According to a report by Morgan Stanley, artificial intelligence (AI) data centers are set up to dramatically increase global water consumption over the next few years.
The report predicts that by 2028, these facilities will be able to cool the systems and use approximately 106.8 billion litres of water per year to generate electricity.
“AI Data Centers will drive annual water consumption for cooling and power generation to approximately 106.8 billion litres by 2028 (basic case) – an 11-fold increase from 2024 estimates.” Annie It is listed.
AI water footprint
The report said that water use in cooling data centres is widely recognized, but indirect water consumption from electricity generation remains less than it is.
The AI's full water footprint spans three “scopes.”
This refers to the direct use of water as data centers require a large amount of water to prevent the server from overheating.
- Scope 2: Power generation
This is the indirect use of water that is involved in generating the large amount of electricity needed to generate electricity in a data center. The report notes that this typically accounts for the largest share of AI water consumption.
- Scope 3: Semiconductor manufacturing
This includes water-intensive processes that produce specialized chips that power AI. Semiconductor facilities can consume as many as 5 million gallons of Ultra Pier water every day.
Key factors for estimation and its impact
The report estimates are based on several factors, including water intensity, consumption factors, cooling techniques, and assumptions regarding local energy mix.
It also presents a variety of scenarios that can be achieved through 2028, with annual water consumption potentially changing between 637 billion litres and 148.5 billion litres depending on the adoption of more efficient solutions.
While AI's overall global water consumption may seem modest, the report highlighted the highly localized impact of its effects. He noted that more than half of the hubs in major global data centres are located in areas already facing moderate to high risks of water scarcity, drought, flooding and water quality decline.
This trend is common in many secondary markets and new data center hubs in areas with high water stress or very high water stress.
“More than half of the world's top data center hubs are in regions that are already facing moderate basin physical risks (i.e. moderate levels of vulnerability) to threats from drought, flooding and water quality,” he said.
As AI demand continues to increase, so will the need for efficient management of water resources. This issue is evolving rapidly and remains a critical issue for industry growth and sustainability. Annie It has been reported.
