Global trade war may produce headwinds for nascent AI sector, IEA says

By Forrest Crellin

PARIS (Reuters) - An escalating global tariff war could provide challenges for the emerging data centre sector and cause slower growth, Laura Cozzi, the International Energy Agency (IEA) Director of Technology told Reuters.

The U.S., China and the European Union together are set to account by 2030 for 80% of the forecast growth in data centre demand growth, which is expected to be dominated by Artificial Intelligence (AI) use, the IEA said in a report on Thursday.

The report's headwind scenario "encompasses many of the things we are seeing - slower economic growth, more tariffs in more countries, so indeed yes (the current tariff environment) is a scenario in which AI would see a slower growth than what we see in our base case," Cozzi said.

Global electricity consumption from data centres is expected to rise to around 945 terawatt hours (TWh) by 2030 in the IEA's base case scenario, but the "headwind scenario" would see that drop to 670 TWh, IEA data showed.

In the United States, data centres are expected to account for nearly half of electricity demand growth between now and 2030, and the country is expected to lead in data centre development globally, according to IEA data.

U.S. electricity utilities have been fielding massive requests for new capacity that would exceed their peak demand or existing generation capacity, raising concerns that tech companies are approaching multiple power utility providers, inflating the demand outlook.

The report aims to work with tech companies and industry to make sense of the real queue for data centres, which is ultimately going to be essential for AI to get the electricity it needs, Cozzi said.

Strain on grids could also lead to project delays, with about 20% of planned data centre projects at risk. Demand for transmission lines and critical grid and generation equipment are in high demand, reflecting this risk, the IEA report said.

Some 50% of data centres under development in the United States are in pre-existing large clusters, potentially raising risks of local bottlenecks, it said.

(Reporting by Forrest Crellin; Editing by Gareth Jones)