A recent Morgan Stanley report projects that the rise of AI infrastructure could drive global water use to 1,068 billion litres annually by 2028, an eleven-fold increase from today. The study analyses water consumption across three categories:

1. Scope 1: Direct water use for cooling servers.

2. Scope 2: Indirect water use from electricity powering data centres.

3. Scope 3: Water used in producing semiconductors and chips.

Annual water demand could range from 637 billion to 1,485 billion litres, depending on technological efficiency and regional energy choices. Alarmingly, over half of major data centre hubs are in regions already facing moderate to high water stress, which could worsen shortages in communities dependent on the same resources.

The surge in water demand is driven by high-performance AI models, which require more data centre capacity, power, and cooling. Without improvements in cooling efficiency, recycling, or renewable energy adoption, the environmental impact could be significant.

Experts suggest that strategic data centre placement, adoption of water-efficient technologies, and regulatory oversight will be essential to balance AI growth with water sustainability, highlighting the need for industry-wide environmental responsibility alongside technological innovation.