loading
loading
After ChatGPT was released in November 2022, the tech industry was quick to herald the chatbot as the marker of a new wave of technological progress. As companies pushed to create more powerful AI models, the water demand of their ambitions took a while to become clear. In April 2023, a research paper out of the University of California estimated that training the large language model behind the initial version of ChatGPT had required 185,000 gallons of water. For every extended conversation someone had with the chatbot or every time they asked it to write an email, they were effectively pouring a bottle of water down the drain. That statistic got a lot of play when it was released. It became an early warning sign of what was coming as companies pushed chatbots as a universal technology that could replace psychologists, educators, lawyers, and even people’s friends or partners. As public concern grew, Google and OpenAI came out with their own figures. The companies suggested that querying one of their chatbots used mere drops of water. Independent researchers were not impressed, arguing the companies were taking a narrow view of the water use necessary to power and cool increasingly monstrous data centers. Those water demands can be quite significant too. A hyperscale data center can suck up between 1 and 5 million gallons of water a day, comparable to the amount used by a city of 10,000 to 50,000 people — depending on the part of the world we’re talking about. Consider how these companies often cluster data centers together and it becomes obvious what a challenge it can be to supply them. The water demands of individual data centers are also continuing to grow. It’s not just that they need more water for cooling as they become larger and contain more servers. The graphics processing units (GPUs) that power generative AI are much more resource-intensive than the chips and storage devices that traditionally filled data center racks. GPUs run harder and throw off more heat, meaning they also need more energy or water to keep them cool. Data centers cannot just use any kind of water either; it has to be clean enough to ensure the equipment and pipes are not damaged. Tapping into public drinking water is both easier and cheaper than paying the cost of treatment themselves. That means data centers end up competing with communities over their potable water supply. If we want to look at a part of the world where the issue of data center water use has taken center stage, there are few better countries to turn to than Chile. In October 2019, mass social demonstrations broke out across the country. One of the movement’s driving slogans was about water. It was poor timing for Google as it was fighting for approval and public support of its latest data center project, a $200 million data center project in Cerrillos, on the southwest side of Santiago. The company made a pitch similar to the one residents in communities the world over have heard whenever a data center is proposed: that the facility would bring jobs and prosperity, and give the area a leg up on further tech investments. The reality was not so straightforward. Google’s new data center would bring a lot of short-term construction jobs but not much long-term employment. And it would need a lot of water — something Cerrillos did not have much of to spare. Local residents were not as passive as the company had hoped. A group of concerned citizens established an organization called MOSACAT, which stands for Community Socio-Environmental Movement for Water and Land, and started poring over whatever information they could get their hands on to try to understand what the project would mean for them. “Usually these reports use quite technical language, and this is a working- class community where not many people are lawyers, for example, or experts on ecology,” explains Sebastián Lehuedé, a researcher at King’s College London who is from Chile and spent time in Cerrillos. According to its own projections, the data center Google wanted to build in Cerrillos was poised to use 169 liters of water every single second. In just one day, the data center would consume as much water as all 80,000 residents of Cerrillos use for drinking in a year. To make matters worse, it wanted to draw that water from an underground aquifer that had been under a decree that restricted how much water could be removed since 2005. Just weeks before the project was rubber-stamped by the environmental authority, that decree had been upgraded to a complete prohibition on new extraction. The water use estimates set off alarm bells for the residents. Chile had already been in an unprecedented drought since 2010. By 2021, more than half the country’s 19 million residents lived in areas experiencing severe water scarcity, forcing hundreds of rural communities to rely on tankers to deliver drinking water. The local activists ultimately built the case against one of the largest companies on the planet. After months of meetings and advocacy to spread awareness, they were able to put a referendum question to the residents of Cerrillos about the project in December 2019. The results were damning: 49% of voters chose to reject the project, while 38% wanted it to go forward. After both the municipality and community members filed claims against Google in 2020, the company returned with a new proposal in February 2022. It offered to cool the data center exclusively with air-conditioning rather than with water. Across the city in Quilicura, opposition was building too. Microsoft announced its intention to build a data center in the community in 2022, the same year strict water-rationing rules had come into effect in Santiago. As in Cerrillos, residents were worried that another large data center would threaten their access to water and provide little in return. Vicente Bardales, a member of the Quilicura Socio Environmental Resistance group, told Rest of World that the companies touch down in places like Quilicura only to “use water, [and] dry up the territory, but don’t give anything back to the population.” Similar scenarios to that of Chile played out across Latin America. The following year, the Uruguayan capital of Montevideo found itself under a state of emergency as the country was hit by its worst drought in 70 years. As residents demonstrated against the misuse of water by industry, they found out Google was planning a data center in Canelones, a little less than an hour north of Montevideo. “Only a tiny proportion of water in Uruguay is used for human consumption,” explained Daniel Pena, a researcher at University of the Republic. “The majority is used for big agro industries, such as soya, rice, and wood pulping. Now we have Google planning to use enormous quantities of water.” Pena had to go to court to get details of its planned water use from the Ministry of Environment, ultimately finding Google’s 72-acre site was planning to use 2 million gallons of water each day to power its cooling systems. Uruguayans had fought for fresh drinking water to be considered a basic right in the national constitution two decades earlier, and as companies like Google moved in, they feared its realization was slipping from their grasp. Uruguayans suspected Google had shifted its focus to their country because it had run into such opposition in Chile. Neighboring Brazil already has over 180 data centers, even as many regions regularly experience drought conditions. On the outskirts of Fortaleza, Brazil’s fourth-largest city, at least four major data centers were looking to set up shop to take advantage of a free-trade zone in the municipality of Caucaia. One in particular would be the size of 12 football fields, costing an estimated $10 billion, and though it will technically be owned by a local wind energy company, it is actually being developed to serve TikTok. This is in a region that experienced emergency drought conditions for 16 of the 21 years between 2003 and 2024. What Cerillos, Canelones, and Caucaia are experiencing is increasingly a global phenomenon. Around the world, data center companies claim they will be respectful and considerate neighbors, but too often they hide the extent of the damage that follows from the scale of their demands on local resources. This is particularly acute in parts of the Global South, but the effects are now also being felt across the United States and Europe. Adapted from “Hyperscale: The Ambition and Excess of Big Tech’s Data Empires” by Paris Marx published by Riverhead Books, an imprint of Penguin Random House © 2026 by Paris Marx, and reprinted with permission.
document fetched 2026-10-24 11:24 · canonical https://restofworld.org/2026/ai-datacenters-water-resistance-global