AI data centers to double electricity use by 2030

The electricity consumption of AI-focused data centers is projected to increase from 485 terawatt-hours in 2025 to 950 terawatt-hours by 2030.

According to the International Energy Agency (IEA) report titled “Key Questions on Energy and Artificial Intelligence,” data centers are expected to account for approximately 3% of global electricity demand by 2030. The report estimates that electricity consumption by these centers will rise from 485 terawatt-hours in 2025 to 950 terawatt-hours by 2030, CE Report quotes Anadolu Agency.

With advances in artificial intelligence, simple text queries now consume relatively small amounts of electricity, while tasks such as video generation and advanced reasoning can require thousands of times more energy.

Although there are no comprehensive global statistics on the frequency and depth of AI use, major model providers report that active user numbers have tripled and revenues have increased fivefold over the past year, indicating rapid growth in demand.

The report notes that investments aimed at eliminating bottlenecks in energy equipment and chip production, combined with the rapid expansion of energy-intensive AI applications, could lead to even higher demand scenarios after 2030.

By 2027, a single server rack in an advanced data center could require as much instantaneous power as 65 households.

Natural gas emerges as a key solution for data centers

Unlike traditional data center operations, AI training and model usage create large and rapid fluctuations in power demand. For this reason, energy storage systems are considered critical to ensuring uninterrupted and reliable electricity supply.

In the United States, slow progress in grid connections is pushing data center developers toward on-site natural gas-based electricity generation projects. Satellite monitoring of these projects indicates that land preparation or construction has already begun in around one-fifth of them.

This trend suggests that on-site natural gas power generation is emerging as a major solution, although important questions remain regarding design, supply chains, regulation, and financing.

According to the IEA’s analysis, in order to reliably meet the critical and variable load demand of data centers through on-site natural gas generation, power infrastructure would need to be built 30% to 70% larger than the expected demand.

However, due to supply shortages in gas turbines, it remains unclear whether on-site generation can always provide a faster solution for data centers.

Despite the uncertainties, the report estimates that by 2030, on-site natural gas capacity supplying data centers could reach between 15 and 27 gigawatts globally, with most of it located in the United States.

The IEA also forecasts that around 20–25 gigawatts of battery storage capacity could be installed in data centers worldwide by 2030, potentially becoming a major asset for power grids if appropriate incentives are introduced.

The report additionally noted that, in recent months, a data center operator signed an agreement for what would become the largest battery project ever in terms of energy capacity — four times larger than the previous record — helping accelerate the commercialization of long-duration energy storage technologies.

Photo: Chat GPT