The escalating demand for data centres highlights their critical role in the digital economy, with projections indicating that electricity consumption for these facilities could reach approximately 415 terawatt-hours by 2024, comprising 1.5 percent of global electricity use. This marks a significant increase of 12 percent annually, and some estimates suggest that by 2025, consumption could rise to 448 terawatt-hours.
Professor Mark Gahegan from the University of Auckland emphasizes that data centres are essential for housing the necessary computing resources, which demand significant energy for both operation and cooling. The environmental impact of artificial intelligence (AI) extends beyond its direct use, stemming largely from the infrastructure that supports it, particularly these energy-intensive data centres.
While AI's operational energy demands are relatively low, the initial training of extensive models such as GPT-4 can consume between 52 million and 62 million kilowatt-hours of electricity. This situation underscores the significant energy requirements of the technology sector, especially as reliance on AI continues to grow across various fields, including medicine and climate science.