AI and Climate Change: An Energy Ally or a New Strain on the Grid?
On the occasion of the International Day against Climate Change, artificial intelligence (AI) is emerging as a key tool for observing, anticipating and responding to complex environmental processes. By analysing satellite imagery, meteorological data, sensors and historical records, it is now possible to detect changes in ecosystems, identify sources of emissions, predict extreme weather events and guide mitigation and adaptation measures with greater precision.
One of the fundamental pillars of climate change mitigation is the transition towards a CO₂-free energy system. In this field, AI is already being actively used to optimise the management of energy assets.
For example, by analysing sensor time series, thermal images and meteorological variables, machine learning algorithms can detect early signs of degradation in solar modules, abnormal vibrations in wind turbines or hotspots in substations.
This enables a shift from corrective to predictive maintenance, reducing downtime and extending the operational life of facilities. At grid operation level, AI has numerous applications: it improves wind and solar generation forecasting, optimises battery charging and discharging cycles, reduces technical losses and helps manage electricity demand in real time.
The other side of the coin: the challenge of growing demand
However, the same technology that helps optimise the energy system and accelerate its transition also introduces a new source of concentrated and rapidly growing electricity demand.
According to the International Energy Agency (IEA), data centres consumed around 415 TWh (terawatt-hours) in 2024, approximately 1.5% of global electricity consumption. This figure is expected to exceed 945 TWh by 2030, approaching 3% of global electricity demand, driven largely by AI.
Although this percentage may appear relatively small at a global level, the real impact lies in its geographical concentration: when these facilities are located in specific areas, their electricity demand can become highly concentrated.
A single data centre may require power equivalent to that of a medium-sized city within an extremely short timeframe. This can place critical pressure on substations, local distribution networks and regional development plans.
The paradox is clear: a technology capable of reducing losses and improving efficiency can simultaneously become an inflexible load that requires physical infrastructure to be expanded or redesigned.
There is a significant difference between a data centre capable of shifting certain processes to periods of high renewable generation and one that maintains constant electricity consumption, even during peak demand hours or periods of high carbon intensity.
Will AI be integrated as a flexible asset that contributes to the stability of the electricity system, or will it become a rigid source of demand that shapes regional energy planning?
Four criteria for assessing its climate value
The fundamental question is not whether AI consumes energy, but whether its net contribution is positive and how it influences resource planning.
Its relationship with the electricity sector works both ways: AI needs electricity to operate, but it also helps generate, distribute and consume it more intelligently. Its real climate balance will depend on whether the emissions and resources it saves exceed those it consumes.
Looking ahead, the climate value of AI should not be measured solely by its potential, but by what it can actually prevent: emissions, energy losses, equipment failures and unnecessary fossil fuel infrastructure. To achieve this, we should assess AI against four fundamental criteria:
Net emissions balance
Positive and independently verified.
Optimisation capacity
Reducing energy losses and extending the operational life of assets.
Operational flexibility
Ability to adapt to the conditions of the electricity grid.
Radical transparency
Regarding energy consumption, geographical location and the source of its electricity supply.
When properly designed and integrated, AI can become a decisive driver of efficiency and resilience. If poorly planned, however, it risks becoming another source of pressure on the very energy system it aims to decarbonise.
Source: International Energy Agency (IEA). Energy demand from AI – Energy and AI .
