Global electricity generation required to supply data centres will grow from 460 TWh in 2024 to over 1 000 TWh in 2030 and 1,300 TWh in 2035, by rough IEA estimations.
A recent report by the International Energy Agency (IEA), revealed that global electricity demand from data centers is expected to more than double in the next five years, driven by the widespread surge in AI use. Thus, global electricity generation needed to power data centres is projected to rise from 460 TWh in 2024 to over 1,000 TWh by 2030, and to 1,300 TWh by 2035.
The estimations go in line with earlier forecast made by Barclays. This study estimated that AI growth could triple U.S. data-centre usage by the decade’s end, increasing energy consumption from 150-175 terawatt hours (TWh) in 2023 to 560 TWh, or 13% of current U.S. electricity demand in 2030.
Although the two numbers given significantly differ, they reflect a clear and predictable trend: people and businesses use AI more. That drives the growth of data center energy demand, leading to greater energy use overall. IEA estimates data centres’ electricity use has grown at a rate of 12% per year over the last five years.
The U.S., China, and Europe will expectably stay the top regions for data centre electricity use, though other areas are growing fast as well. Notably, Southeast Asia, boosted by hubs in Singapore and Malaysia, is set to more than double its demand by 2030. By the same year, the U.S. and China are projected to drive 80% of global growth. U.S. demand will rise by 240 TWh (130%), China by 175 TWh (170%), Europe by 45 TWh (70%), and Japan by 15 TWh (80%).
Comparing per-capita data centre electricity use shows significant regional differences. In 2024, Africa used less than 1 kWh per person, which might rise to just under 2 kWh by 2030. South Africa stands out, reaching over 25 kWh per person — 15 times the regional average. The U.S. leads globally, with 540 kWh per person in 2024, set to rise to over 1,200 kWh by 2030 — about 10% of a household’s annual electricity use, and ten times higher than the per-capita rate in any other region.
Despite the AI-driven surge in energy demand, data centre electricity use growth is still expected to account for less than 10% of global electricity demand growth between 2024 and 2030. In some forecast scenarios, this amount might be as low as 2%-3%.
As for the environmental impact, CO₂ emissions from data centre electricity are forecast to peak at 320 Mt in 2030, then decline slightly to 300 Mt by 2035. Despite growth, data centres will emit under 1% of total CO₂.
Over the next five years, renewables will supply nearly half of the additional electricity needed by data centres, followed by natural gas and coal. Nuclear will become more important later in the decade. Coal is expected to remain the top source at 30%, especially in China. Renewables (mainly wind, solar, and hydro) will provide 27% of the total energy supply to data centers, natural gas 26%, and nuclear 15%. These figures reflect the actual energy used based on local grids and onsite generation, not just contractual sources.
Renewables use for the purpose is growing at the fastest rates, with a 22% annual increase expected through 2030, covering nearly half of the new demand. Wind and solar lead this growth, often supported by tech company PPAs or direct investments. Still, natural gas and coal will meet over 40% of new demand by 2030, via existing and new plants. After 2030, small modular reactors (SMRs) will begin supplying low-emission nuclear power, backed by major tech firms. This, along with renewables, will lead to a drop in coal use by 2035.
However, these projections are true for the so-called Base Case – a plausible scenario based on conditions similar to the ones we have today. However, there’s significant uncertainty around current and future data centre electricity use. Because of this, scenario-based analysis helps explore different paths and timelines for energy planning in three more different modes. Other scenarios, Lift-Off, High Efficiency, and Headwinds, illustrate key uncertainties in tech efficiency, AI growth, and energy sector challenges.
Thus, in the Lift-Off Case, nearly 50% of the extra electricity for data centres between 2024 and 2030 will come from fossil fuels. Natural gas is expected to grow 1.5 times faster than in the Base Case, led by the U.S., while coal grows twice as fast, mainly due to China. By 2035, in the same alternative scenario, fossil fuels will provide about 35% of added data centre electricity globally.
In other scenarios, fossil fuels supply 35% in the High Efficiency Case, 15% in the Headwinds Case, and 28% in the Base Case. Across all cases, fossil fuels make up around 40% of total data centre electricity by 2035, remaining key options for handling demand surges.
