AI development to boost energy demand at data centres

The technology will comprise 85% of the energy demand growth until 2026.

The rise of artificial intelligence (AI) will drive energy demand as data centres that provide computation and storage capabilities needed to develop such technology with require the consumption of more power.

In a report, Schroders said AI models in the training phase, where they learn to make predictions and data-based decisions, already use six times more energy than non-AI computational uses.

During the inference phase, where AI models are trained to draw conclusions from data and queries, they need two to three times more energy than traditional workloads.

Citing data from research firm SemiAnalysis, Schrodes noted that critical IT power capacity at data centres globally will increase to 96,000 megawatts (MW) by 2026 from 49,000 MW in 2023.

This represents a 25% compound annual growth rate (CAGR) over the next three years, with AI workloads comprising 85% of the increase.

The growth rate exceeds the projected 13% rise seen from 2014 to 2023.