Data Center Energy Consumption Surges Amid AI Boom

Global electricity demand from data centers is expected to more than double in the next five years, largely driven by the widespread uptake of AI. This is according to a report by the International Energy Agency (IEA), released on Thursday.

By the end of the decade, around half of the global energy demand for data centers is expected to come from renewables, including solar PV, wind and hydro. Combined, renewables are the fastest-growing source of electricity for data centers. The next most important energy supply in 2030 will be natural gas and coal, while the role of nuclear power is also expected to become more prominent.

As the following chart shows, a range of energy sources will be needed to meet global data centers’ growing electricity needs. The composition of electricity sources used to power data centers depends on the availability and cost-competitiveness of different sources in respective markets. In China, for example, coal represents a higher share of electricity generation than in the U.S., which has a higher share of natural gas.

In both countries, nuclear power is set to increase significantly, as well as solar and wind. In China, solar PV electricity generation is expected to surge from 7 TWh in 2025 to 83 TWh in 2035, not far behind the U.S.’ 97 TWh in 2035. In terms of wind, the U.S. will scale up from 28 TWh to 83 TWh over the ten years, nearly double that of China’s 44 TWh in 2035. Meanwhile, the U.S. will phase down its coal power generation with data centers expected to see a decline from 28 TWh to 7 TWh.

The report states that the demand for electricity for data centers will not be felt equally. For example, in the U.S. it will account for almost half of the growth in electricity demand, while it will account for more than half in Japan and one-fifth of the demand in Malaysia.

This relationship between energy and AI has raised questions over the potential environmental impacts of the technology. According to the report, data centers around the world currently account for around 180 Mt of indirect CO2 emissions from the consumption of electricity, which equates to roughly 0.5 percent of all combustion emissions released. IEA experts explain that while on the one hand, the technology requires electricity to be generated and therefore more emissions created in order to run, these could be somewhat offset by AI being used to make systems more efficient and to accelerate innovation in energy technologies such as batteries and solar PV.

Speaking on the topic EA Executive Director Fatih Birol said: “With the rise of AI, the energy sector is at the forefront of one of the most important technological revolutions of our time. AI is a tool, potentially an incredibly powerful one, but it is up to us – our societies, governments and companies – how we use it."

According to the IEA, this data is based on the fuel mix of the electricity physically consumed by data centres (considering both onsite generation and electricity received through the grid, taking into account the fuel mix of the local electricity systems they are located in), rather than the contractual mix of different data center operators.