International Energy Agency (IEA), has published its first major study on the impact of AI gold rush on global energy consumption. Its findings paint a worrying and perhaps contradictory picture.
The energy consumption of data centres, which includes artificial intelligence applications, will double in the next five year to 3%. The report concluded that AI-specific energy consumption could be responsible for more than half of the growth.
Some of today’s data centres consume as much electricity per household as 100,000 households. According to the IEA, hyperscalers in the future may consume 20 times that amount. The IEA predicts that by 2030, data centres will run on 50% renewable power and the rest will be a mixture of coal, nuclear energy, and new natural-gas-fired plants. The IEA stated that the findings paint a bleak climate picture, but there is a silver lining. AI will consume more energy but its ability to unlock efficiencies in power systems and discover novel materials could provide a counterweight.
The of EU tech
The latest rumblings on the EU tech scene
A story from our wise old founder Boris and some questionable AI artwork. Every week, it’s in your inbox for free. Sign up today!
Fatih Birol is the executive director of the IEA. He said, “With the rise in AI, the energy industry is at the forefront one of the most significant technological revolutions of our times.” “AI is an incredible tool, but how we use it is up to our societies, governments and companies.”
AI is able to optimize power gridsand increase the energy output from solar and wind farms by improving weather forecastingand detect leaks within vital infrastructure. The technology can also be used to better plan transportation routes or design cities. AI has the potential to discover new green materials for tech like batteries.
But the IEA warned governments that the combined impact would be “marginal,” unless they created the necessary “enabling circumstances.”
“the net impact of AI will depend on the way AI applications are rolled-out, what incentives and businesses cases arise, as well as how regulatory frameworks react to the evolving AI environment,” the report stated.The AI energy debate is divided
Major questions remain about AI’s potential to reduce energy consumption. The negative climate impact of the technology is already in place.
According to the IEA, data centres will contribute 1,4% of global “combustion emission” by 2030. This is almost triple the current figure and almost as much as air travel. The IEA’s estimate is not accurate because it does not include the embodied emissions from building all these new data centres, and the materials used to construct them.
Alex de Vries, a researcher and founder of Digiconomist at VU Amsterdam, believes that the IEA underestimated the growth of AI’s energy use. “We’re talking about several percentages of our global consumption of electricity,” De Vries Said. This increase in data centre electricity consumption “could pose a serious threat to our ability to meet our climate goals,” said de Vries.Claude Turmes is the Luxembourg’s Energy Minister. He accused the IEA for presenting a too optimistic view, and failing to address the harsh realities that policymakers must hear.
Instead of making practical recommendations for governments on how to regulate AI and new mega-data centres and minimise their negative impact on the energy system the IEA [executive director] Fatih Birol is making a welcome present to the new Trump Administration and the tech companies that sponsored this new US Government,” he said toThe Guardian.
In addition to AI, there are other proven methods of reducing energy consumption in data centres. Immersion cooling is one of thesolutions thatstartups such as Asperitas from the Netherlands, Submer in Spain, and Iceotope from the UK are pursuing. DeepGreen, a UK-based venture, proposes repurposing data centre heat for other uses. All these strange and wonderful solutions need to be scaled up quickly if they want to make a dent on the data centres’ thirst of electricity. We also need to use computing power in a more efficient way.
The debate on sustainable AI will be continued at the TNW Conference,which takes place in Amsterdam on June 19-20. Tickets are on sale now. Use the code TNWXMEDIA2025 to get 30% off.
