While artificial intelligence investments on a global scale have reached hundreds of billions of dollars in the last year, discussions continue over the consumption of water resources and carbon emissions. Latest research reveals that the total water footprint of artificial intelligence systems can be between 312.5 billion liters and 764.6 billion liters annually. This amount is on the same order as the global annual bottled water consumption of approximately 350 billion liters. In the analysis, it is emphasized that the annual water footprint of 312.5 billion liters corresponds to an average of 856 million liters per day, while the upper limit of 764.6 billion liters means approximately 2.1 billion liters per day.
WATER CONSUMPTION OF TECHNOLOGY GIANTS
According to International Energy Agency (IEA) data, the total water use of global data centers is estimated to be approximately 600 billion liters. It is predicted that this amount may reach 1 trillion 200 billion liters by 2030 with the introduction of new investments focused on artificial intelligence. Experts emphasize that a data center with an average capacity of 100 megawatts consumes 2 million liters of water per day, and this amount corresponds to the daily water use of approximately 6,500 households. The water consumption charts of global technology giants are also quite striking! Google consumes approximately 22.7 billion liters of water annually, Microsoft approximately 13 billion liters, Apple approximately 6 billion liters, and Meta approximately 3.9 billion liters.
EMISSIONS ARE INCREASING
Water consumption in artificial intelligence systems is not only due to the direct need for cooling. As the energy consumption of data centers increases, the amount of water used in electricity generation also increases. The cooling systems of thermal power plants, hydroelectric production processes and some renewable energy technologies also require water. In the evaluation published by Vrije Universiteit Amsterdam, it is stated that indirect water use is often missing from companies’ official reports and that the real impact can be up to three to four times the declared levels. Research also shows that the carbon footprint of artificial intelligence systems is between 32.6 million tons and 79.7 million tons of carbon dioxide, and it is noted that this amount is equivalent to the annual emissions of a small European country.
Türkiye IS INDIRECTLY AFFECTED
Türkiye is among the countries indirectly affected by the current situation. The total annual water consumption across the country is calculated as approximately 63 billion cubic meters. This 73 percent of consumption is used in agricultural irrigation, 16 percent is used as drinking and potable water, and 11 percent is used in industry. Although publicly available water consumption data for data centers in Türkiye is limited, it is emphasized that the water consumption of operating data centers is quite limited compared to the consumption of global technology giants. According to experts, the real risk is related to Türkiye being a country experiencing water stress.
WATER STRATEGIES SHOULD BE DEVELOPED
Water Policy Expert Dursun Yıldız, while mentioning that Türkiye is putting new investments on the agenda in line with its goal of becoming a regional data center base, said, “While Turkey is putting new investments on the agenda in line with its goal of becoming a regional data center base, the issue of water efficiency also stands out among the planning topics. By 2027, the annual water consumption of artificial intelligence systems is expected to be between 4.2 and 6.6 billion cubic meters. This amount indicates a size equivalent to approximately 7 to 10 percent of Türkiye’s current annual total water consumption. “Therefore, it is a critical necessity for Türkiye to develop a framework that considers the sustainability of water resources while encouraging artificial intelligence investments, in terms of water planning as well as energy.”
(email protected)
