More than half of the electricity consumed by data centers comes from fossil fuels. The question is not to give up on innovation, but to manage it intelligently.
The Shift Project has just published a report which calls out: the electricity consumption of data centers could triple between 2023 and 2030. As a presenter of conferences on climate issues, I regularly meet managers faced with a complex equation: how to integrate AI into their strategy while respecting their environmental commitments?
The question is not to give up on innovation, but to manage it intelligently. Because the figures are clear: in 2024, more than half of the electricity consumed by data centers comes from fossil fuels. It is time to move away from the incantatory discourse on “green AI” and enter into a pragmatic and measured approach.
Understand before acting
For years, energy efficiency gains have offset digital growth. This period is over. The Shift Project report demonstrates this: global electricity consumption in data centers is expected to increase from 460 TWh in 2022 to 650-1050 TWh by 2026. In France, digital technology already represents 4.4% of our national carbon footprint.
These figures should not paralyze businesses, but enlighten them. I observe the same dynamic among decision-makers: those who anticipate these issues are better positioned than those who are subject to them.
The opportunity for differentiation
The most inspiring leaders are not those who renounce innovation, but those who question it. They ask themselves: what use of AI really creates value? What is the benefit/impact ratio?
This approach is not a constraint. It’s a competitive advantage. In a context where investors, customers and talents are increasingly scrutinizing environmental commitments, knowing how to manage technological choices becomes a strategic asset.
As the Shift Project points out, it is necessary to prioritize use cases rather than deploying AI in a generalized manner. This prioritization may seem restrictive, but it forces us to clarify what really matters to the company.
Three concrete levers of action
1. Establish a climate analysis grid for AI projects
Before each deployment, ask three questions: What is the estimated energy footprint? What business benefit justifies this energy investment? Is there a less greedy alternative?
This grid does not block innovation. It directs it towards uses with higher added value. Using AI to optimize your supply chain and reduce your transportation emissions makes sense. Using it to generate automated marketing content may be worth questioning.
2. Dialogue with your cloud providers
Companies have considerable power to influence their service providers. By systematically asking about the energy mix of their data centers, the carbon intensity of their services, and their decarbonization trajectory, you create positive pressure.
Some suppliers will be transparent, others evasive. This information must become a selection criterion in the same way as technical performance or cost.
3. Communicate about your decisions
The Shift Project is clear: if an AI solution cannot be deployed in a manner compatible with the carbon constraint, it must be abandoned. Rather than hiding these renunciations, value them. Explaining why you have chosen not to deploy a particular AI service for environmental reasons means asserting strategic coherence.
From field ecology to tech strategy
We talk a lot about “office ecology”, this approach disconnected from physical realities. The alternative is field ecology: that which concretely measures the impact of each choice, which knows the real cost of each resource.
This approach applies perfectly to technological choices. Not all uses of AI are equal. AI is an enabler that accelerates the system in which it is placed. It is up to companies to decide which system they want to accelerate: that of overconsumption or that of energy efficiency?
Anticipate rather than endure
RTE anticipates a possible tripling of data center consumption by the middle of the next decade, to reach 23 to 28 TWh, or nearly 4% of national consumption. In a context of reindustrialization and electrification of transport, this energy tension will create trade-offs.
Companies that have mapped their AI uses, assessed their impact, and built a prioritization strategy will be better equipped. Those who wait will potentially face regulatory constraints or increases in energy costs.
Towards sustainable digital maturity
AI is neither good nor bad. It is a powerful tool whose impact depends on the intelligence with which we deploy it. The Shift Project offers a clear methodology for assessing the climate compatibility of each AI deployment. These tools exist and are available to businesses.
Digital sobriety is not a barrier to competitiveness. It is a principle of efficiency. The most advanced leaders are not those who renounce innovation, but those who direct it with lucidity.
The Shift Project report offers us a valuable framework for structuring this reflection. It is up to us, economic actors and citizens, to make it a lever for transformation rather than a source of anxiety. Companies that know how to combine technological innovation and energy efficiency are already building the competitive advantage of tomorrow.
