Global data centres are projected to consume almost three times the annual electricity used by Nigeria, Pakistan and Bangladesh combined by 2030, according to a new report released by the United Nations on Wednesday.

The report, published by the UN University Institute for Water, Environment and Health, attributed the anticipated increase in electricity demand largely to the rapid growth in everyday artificial intelligence usage rather than the energy consumed during the training of AI models.

According to the report, electricity consumption by data centres is expected to rise from 448 terawatt-hours in 2025 to 945 TWh by 2030. Artificial intelligence is projected to account for 40 per cent of that total by the end of the decade, compared to about 20 per cent currently.

Pakistan, Bangladesh and Nigeria together have a combined population exceeding 650 million people, underscoring the scale of the projected energy demand from global data centre operations.

The report also forecasts a significant increase in water consumption by data centres. Water usage is expected to more than double from 4.5 trillion litres to 9.3 trillion litres by 2030. The projected volume is enough to meet the basic water requirements of approximately 1.3 billion people across Sub-Saharan Africa.

Carbon emissions associated with data centre operations are expected to increase from 189 million tonnes to 399 million tonnes during the same period. In addition, the land area occupied by data centres worldwide is projected to grow from 6,900 square kilometres to more than 14,500 square kilometres.

The report identified inference, the continuous operation of deployed AI systems to answer users’ daily queries, as the biggest contributor to AI-related energy consumption.

According to the findings, inference currently accounts for between 80 and 90 per cent of total AI energy use, making it far more energy-intensive overall than model training because of the vast number of user interactions taking place every day.

The report estimated that ChatGPT alone processes around 2.5 billion prompts daily and consumes approximately 383 gigawatt-hours of electricity each year.

It added that offsetting the emissions generated from ChatGPT’s annual electricity consumption would require the cultivation of 2.6 million tree seedlings over a ten-year period.

AI image generation was also highlighted as particularly energy intensive, requiring roughly 1,450 times more energy than basic text-classification tasks.

The report further noted that a single short AI-generated video consumes as much electricity as processing 200,000 spam classifications.

A more complex AI-generated video carries a water footprint of approximately 4.1 litres, an amount equivalent to nearly two days’ drinking water supply for one person.

The report warned that if current growth trends persist, electricity consumption by global data centres could nearly double to 945 terawatt-hours by 2030.

“At that level, data centers would consume almost three times as much electricity annually as Pakistan, Bangladesh, and Nigeria combined, countries that together are home to more than 650 million people”.

The report highlighted examples where the expansion of data centres has already strained local resources.

In Ireland, data centres accounted for 21 per cent of total metered electricity consumption in 2023, exceeding the combined electricity use of all urban households in the country.

The growing pressure on Ireland’s electricity infrastructure prompted the national grid operator to suspend approvals for new data centres around Dublin until 2028.

The report also revealed a significant imbalance in the global distribution of AI infrastructure.

Only 32 countries currently host AI-specialised data centres, while more than 90 per cent of the world’s AI computing capacity is concentrated in just two countries, the United States and China.

More than 150 countries currently have little or no access to sovereign AI computing infrastructure, highlighting a widening digital divide in the global AI ecosystem.

Another concern raised in the report is the increasing volume of electronic waste expected from AI infrastructure.

According to the findings, AI infrastructure could generate as much as 2.5 million tonnes of electronic waste annually by 2030.

The report said this volume is equivalent to discarding nearly 250 Eiffel Towers every year, with much of the waste processed in low-income countries that often have limited environmental protection measures.

It also noted that the critical minerals needed for AI hardware production are concentrated in regions with weaker environmental oversight, particularly in parts of the Global South, including several African countries.

Despite these concerns, Nigeria is actively positioning itself to benefit from growing global investment in data centre infrastructure.

The country’s strategy is being driven by the National Cloud Policy 2025 and strengthened by recent investments such as the commissioning of Digital infrastructure provider, Kasi Cloud’s LOS1 facility in Lagos in May 2026.

The report noted that these developments place Nigeria in a position to capture some of the economic opportunities emerging from the rapid expansion of AI infrastructure globally.

Kasi Cloud recently commissioned the first phase of its planned 100-megawatt data centre project.

The facility is expected to reduce Nigeria’s reliance on foreign cloud service providers and help retain an estimated $850 million currently spent each year on offshore cloud infrastructure.