Artificial intelligence in Morocco: what is its environmental impact?
Author
Oussama BOUOUAROUR
Artificial intelligence can sometimes feel almost immaterial.
You open an app, ask a question and, a few seconds later, an answer appears. No smoke, no engines, no visible waste.
And yet, behind that simplicity lies a very physical infrastructure.
Servers. Data centres. Cooling systems. Electricity. Water. Electronic components.
As Morocco accelerates its development in artificial intelligence, one question deserves to be asked now:
Can the country become a major player in AI without repeating the environmental mistakes made by other industries?
Morocco no longer wants to simply use AI
Morocco’s ambitions now go far beyond adopting a few digital tools.
Through Maroc Digital 2030 and the AI Made in Morocco 2030 roadmap, the country aims to develop its own artificial-intelligence capabilities, strengthen its digital infrastructure and build an ecosystem that brings together universities, businesses and research centres. Maroc Digital 2030 et de la feuille de route AI Made in Morocco 2030, le Royaume cherche à développer ses propres capacités en intelligence artificielle, renforcer ses infrastructures numériques et construire un écosystème associant universités, entreprises et centres de recherche.
At the beginning of 2026, the government announced its intention to generate up to 100 billion dirhams in AI-related economic value by 2030create around 50,000 jobs in the sector and train 200,000 graduates in artificial-intelligence skills.
The country also plans to expand its sovereign cloud and data-storage capacity. Among the announced projects is a 500 MW data centre in Dakhla, expected to be powered by renewable energy. 500 MW à Dakhla, qui devrait être alimenté par des énergies renouvelables.
From an economic and technological perspective, the potential is considerable.
Mais si l’on veut parler d’IA responsableBut if we are going to talk about responsible AI, the discussion cannot stop at data protection, jobs or algorithmic ethics.
We also need to talk about its environmental footprint.
Behind every algorithm, there is a machine somewhere
When we use generative artificial intelligence, the calculations do not happen “in the cloud” in an abstract sense.
They take place in data centres filled with processors that may run continuously.
And this infrastructure requires energy.
According to the International Energy Agency, global electricity consumption from data centres reached around 485 TWh in 2025 and could approach 950 TWh by 2030. Consumption from data centres specifically geared towards AI is growing even faster and could triple during that period. 485 TWh en 2025 et pourrait approcher 950 TWh en 2030. La consommation des centres spécifiquement orientés vers l’IA augmente encore plus rapidement et pourrait tripler sur cette période.
At a global level, this still represents only a fraction of total electricity consumption.
But the figure alone does not tell the whole story.
A data centre is not spread evenly across the planet. It is built somewhere. quelque part.
Connected to a specific electricity grid.
Located in a region with its own energy constraints.
And sometimes in a place where water resources are already under pressure.
It is this local dimension that becomes particularly relevant for Morocco.
Water could be just as important a question as electricity
We naturally associate data centres with energy consumption.
Much less with water use.
And yet, servers generate a great deal of heat. Depending on their design, data centres use different cooling systems, some of which require water. The indirect water use associated with electricity generation must also be taken into account.
And in Morocco, this deserves particular attention.
The country is already facing structural water stress. The World Bank estimates that Morocco’s water resources have declined by almost 30% over the past six decades. stress hydrique structurel. La Banque mondiale estime que les ressources en eau du Maroc ont diminué de près de 30 % au cours des six dernières décennies.
This obviously does not mean that data centres should not be built.
But it does mean that their location, cooling technologies and water sources should become criteria that matter just as much as their computing power.
An infrastructure can be highly innovative while still being poorly suited to its environment.
And this is probably one of the most important lessons to remember when we talk about sustainable technology.
“Powered by renewable energy” does not solve everything
A data centre powered by solar or wind energy is clearly a major improvement on one powered mainly by fossil fuels.
But the word renewable does not automatically mean impact-free. sans impact.
We also need to consider:
- the total amount of energy consumed;
- storage and grid requirements;
- cooling systems;
- water use;
- the lifespan of servers;
- the materials required to manufacture them;
- and what happens to equipment when it becomes obsolete.
The United Nations Environment Programme reminds us that AI’s footprint extends across its entire lifecycle: mineral extraction, equipment manufacturing, energy use, water use and electronic waste.
In other words, replacing fossil-based electricity with renewable electricity is essential.
But it is only one part of the equation.
Should we see AI as an environmental problem, then?
Not really.
And this is where the conversation becomes much more nuanced.
Artificial intelligence can itself become a powerful tool for reducing certain environmental pressures.
It can help us anticipate energy demand more accurately, optimise electricity grids, detect anomalies, improve infrastructure maintenance and analyse vast volumes of environmental data much faster than a human team could.
For example, the International Energy Agency estimates that the large-scale adoption of existing AI applications could help the global electricity sector optimise its operations and unlock up to 175 GW of additional capacity on transmission gridswhile reducing certain costs.
AI could therefore both:
consume more resources and help us use the resources we already have more effectively. et nous aider à mieux utiliser celles que nous avons déjà.
Both realities can be true at the same time.
So the real question is probably not:
Is artificial intelligence good or bad for the environment?
But rather:
Which uses are truly worth the resources they require?
And that question deserves much more attention.
Not every use of AI has the same value
Using a powerful model to analyse complex climate data does not necessarily have the same value as using it to generate fifty variations of an image that no one will ever use.
In the same way, not all models require the same amount of computing power.
Not all data centres are equally efficient.
And their footprint depends heavily on where they are located, the electricity system that powers them and the cooling technologies they use. Recent research shows that these factors can significantly affect the energy and water footprint of AI infrastructure.
There is therefore an idea that we will probably need to gradually build into our digital habits:
digital sobriety.
Not stopping the use of technology.
But choosing the right tool for the need.
Because making a technology more efficient is not always enough.
If it becomes ten times more accessible and we use it one hundred times more, total consumption may still continue to rise.
Morocco may have a particular opportunity
There is an advantage to arriving at a moment when these questions are only just beginning to be asked.
Much of Morocco’s AI-related infrastructure is still yet to be built.
That means some environmental considerations can be integrated from the design stage, rather than having to be corrected years later. dès leur conception, plutôt que d’essayer de les corriger plusieurs années plus tard.
Morocco also has significant renewable-energy potential, while its digital strategy includes an expansion of data and cloud infrastructure.
The opportunity is therefore clear:
not simply to build more data centres,
but to try to build better data centres. de meilleurs centres de données.
This could involve more efficient equipment, greater use of renewable energy, cooling systems adapted to the local climate, responsible water management, better use of servers and transparent measurement of their environmental footprint. UNEP specifically recommends that these elements be integrated into the planning and procurement of new digital infrastructure.
Perhaps we need to broaden our definition of “responsible AI”
When we talk about responsible artificial intelligence today, we naturally think about privacy, algorithmic bias, misinformation or jobs.
These are essential issues.
But truly responsible AI should also be able to answer far more practical questions:
How much energy does it use?
Where does that energy come from?
How much water does its infrastructure require?
How long does the equipment remain in service?
And is the environmental impact created justified by the value produced?
Morocco has clearly chosen to make artificial intelligence one of the drivers of its digital development.
So the question is no longer whether this transformation will happen.
It has already begun.
The real challenge will be deciding what form we want it to take. quelle forme nous voulons lui donner.
Because technological progress does not automatically become sustainable progress.
But if the right questions are asked early enough, it can certainly become just that.