AI data centers are less thirsty now, tech giants say
Tech companies are addressing concerns about the high water usage of AI data centers. These facilities, which house servers critical to internet and AI operations, require significant amounts of water for cooling. However, advancements like Nvidia's DSX system aim to reduce water consumption by using closed-loop cooling systems where liquid circulates directly through servers. This approach minimizes water use while maintaining efficient cooling, although there remains a balance between water and energy consumption. Companies such as Microsoft, Amazon Web Services, and Meta have reported improvements in water use efficiency despite increased overall water usage. The lack of standardized reporting on ESG efforts complicates assessments of progress across the industry.
AI data centers are less thirsty now, tech giants say NEW YORK Tech giants are claiming that their latest advancements in AI data center design have significantly reduced water consumption, addressing growing public concerns over the sector’s environmental impact. Companies such as Meta, Amazon Web Services (AWS), and Microsoft have adopted closed-loop cooling systems, which they assert require no net water loss. Meanwhile, NVIDIA has introduced a new system called DSX, allowing it to nearly eliminate water usage at certain facilities by employing a closed-loop approach where liquid circulates directly through servers, maintaining optimal temperatures without relying heavily on external water sources. According to Rystad Energy, global data centers consumed 222 billion liters of water in 2025 for cooling purposes. Without adaptive measures, this figure could nearly triple by 2030, reaching 644 billion liters. However, steps to curtail this growth could maintain usage below 400 billion liters. NVIDIA’s DSX system operates by allowing liquid to enter servers at a higher temperature, around 45 degrees Celsius, compared to the typical 32-degree Celsius used in many existing closed-loop systems. This reduces the need for continuous air cooling, thereby lowering both water and energy demands. Andy Masley, an independent researcher specializing in AI and data centers, notes that there is a direct trade-off between water and energy use for cooling. Reducing water typically increases energy consumption because the liquid within sealed pipes still requires cooling, often via air circulation, which relies on electricity. At NVIDIA, simple fan-based cooling is frequently sufficient, though mixed methods may be necessary in extreme conditions. Josh Parker, NVIDIA’s head of sustainability, emphasized that these strategies help mitigate the broader environmental impact of data centers. While NVIDIA claims near-complete elimination of water use in some cases, other companies remain cautious. Microsoft, AWS, and Meta have confirmed they also utilize closed-loop systems, but they have not disclosed specific figures on water savings. Their sustainability reports indicate that despite increased overall water use between 2022 and 2025, efficiency has improved, by 25% at Microsoft and 37% at AWS. However, there is no uniformity in how companies report their environmental, social, and governance (ESG) efforts. Elon Musk’s SpaceX, which recently acquired xAI, has yet to publish an ESG report, earning it the lowest rating from MSCI in June. Shaolei Ren, an engineering professor at the University of California, Riverside, explained that companies have less financial incentive to reduce water use since it is generally cheaper than electricity. Public relations considerations, however, play a larger role given the rising backlash against data centers in the U.S. Upgrading older facilities to newer technologies is costly, but older data centers are inherently smaller and less powerful, requiring less cooling. Minh K. Le, leading data center and hydrogen research at Rystad, noted that this factor may lessen the urgency for upgrades. Beyond direct water use, the environmental impact extends to energy generation. Many data centers rely on fossil fuels, contributing to greenhouse gas emissions. A 2024 study revealed that over half of U.S. data centers are powered by fossil fuels, emitting more than 105 million tons of CO2 annually. Efforts to offset these emissions would require over 31,000 wind turbines operating continuously. Additionally, the expansion of data centers has delayed the retirement of coal plants and spurred the construction of new gas-fired facilities, further complicating climate goals.
Go to the primary sources (2)
The official sources this coverage is built on. Read them directly to bypass framing.
Data centers, which support AI models, cloud storage, streaming, and social media, have a significant environmental impact due to their high energy consumption. They require vast amounts of electricity, water, and land, and if they were a country, they would rank as the 11th largest electricity consumer globally. In cities like Frankfurt and Dublin, data centers account for a large share of local electricity usage, and global demand is expected to double by 2030. Much of this energy comes from fossil fuels, contributing to increased carbon emissions and delaying the phase-out of coal plants. Some data centers are transitioning to renewable energy to reduce their environmental footprint.
Bias read (Center): The article presents factual information about the environmental impact of data centers without overtly favoring one perspective over another. It cites multiple research institutions and reports, providing balanced context on both the challenges posed by data center energy consumption and potential,
Why factuality (90): The article accurately references the IEA's projection of data center electricity consumption doubling by 2030 to 945 TWh, matching the primary source. It also correctly notes the rapid growth compared to other sectors, though it doesn't mention the specific breakdown of accelerated vs. conventional
Why objectivity (85): The article maintains a generally neutral tone, discussing the environmental impact of data centers without overtly favoring any perspective. It acknowledges the scale of energy use and potential environmental impacts without taking a clear stance.
Tech companies are addressing concerns about the high water usage of AI data centers. These facilities, which house servers critical to internet and AI operations, require significant amounts of water for cooling. However, advancements like Nvidia's DSX system aim to reduce water consumption by using closed-loop cooling systems where liquid circulates directly through servers. This approach minimizes water use while maintaining efficient cooling, although there remains a balance between water and energy consumption. Companies such as Microsoft, Amazon Web Services, and Meta have reported improvements in water use efficiency despite increased overall water usage. The lack of standardized reporting on ESG efforts complicates assessments of progress across the industry.
Bias read (Center): The article discusses technological advancements in reducing water usage in data centers without taking a stance on political issues. It presents information objectively, focusing on technical solutions and industry practices rather than political debates or ideological positions.
Why factuality (75): The article mentions data centers' water usage and Nvidia's closed-loop cooling system, which is not detailed in the primary source document. However, it does reference the IEA's projection of doubling electricity consumption by 2030, aligning with the primary source. Some details are extrapolated b
Why objectivity (70): The article presents the issue of data centers using water and power in a somewhat alarmist tone, emphasizing 'public anger' and 'mounting pressure.' This framing may introduce bias rather than presenting a neutral analysis of the situation.
How each side covered it
The same event, grouped by the political lean of the outlets covering it.
progressive
center
conservative
★
How each side covered it
Support independent, bias-aware news and unlock the social pulse, community voting, and every other Supporter feature.