The use of Artificial Intelligence (AI) is increasing in the modern world and among a multitude of concerns raised, the environmental impacts of such technology have been emphasised the most. There has been a lot of public debate about the energy consumption of AI data centres, yet there has been almost no discussion about their water consumption – until now.
What are AI data centres?
AI data centres are the physical facilities that house the massive computing power and storage needed to train and run AI models, like the kind used in chatbots, language processing or predictive systems. These centres often operate 24/7 in vast warehouses, requiring significant energy to maintain ideal conditions for high-performance computing hardware.
Energy consumption at these centres is well documented. A study titled “Making AI Less ‘Thirsty’: Uncovering and Addressing the Secret Water Footprint of AI Models” estimates that AI data centres now consume around 2% of all global electricity.¹
But the water usage figures are just as eye-opening. In Ireland, the expansion of hyperscale data centre infrastructure has prompted concern over water use. A 2020 Eolas Magazine report found that centres in Greater Dublin were consuming between 0.5 and 5 million litres of water per day, with one Facebook site in Meath using 395 million litres in 2019 - the equivalent of a medium-sized Irish town.² This level of demand raises concerns, especially during dry summer periods, when water restrictions and hosepipe bans are already in place.
Why do data centres use so much water?
AI data centres consume water both indirectly and directly, with the most usage occurring during the “training” phase of AI development when the system is being fed data and taught how to respond.
Indirect water use refers to water consumed off-site in electricity generation (e.g. cooling towers at thermal power plants). While data centres don’t use this water directly, their energy demands can significantly amplify it.
Direct water use happens on-site, mostly for cooling. Servers generate enormous heat, and to keep systems stable, water is often run through heat exchange systems. Seasonal temperature shifts also demand increased cooling capacity, something especially concerning as climate change causes more frequent heatwaves, impacting water supply and distribution.
Alarmingly, the “Making AI Less Thirsty” study also noted that data centres in Asia may use up to three times more water than those in Europe or the U.S. This is particularly worrisome given projections that half of the world’s population will face severe water stress by 2030.
Two important caveats
While the data is striking, it’s important to note that water consumption figures for AI data centres remain largely underreported. This is due in part to a lack of research and, reportedly, a reluctance among tech companies to publicly disclose usage data. Additionally, the core study referenced above, though widely cited in the media, has not yet been peer reviewed as of this writing. This doesn’t invalidate its findings, but further independent investigation will be essential.
What can be done about it?
The authors of “Making AI Less Thirsty” point out that where and when AI is trained makes a significant difference in water usage. Therefore, the geographic location of data centres, and the timing of high-power computing operations, should be key planning factors.
In the meantime, innovative technologies can help reduce direct water consumption in cooling processes. For example:
Our SIRION™ Mega SF and TF models offer high high-flux, low-energy reverse osmosis (RO), removing up to 98% of dissolved organics from wastewater. This water can then be recycled into cooling towers, cutting fresh water demand and saving up to 50% in electricity compared to traditional systems.
Paired with Ionsoft™ systems, these RO units help reduce discharge volumes, enabling more circular and cost-effective water management within data centre operations.
And from a compliance perspective, Ireland is now taking action. As of February 2025, under EU Delegated Regulation 2024/1364, data centre operators in Ireland (with ≥500 kW IT capacity) must report key sustainability metrics, including water consumption, withdrawal and source data, to a European data centre database.³
This means that water-conscious operation is no longer just environmentally responsible; it's becoming a regulatory requirement. Our engineering teams can work with facilities to audit existing systems, identify inefficiencies and design sustainable upgrades that reduce both environmental impact and operational cost.
For more information about the solutions we can provide for data centre facilities, get in touch with one of our experts here.
1 Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models - https://arxiv.org/abs/2304.03271
2 Eolas Magazine: Data Centre Water Usage in Ireland - https://www.eolasmagazine.ie/data-centre-water-usage
3 Gov.ie: Data Centre Energy and Sustainability Performance Reporting Obligations - https://www.gov.ie/en/department-of-climate-energy-and-the-environment/publications/data-centre-energy-and-sustainability-performance-reporting-obligations
