AI’s growing thirst: Data centres may use 9.3 trillion litres of water by 2030 – Business News

The projected electricity demand alone is nearly triple the combined annual power consumption of Pakistan, Bangladesh and Nigeria, countries with a combined population exceeding 650 million people, and would account for almost 3% of global electricity use by the end of the decade.

The associated water footprint is equally staggering. The report estimates that AI-driven data centres will consume enough water to meet the basic annual domestic needs of 1.3 billion people in Sub-Saharan Africa, while the land footprint would be roughly twice the size of the Jakarta metropolitan region, home to more than 32 million people.

The findings come as governments and technology companies worldwide race to build AI infrastructure, with demand for computing power surging following the rapid adoption of generative AI applications.

The report notes that global data centres consumed an estimated 448 TWh of electricity in 2025. If considered a country, they would already rank as the world’s 11th-largest electricity consumer, behind France but ahead of Saudi Arabia. Within just five years, that figure is expected to more than double.

Everyday Prompts

One of the report’s most significant findings is that the environmental burden of AI is increasingly driven not by training large models but by their everyday use. Researchers estimate that 80-90% of AI energy consumption now comes from inference — the continuous operation of deployed models responding to user requests.

ChatGPT alone is estimated to process around 2.5 billion prompts every day, translating into roughly 383 GWh of electricity consumption annually for a single product. According to the report, offsetting the associated emissions would require 2.6 million tree seedlings grown for 10 years, enough trees to cover an area roughly equivalent to Manhattan.

The study also highlights the rapidly rising energy intensity of AI-generated content. A typical AI-generated image consumes around 1,450 times more energy than a basic text-classification task, while a short AI-generated video can consume as much electricity as 200,000 spam classifications.

“What surprised us most is how often the choices that look greenest from a carbon perspective end up worse for water or for land,” said Dr Miriam Aczel, lead author of the report.

Efficiency Paradox

The report warns that efficiency gains alone may not solve the problem. As AI systems become cheaper and more efficient, overall consumption rises even faster — a phenomenon known as the Jevons Paradox.

Beyond electricity and water, the report identifies a growing global imbalance in AI infrastructure. More than 90% of AI-specialised computing capacity is concentrated in the United States and China, while only 32 countries host AI-specialised data centres. More than 150 countries currently have little or no sovereign AI computing capacity.

“This report is not a case against artificial intelligence,” said Professor Kaveh Madani, Director of UNU-INWEH. “It is a call for using it responsibly and addressing its unintended impacts proactively.”

The report argues that unless governments integrate AI infrastructure into energy planning, water governance and land-use policies, the environmental footprint of the technology could soon rival that of some of the world’s largest industrial sectors, even as demand for AI services continues to surge globally.