily consume over a billion liters of water annually for cooling.
In drought regions, this level of consumption means what it says without further explanation.
Building water guzzlers on the edge of the desert.
In April 2025, an investigative report by the British (The Guardian) pointed out that Amazon, Microsoft, and Google are operating and expanding data centers in some of the driest regions globally, with the scale of the three companies’ data centers expected to expand by 78%. Behind these numbers are a series of ongoing conflicts.
In Querétaro state, Central Mexico, 17 out of 18 towns are suffering from severe drought. Meanwhile, the state has become a cluster for data centers of international tech giants. Local residents held signs outside data centers saying: 'No queremos centros de datos, queremos agua' — we don't want data centers, we want water. BBC provided detailed coverage of this conflict.
In Mesa, Arizona, according to a Business Insider report in June 2025, Meta reached a water agreement allowing its facilities to use up to 4 million gallons of water daily. What does 4 million gallons mean? Based on the average daily water usage of approximately 82 gallons for U.S. residents, that's equivalent to the daily water usage of nearly 49,000 people. Arizona itself is one of the most water-scarce regions in the U.S., with the Colorado River's water level declining year after year, leading to ongoing disputes between states over water allocation. A data center drawing 4 million gallons of water daily, legally and compliantly, does not mean it is without controversy.
Similar voices have emerged in Australia. (The Guardian) reported in December 2025 that as the construction of large-scale data centers accelerates, drinking water supplies in some regions are facing direct competition. In developed countries with relatively mature water resource planning systems, the sudden spike in data center water usage has brought significant shocks, illustrating that this is not an isolated incident of governance but a widespread contradiction between scale expansion and resource limitations.
These controversies share a commonality not in tech companies 'violating water use.' They have committed no violations. Every water use agreement has been legally approved, and every water fee has been duly paid. The root of the problem is that the existing water resource allocation framework was established in an era when data centers had not yet become major water users. When the daily water usage of a data center equals that of a town, compliance itself becomes an issue. The system has not kept pace with the growth rate of water guzzlers.
A report from (The Guardian) in October 2025 revealed another dimension. Amazon has long refused to disclose detailed water usage at its data centers, accused of strategically concealing its complete water footprint. Google disclosed single site data, Meta released aggregated global data, while Amazon provided the least information. This disclosure differentiation has begun to be viewed by analysts as a variable in risk assessment. The less a company is willing to tell you about its water usage, the more likely its water consumption is to spark controversy.
Projects are getting blocked, and water is the reason.
Water resource controversies are no longer limited to public opinion. They are substantively blocking project implementation.
A report from Data Center Watch indicates that over the two years since mid-2024, approximately $64 billion worth of data center projects in the U.S. were halted or delayed due to local community opposition. Water resource consumption is one of the core reasons for protest, alongside electricity usage and noise pollution. The report documented 142 bipartisan grassroots opposition organizations spread across different states in the U.S., unified in their rare consensus against giant data centers.
Water is becoming a new weapon for NIMBY (Not In My Back Yard) effects. In the past, NIMBY movements mainly revolved around substations, waste disposal plants, and highways. Now data centers have joined this list. The reasons have changed, but the logic remains the same. Residents’ logic is simple: you say your data center contributes to the economy, but if the cost is a decrease in my household water pressure, an increase in my water bill, and a drop in well water levels, that’s a price I’m not willing to pay.
Once this opposition forms, it's not something a company can resolve by just giving a few community talks or promising a couple of jobs. Electricity can be generated from new power plants, fiber optics can be laid out anew, and land can be bought at a premium. But water, in the eyes of residents, has no substitutes. When there's no alternative, there's hardly any room for negotiation.
Throughout 2025, industry statistics show that about half of the data center projects initially planned for launch in 2026 were canceled or delayed. This ratio is enough to prompt any company planning to expand AI infrastructure to reassess site selection logic. Previously, the order of data center site selection was: electricity, fiber optics, land cost, climate. Now, the importance of water is catching up.
The University of California, Berkeley's Center for Law and Energy published a specialized report in February 2026, exploring how to regulate water use in California's data centers. This is the first time academia has directly addressed this issue in a specialized report. The release of this report itself is a signal: when top law schools and energy policy think tanks begin to systematically study the regulatory framework for data center water use, it indicates that this issue has crossed the boundaries of internal industry discussion and entered the public policy agenda.
Investors are starting to calculate water costs.
The capital markets are following suit.
In April 2026, according to the Journal Record, investors formally urged Amazon, Microsoft, and Google to disclose more data regarding water usage in their data centers. The report simultaneously cited a set of macro data: North American data centers had already used nearly 1 trillion liters of water by 2025.
1 trillion liters is a figure that is hard to intuitively grasp. To express it in another way: it is roughly equivalent to the storage capacity of a large freshwater lake. Lawrence Berkeley Lab's estimate for 2023 is already substantial, but it may skew conservative in light of actual consumption in 2025.
The shift in investor sentiment towards communities is traceable. In the past, water resources appeared in ESG reports alongside other environmental indicators, often just a form to be filled out by the corporate social responsibility department. That’s changed now. Water resources have jumped from the 'corporate image' section to the 'operational risk' section. Shareholders are now concerned not about environmental issues but whether there is enough water to keep servers running. When the stability of water supply starts to impact revenue expectations, it is no longer an ESG topic but a financial one.
Different companies are showing significant differentiation in their response strategies. Google continues to release water usage data for individual sites, claiming in 2024 that it returned 4.5 billion gallons of water through water replenishment projects. Meta publishes aggregated data. After the investigation from (The Guardian), Amazon has yet to disclose detailed water usage at its sites. This differentiation further reinforces the viewpoint: the transparency of water usage data itself is becoming a variable for analysts to assess the risk exposure of AI infrastructure companies.
Companies are also trying to respond on a technical level. Switching to air cooling can reduce direct water use but often increases electricity consumption. Liquid cooling technology can use higher temperature water (NVIDIA’s Vera Rubin platform supports cooling with water at 45°C), but the deployment costs are higher. Each technical route weighs the trade-offs between water consumption and electricity usage, with no perfect universal solution. Ultimately, the cooling solution chosen for a data center may not be the most technically optimal but rather dictated by local water prices, electricity prices, and policy tolerance. Technical choices become compromises under resource constraints.
A sarcastic contrast.
In March 2026, OpenAI CEO Sam Altman said a widely circulated statement during a public speech. According to Business Insider, he stated: 'We see a future where intelligence becomes a utility like electricity or water, and people pay us based on usage.'
This statement sparked extensive discussion on copyright and business models, but it carries a more personal implication. When Altman compared AI to water and electricity, the actual physical operation of AI is consuming real water in the real world. The industry's imagination of business models is packaging AI as an inexhaustible infrastructure, charging by the amount like turning on a faucet. Meanwhile, SpaceX's prospectus is candidly admitting: without enough water, AI may not run.
Before a service is compared to water and electricity, its infrastructure has already racked up huge bills for water and electricity. This contrast itself is the most accurate description of the AI industry’s state of awareness in 2026.
Looking back at the timeline, the narrative path is quite clear.
From 2023 to 2024, annual water usage data from the world's top cloud providers will be passively disclosed through sustainability reports. Lawrence Berkeley National Laboratory released estimates that provide a macro view of water consumption across U.S. data centers for the first time. Community conflicts in places like Querétaro, Mexico, and Mesa, Arizona, are starting to enter mainstream media coverage.
In 2025, (The Guardian) and BBC conducted systematic investigative reports linking the expansion of data centers in drought-affected areas with local water pressures into public discussion. Data Center Watch released quantified statistics on $64 billion in stalled projects. Investors began formally requesting increased transparency regarding water footprints.
In 2026, SpaceX pulled this chain from public discussion and industry reports and placed it into the 'risk factors' section of its IPO prospectus. This marks the formal transition of water issues from being a public discourse topic to an investment pricing factor. An investor purchasing SpaceX shares must sign to confirm they are aware that the company’s AI operations may face issues due to water shortages.
The capital market's pricing approach to resource constraints is just like this. It doesn't care about sentiment, corporate commitments, or sustainable development visions in PR materials. It only cares about one thing: what factors under what conditions could lead to a shortfall in expected returns. Water supply can be affected by weather, water prices can be influenced by policies, and access to water can be hindered by community opposition—these three factors are beyond a company's control. What cannot be controlled is risk. Risks need to be written into documents to inform investors.
This mechanism itself is reshaping the logic of AI infrastructure expansion.
In recent years, the narrative around AI competition has primarily focused on a computing power arms race. Chips, electricity, and talent are the three key elements. Water has been a hidden condition, assumed to be available locally. Now, that assumption is shaky. In drought-prone areas, in cities where local water is already scarce, and where regulations are tightening water quotas, 'local water availability' is no longer an assumption that can be taken for granted.
The expansion of AI infrastructure is no longer just a game of technology and capital. It has entered a phase where resource allocation needs to be negotiated simultaneously with local residents, local governments, regulatory bodies, and investors. The pace of the computing power race may not be determined by the fastest company but by the slowest water meter.