Image: IEA
The new IEA energy and AI observatory is proposed to monitor and analyse the energy consumption and linkages between AI and energy.
Underlying its formation is the growing energy consumption of data centres and AI data centres in particular and the limited data on this consumption.
It follows the release in April by the IEA of its first comprehensive report on the impact of AI in the energy sector, which found that electricity demand from AI-optimised data centres could more than quadruple by 2030.
However, at the same time, AI is already being deployed in the energy industry, unlocking opportunities to cut costs, enhance competitiveness and reduce emissions.
Have you read?
Access to power biggest challenge for Europe’s data centres
New project spotlight: GRAVITEQA and EnergyGuard look to next gen AI energy solutions
Fatih Birol, IEA Executive Director, commented that AI is quickly emerging as one of the most important technologies of our time.
“Building on our recent major report on this subject, this new energy and AI observatory underscores our commitment to supporting decision makers around the world as they plan for the future. Reliable data and analysis are the cornerstone of navigating this fast-moving space.”
The energy and AI observatory, which is publicly accessible, includes comprehensive datasets with accompanying interactive tools to explore data centre electricity consumption and digital infrastructure by region and the aim is for these to be updated regularly.
It also features 20 case studies to show how AI is being deployed in a wide range of applications across the energy sector, following a public call for submissions that showcase current best practices.
The report estimated current data centre energy consumption at around 1.5% of the world’s electricity consumption and that it is set to more than double by 2030.
While this is still small at the global level, individual countries face challenges as data centres tend to cluster, as is apparent in the AI observatory mapping.
AI data centres in particular are also a concern due to their scale necessary to train and run AI models, which has resulted in the emergence of gigawatt-scale clusters in numerous regions across the US and for example in the Frankfurt, London, Amsterdam, Paris, and Dublin hubs in Europe.
Other useful data in the mapping is that the data centres can be seen alongside the associated power and digital infrastructure.
