AI revolution to drive increase in data center power demand

The growing adoption of AI technologies is expected to significantly impact global power consumption. According to Goldman Sachs Research, the power demand of data centres will rise by 160% by 2030, driven by the high energy requirements of AI processing. Currently, data centres consume 1-2% of the world’s power, a figure projected to increase to 3-4% by the end of the decade.

Data centers have maintained stable power usage due to efficiency gains, despite increasing workloads. However, since 2020, these gains have slowed, and power consumption has risen. For instance, a ChatGPT query uses nearly 10 times more electricity than a Google search, requiring 2.9 watt-hours compared to 0.3 watt-hours.

Goldman Sachs estimates AI will add 200 terawatt-hours per year to data centre power consumption from 2023 to 2030. By 2028, AI is expected to account for about 19% of data centre power demand. This increase in demand will have a significant carbon footprint, with data centre CO2 emissions potentially more than doubling by 2030, resulting in a “social cost” of $125-140 billion.

In the US, power demand has been flat for the past decade, but is set to surge by 2.4% between 2022 and 2030, with data centres contributing to 0.9 percentage points of this growth. Data centres will use 8% of US power by 2030, up from 3% in 2022. This will necessitate around $50 billion in new generation capacity investments and drive 3.3 billion cubic feet per day of additional natural gas demand.

Europe’s power demand, which has declined by 10% since 2008 due to economic crises and de-industrialization, is poised to grow by 40-50% from 2023 to 2033. This growth will be driven by data centre expansion and increased electrification. Europe’s ageing power grid will require €800 billion in transmission and distribution investments, plus €850 billion in renewable energy.