Increasing demand for AI power poses global supply risks, leading to potential regulatory measures.

Tech companies’ increasing electricity consumption for AI training poses global power supply stability risks, says Hitachi Energy CEO Andreas Schierenbeck. AI data centres exhibit volatile power demands, spiking up to 10 times normal usage within seconds. Regulatory measures similar to other industries are needed to manage this.

The International Energy Agency projects data centre electricity usage to double to 945 TWh by 2030, exceeding countries like Japan’s current consumption. Ireland and the Netherlands have restricted new data centre developments due to electricity network concerns. A US DOE report forecasts data centre power demand doubling or tripling by 2028.

Lawrence Berkeley National Lab’s report shows data centre electricity usage jumped from 58TWh in 2014 to 176TWh in 2023, possibly reaching 325TWh to 580TWh by 2028. Analysts suggest AI power demands could stabilize grids if companies set limits and align training with renewable energy. Hitachi Energy faces global power transformer shortages, which may take up to three years to address.

Read more at Yahoo Finance: AI power demand poses global supply risks, says Hitachi Energy