Artificial Intelligence (AI) data centres are set to dramatically increase global water consumption in the coming years, according to a report by Morgan Stanley.

The report forecasts that by 2028, these facilities could use roughly 1,068 billion litres of water annually for cooling the systems and for electricity generation.

“AI data centers to drive annual water consumption for cooling and electricity generation to approximately 1,068 billion liters by 2028 (our base case) – an 11x increase from 2024 estimates,” the report quoted by ANI stated.

Water footprint of AI

The report noted that while water use for cooling data centres is widely acknowledged, its indirect water consumption through electricity generation remains underappreciated.

The full water footprint of AI extends across three “scopes”:

This refers to the direct use of water as data centres require large quantities of water to keep servers from overheating.

Scope 2: Electricity generation

This is an indirect use of water which in involved in generating massive amounts of electricity needed to power data centres. The report notes that this is an typically accounts for the largest share of AI’s water consumption.

Scope 3: Semiconductor manufacturing

This includes the water-intensive process of producing the specialised chips that power AI. Semiconductor facilities can consume as much as five million gallons of ultrapure water every day.

Key factors of the estimate and its impact

The report’s estimates are based on several factors, including assumptions around water intensity, consumption factors, cooling technologies, and regional energy mix.

It also presents a range of possible scenarios till 2028, with annual water consumption potentially varying between 637 billion litres and 1,485 billion, depending on the adoption of more efficient solutions.

While AI’s overall global water consumption may appear modest, the report highlighted that the impact is highly localised. It pointed out that over half of the leading global data centre hubs are situated in regions that already face moderate to high risks of water scarcity, drought, flooding, and declining water quality.

This trend is also prevalent in many secondary markets and emerging data centre hubs which are located in areas with high or very high water stress.

It stated “More than half of the world’s top data center hubs are in areas already facing medium basin physical risk (i.e., medium level vulnerability) to threats from drought, flooding, and declining water quality”.

As the demand for AI continues to grow, so will the need for efficient management of water resources. This issue is rapidly evolving and remains a critical challenge for the industry’s growth and sustainability, ANI reported.