The European Central Bank published Economic Bulletin research examining the surge in artificial intelligence (AI) use and the resulting increase in data centre electricity demand, assessing potential knock-on effects for energy and commodity markets. The analysis finds that, even under extreme assumptions, AI-related demand is likely to add only limited upward pressure to natural gas and critical mineral prices, while posing more material risks of price pressures in national electricity markets where supply and interconnection constraints bind. AI-related energy consumption in data centres is estimated at around 20 terawatt-hours (TWh), but the International Energy Agency projects total data centre electricity use could be 80% higher in 2026 than in 2022, with AI-driven data centres contributing an additional 90 TWh (around 20 percentage points of the growth), equivalent to roughly 4% of the EU’s current electricity consumption. Under a scenario where incremental demand is met entirely by natural gas generation, gas prices are estimated to rise by around 9% in Asia and Europe and by 7% in the United States, with AI-driven data centres accounting for about 2 percentage points of those increases; under a renewables-only scenario, demand for critical minerals such as lithium and nickel increases but is not expected to materially affect prices given market size and transportability. The paper highlights that electricity markets are more fragmented than gas and minerals markets, so countries with high data centre concentration and limited interconnection capacity, such as Ireland, could face sharper local challenges depending on market structure and regulatory requirements for data centres to support power supply.
European Central Bank 2025-03-20
European Central Bank research estimates AI-driven data centres could add 90 TWh of electricity demand by 2026 with limited gas and critical mineral price effects
The European Central Bank's Economic Bulletin research examines AI's impact on data centre electricity demand and energy markets. It finds limited pressure on natural gas and mineral prices but highlights potential electricity price pressures in countries with supply and interconnection constraints. AI-driven data centres are expected to significantly increase electricity consumption, posing challenges in areas with high data centre concentration and limited interconnection capacity.