Interactive IEA tracker shows where AI is guzzling the most energy
(2025/06/19)
- Reference: 1750345449
- News link: https://www.theregister.co.uk/2025/06/19/iea_observatory_ai/
- Source link:
The International Energy Agency (IEA) has unveiled an online platform to closely monitor and analyze the impact of AI across the energy sector worldwide.
The [1]Energy and AI Observatory sports interactive tools to explore datacenter electricity consumption and the scale of digital infrastructure by region, and provides case studies to illustrate how AI is being deployed across the energy sector itself, as well as its effects.
It follows the publication earlier this year of a [2]report from the IEA estimating that energy consumed by bit barns worldwide is set to more than double by 2030, surpassing that currently used by all of Japan. AI is the driver of most of this; the energy required by "AI-optimized" datacenters is projected to outpace the rest and more than quadruple by the end of the decade.
[3]
However, the same report also claimed that AI is already being deployed in the energy industry to optimize systems, minimize greenhouse gas emissions, and cut costs.
[4]
[5]
Divided into two main sections, the Observatory features an interactive map showing the locations of datacenter hubs and their capacity – both operating and planned – alongside major telecoms and high voltage distribution cable networks.
It notes that gigawatt-scale clusters have emerged in specific regions of North America, Europe, and Asia-Pacific, thanks to large-scale investment in the infrastructure that is used used in the development and training of AI models.
[6]
IEA Observatory – click to enlarge
Other charts show the total global installed bit barn capacity, illustrating how it has accelerated in recent years (doubling since 2020), the installed capacity by region (spoiler: the US is by far the largest), and datacenter capacity as a share of regional energy demand.
Meanwhile, the "AI for energy" section details use cases such as energy price forecasting using AI tools; AI-driven models for electricity access forecasting; and building-level electricity planning using AI models.
[7]Schneider Electric says US grid will be less stable by 2030 as datacenter demand rises
[8]AWS says Britain needs more nuclear power to feed AI datacenter surge
[9]Microsoft burnishes green cred by paying Swedes to burn biomass and bury CO 2
[10]Global datacenter electricity use to double by 2030, say policy wonks. Yup, it's AI
According to the IEA, the Observatory was first announced by executive director Fatih Birol at the [11]AI Action Summit held in Paris in February, and aims to inform national and international policy on energy and AI topics, including ongoing dialog between government and industry.
"The IEA is at the forefront of efforts to understand and manage the significant links between energy and artificial intelligence, which is quickly emerging as one of the most important technologies of our time," said Birol.
[12]
He added that the Observatory would support decision-makers around the world in planning for the future, stressing the importance of reliable data and analysis in such a fast-moving sector.
An [13]earlier IEA report also warned that the world is going to need a lot more electricity generation capacity within the next three years to keep up with an unprecedented spike in demand, though it isn't just AI and datacenters to blame. Increasing use of air conditioning, heat pumps, the electrification of industrial processes and electric vehicles are all taking their toll as well. ®
Get our [14]Tech Resources
[1] https://www.iea.org/data-and-statistics/data-tools/energy-and-ai-observatory
[2] https://www.theregister.com/2025/04/12/ai_double_datacenter_energy/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://regmedia.co.uk/2025/06/18/iea-observatory.jpg
[7] https://www.theregister.com/2025/06/03/schneider_electric_says_us_grid/
[8] https://www.theregister.com/2025/05/16/amazon_nuclear_power_britain/
[9] https://www.theregister.com/2025/05/06/microsoft_extends_carbon_removal_agreement/
[10] https://www.theregister.com/2025/04/12/ai_double_datacenter_energy/
[11] https://www.elysee.fr/en/sommet-pour-l-action-sur-l-ia
[12] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[13] https://www.theregister.com/2025/02/14/iea_global_electricity_demand/
[14] https://whitepapers.theregister.com/
The [1]Energy and AI Observatory sports interactive tools to explore datacenter electricity consumption and the scale of digital infrastructure by region, and provides case studies to illustrate how AI is being deployed across the energy sector itself, as well as its effects.
It follows the publication earlier this year of a [2]report from the IEA estimating that energy consumed by bit barns worldwide is set to more than double by 2030, surpassing that currently used by all of Japan. AI is the driver of most of this; the energy required by "AI-optimized" datacenters is projected to outpace the rest and more than quadruple by the end of the decade.
[3]
However, the same report also claimed that AI is already being deployed in the energy industry to optimize systems, minimize greenhouse gas emissions, and cut costs.
[4]
[5]
Divided into two main sections, the Observatory features an interactive map showing the locations of datacenter hubs and their capacity – both operating and planned – alongside major telecoms and high voltage distribution cable networks.
It notes that gigawatt-scale clusters have emerged in specific regions of North America, Europe, and Asia-Pacific, thanks to large-scale investment in the infrastructure that is used used in the development and training of AI models.
[6]
IEA Observatory – click to enlarge
Other charts show the total global installed bit barn capacity, illustrating how it has accelerated in recent years (doubling since 2020), the installed capacity by region (spoiler: the US is by far the largest), and datacenter capacity as a share of regional energy demand.
Meanwhile, the "AI for energy" section details use cases such as energy price forecasting using AI tools; AI-driven models for electricity access forecasting; and building-level electricity planning using AI models.
[7]Schneider Electric says US grid will be less stable by 2030 as datacenter demand rises
[8]AWS says Britain needs more nuclear power to feed AI datacenter surge
[9]Microsoft burnishes green cred by paying Swedes to burn biomass and bury CO 2
[10]Global datacenter electricity use to double by 2030, say policy wonks. Yup, it's AI
According to the IEA, the Observatory was first announced by executive director Fatih Birol at the [11]AI Action Summit held in Paris in February, and aims to inform national and international policy on energy and AI topics, including ongoing dialog between government and industry.
"The IEA is at the forefront of efforts to understand and manage the significant links between energy and artificial intelligence, which is quickly emerging as one of the most important technologies of our time," said Birol.
[12]
He added that the Observatory would support decision-makers around the world in planning for the future, stressing the importance of reliable data and analysis in such a fast-moving sector.
An [13]earlier IEA report also warned that the world is going to need a lot more electricity generation capacity within the next three years to keep up with an unprecedented spike in demand, though it isn't just AI and datacenters to blame. Increasing use of air conditioning, heat pumps, the electrification of industrial processes and electric vehicles are all taking their toll as well. ®
Get our [14]Tech Resources
[1] https://www.iea.org/data-and-statistics/data-tools/energy-and-ai-observatory
[2] https://www.theregister.com/2025/04/12/ai_double_datacenter_energy/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://regmedia.co.uk/2025/06/18/iea-observatory.jpg
[7] https://www.theregister.com/2025/06/03/schneider_electric_says_us_grid/
[8] https://www.theregister.com/2025/05/16/amazon_nuclear_power_britain/
[9] https://www.theregister.com/2025/05/06/microsoft_extends_carbon_removal_agreement/
[10] https://www.theregister.com/2025/04/12/ai_double_datacenter_energy/
[11] https://www.elysee.fr/en/sommet-pour-l-action-sur-l-ia
[12] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aFQ0E3Q0AUJ0cQtTbLfOCAAAAMI&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[13] https://www.theregister.com/2025/02/14/iea_global_electricity_demand/
[14] https://whitepapers.theregister.com/
ChrisElvidge
In addition, anyone building a new datacentre should have to provide their own power and water supplies, not rely on "government" to supply same, paid for by taxes they don't pay.
Water too
retiredFool
Extra water usage by these things is also an issue in many areas. The SW of the US is already seriously over subscribed on water usage. TX just passed a state bill to spend 1B annually for the next 20 years to fund more water supply. Most feel it is likely a "drop in the bucket" compared to funding that will be needed. The NM site they show also would have water constraints as I imagine the one in Saudi Arabia. And when push comes to shove, will the billionaires have sufficient control to divert water/juice to data centers and let the poorer people just die from heat/lack of water? Not to mention that thing called food needs water to grow too. So water for agriculture or water for data center?
Conspicuously missing from this article about AI energy consumption is climate change. In my opinion, every conversation about AI should include the fact that it is accelerating and worsening anthropogenic climate change by hugely increasing energy consumption.