Relationship between artificial intelligence and energy according to the IEA
The relationship between artificial intelligence and energy is one of today’s most controversial issues, especially when considered in conjunction with innovation, sustainability and security. Although neural network based technology has grown to the point where it has already transformed the global economy completely, this development:
- has, on one hand, made the energy system efficient, resilient and sustainable;
- but on the other hand is leading to enormous electricity costs.
According to the International Energy Agency (IEA), as AI cannot exist without energy (electricity in particular), it will be difficult to manage the energy of the future efficiently without the support of AI. This two-way connection is therefore becoming a strategic issue for governments, companies and citizens.
Why artificial intelligence is consuming more and more energy
To better unerstand the situation described above, it is important to start by looking at the growth in demand for electricity. Modern AI systems (especially those using largescale models and generative artificial intelligence), require huge calculation capacity. Their training and use take place inside highly specialised data centres which operate 24 hours a day.
The IEA highlights how data centres are now among the main drivers of growth in electricity demand in several regions of the world. In some countries, they account for a significant proportion of the total energy consumption, with a direct impact on networks, energy planning and costs.
The most advanced neural network based data centres can consume quantities of energy comparable to amounts used by an average sized city. Furthermore, the geographic concentration of data centres greatly amplifies this problem, putting great pressure on local networks.
According to the IEA’s analysis, rapidly increasing investment in digital infrastructures requires more integrated energy planning. It is no longer sufficient to produce more electricity; flexible, resilient, digitalised networks are required. In that sense, the relationship between artificial intelligence and energy is also a question of infrastructure.
Key data in the IEA report
A look at some of the quantitive data provided directly by the International Energy Agency at global level reveals how data centres accounted for around 1.5% of global electricity consumption in 2024 and also how artificial intelligence represents an increasing proportion of this energy load.
A single large size AI based data centre can have an electrical capacity of over 100 megawatts (equivalent to the annual consumption of around 100,000 inhabitants), while the largest structures currently being built can reach levels comparable to the consumption of millions of families.
In some areas the impact is already very clear:
- in Ireland data centres account for around 20% of national electricity consumption;
- while in the State of Virginia that proportion is around 25%.
At the same time, the calculation capacity necessary to train the most advanced AI models has already increased by around 350,000 volts since 2014, highlighting how the relationship beween artificial intelligence and energy is now firmly established and destined to have a profound effect on electrical systems.
Energy efficiency: when AI helps to reduce energy consumption
As mentioned above, one key aspect of this complex relationship which is often overlooked, is that AI itself can contribute to reducing energy consumption. Neural network systems are in fact already used to optimise data centre operation by improving cooling processes, load management and hardware resource use.
Thanks to artificial intelligence, it is possible to obtain the same calculation power with less energy consumption. This shows that the relationship between AI and energy is not necessarily going to worsen over time, but could develop in a more balanced way thanks to technological innovation.
The IEA also emphasises how artificial intelligence can play a crucial role in the integration of renewable sources (such as solar and wind power), which are characterised by varying production levels.
Through advanced predictive models, it improves production forecasts, reduces waste and enables more efficient demand management. This approach makes the energy system more flexible and less dependent on fossil fuels.
Another tangible example of the mutual benefits is what is known as the smart grid. AI allows for real time network monitoring, so as to identify anomalies, prevent breakdowns an strengthen the resilience of the electricity network.
In an increasingly complex scenario, with the increase in renewable energy use and AI related electrical loads, smart networks are becoming essential. Nevertheless, the IEA warns that wider digitalisation will also bring new risks, particularly in terms of cybersecurity, which must be managed with great care.
Climate impact and global inequalities
From an environmental point of view, the relationship between artificial intelligence and energy brings both advantages and disadvantages. Greater demand for electricity can result in higher CO₂ emissions if the energy used derives from fossil fuels. This is one of the main risks highlighted by the International Energy Agency.
At the same time however, the widespread use of AI in the energy, industrial and transport sectors may lead to much greater reductions in emissions than the energy consumed directly by AI. The overall climate balance therefore depends on the choice of energy mix and the decarbonisation policies adopted.
Another key element is the inequalities between countries. Economies with access to reliable, abundant, low-cost energy are at an advantage in terms of AI development. However, this risks widening the global technological divide.
Nevertheless, the IEA highlights AI’s potential to improve energy access in less developed areas, thanks to smart solutions for mini-grids and off-grid systems based on renewable energy sources. Again, the relationship between artificial intelligence and energy may be considered a challenge but also an opportunity.
Also according to the Agency, the future of this relationship will largely depend on decisions made now. It is essential that governments andd companies integrate AI into their energy planning, by adopting high efficiency standards, investing in networks and promoting a regular dialogue between the technological and energy sectors.
Artificial intelligence must definitely not by seen as just a new source of demand, but rather as a genuine strategic lever for the sustainable transformation of the energy supply system.
