A recent opinion piece by Alex de Vries-Gao, a PhD researcher at the VU Amsterdam Institute for Environmental Studies, explores how artificial intelligence could dramatically reshape global energy use. His analysis, published in the journal Joule, estimates current and projected electricity consumption linked to AI activity in data centers.
Drawing from a straightforward study, de Vries-Gao calculated how much energy AI providers might be using to process user queries. His work follows new findings from the International Energy Agency, which reported that data centers accounted for up to 1.5% of global electricity consumption in 2024, a figure that continues to grow. While AI is a major factor, data centers also support services like cloud storage and bitcoin mining.
Large AI models like ChatGPT require significant computing resources, and some tech companies have even started producing their own energy to support operations. However, de Vries-Gao points out a recent decline in transparency from AI firms about their energy usage, prompting him to conduct independent estimates.
His findings suggest that AI workloads could consume around 82 terawatt-hours of electricity this year is comparable to the total annual power use of a nation like Switzerland.
Furthermore, he projected what might happen if AI demand continues to grow at its current rate. If it doubles before year-end, AI-driven applications could represent nearly 50% of the energy used by data centers globally.
Beyond sheer energy demand, de Vries-Gao emphasises the environmental stakes. As many AI companies still rely on grid electricity, which is often generated using fossil fuels, the resulting greenhouse gas emissions could accelerate climate change and push energy prices even higher.
