Jakarta, INTI - The rapid development of Artificial Intelligence (AI) is driving demand not only for electricity but also for significant amounts of water. Data centers that power AI models require substantial water resources for cooling their high-performance computing infrastructure.
According to the World Economic Forum (WEF), a 1-megawatt (MW) data center can consume up to 25.5 million liters of water annually for cooling alone. This is comparable to the daily water consumption of around 300,000 people.
Such levels of consumption can put additional pressure on water resources, particularly in regions already facing water scarcity.
Data centers rely on various cooling technologies, including chillers, cooling towers, and liquid cooling systems. Water is also consumed indirectly through electricity generation, particularly at power plants that use water-intensive processes such as steam generation.
The WEF estimates that a medium-sized 15 MW data center could consume an amount of water annually equivalent to the yearly needs of three average-sized hospitals or more than two 18-hole golf courses.
One factor contributing to high water consumption is the limited ability to reuse water within cooling systems. Some freshwater evaporates during the cooling process, while the remaining water may become wastewater containing dust, chemicals, and minerals.
As a result, wastewater cannot always be immediately reused because contaminants can reduce the efficiency of cooling systems.
AI Growth Adds to Water Pressure
The rapid expansion of AI is further increasing the challenge. AI models such as GPT-3 and GPT-4 require substantial computing power, which increases both energy consumption and the amount of cooling required to keep data center equipment operating at safe temperatures.
The WEF estimates that GPT-3 can require around 500 milliliters of water for every 10–50 responses. When multiplied across billions of AI interactions, the technology's overall water footprint can become significant.
The WEF also estimates that global AI demand by 2027 could result in annual water withdrawals of 1.1–1.7 trillion gallons, equivalent to approximately 4.2–6.6 billion cubic meters. This would be more than four to six times Denmark's total annual water withdrawals.
According to the WEF, the water consumption of data centers has received less attention than their carbon emissions. Improving circular water management, including optimizing water use and replacing freshwater sources with alternative supplies, therefore represents an opportunity to reduce the industry's environmental impact.
Circular Water Solutions Could Reduce Consumption
To mitigate the broader environmental and social impacts of data centers, operators can adopt innovative circular water management solutions.
These approaches involve treating, recycling, and reusing water to create more sustainable and efficient water management systems.
Technologies such as wastewater-recycling cooling systems and rainwater harvesting can significantly reduce reliance on freshwater. In some locations, these practices have the potential to reduce freshwater consumption by 50–70%.
Integrating circular water management principles could become increasingly important as data centers expand to support the next wave of technological development while operating in a world facing growing pressure on water resources.
Conclusion
The rapid growth of AI is creating a new challenge for data centers: rising water consumption. As demand for computing power increases, adopting water-efficient cooling systems, recycling, and circular water management will be increasingly important to reduce environmental pressure.
Read more: PLN Strengthens Power Infrastructure to Support Indonesia’s Growing Data Center Industry