AI in energy consultancies: uses and legal framework

AI and energy: a two-way link

Artificial intelligence has become an energy issue in its own right. It consumes electricity, sometimes a lot. It can also save electricity, provided it is put to work for energy efficiency.

The International Energy Agency (IEA) devoted a special report to this two-way link, Energy and AI, published on 10 April 2025. Its figures give the order of magnitude of what is at stake.

According to the IEA, wider adoption of existing AI applications to optimise industrial processes would deliver energy savings greater than the total current consumption of Mexico. For buildings, the agency also identifies the obstacles: fragmented ownership, low digitalisation and insufficient incentives.

These figures are potentials, not results already achieved. They nevertheless show that energy efficiency is one of the areas where AI can produce measurable gains. They also remind us that its use has an energy cost of its own.

Putting consumption in perspective

The growth of data centres raises legitimate concerns. The IEA places it in a wider picture: according to its report, data centres account for about a tenth of the global increase in electricity demand by 2030. That is less than industrial motors, air conditioning in homes and offices, or electric vehicles.

The picture differs from one country to another. In advanced economies, where electricity demand had stagnated for decades, the IEA estimates that data centres will account for more than 20% of demand growth by 2030. For grid operators and large consumers alike, AI is therefore also a question of connection and capacity.

For a company that uses AI, the practical conclusion is this: considered use, focused on the tasks where the tool brings a real gain, is better than systematic use. Digital sobriety joins energy efficiency here.

What AI changes in an energy consultancy

An energy consultancy handles a large volume of information every day: site surveys, consumption data, equipment datasheets, regulatory texts and specifications. A large part of the work consists in matching this information, checking that it is consistent and drawing a reliable deliverable from it.

Generative AI and data analysis tools come into play precisely on these tasks. They can speed up the reading of a long regulatory text, spot an inconsistency in a data set, prepare a first report structure or track the evolution of a body of texts. The gain is measured in time given back to engineers for analysis and advice.

These tools do not replace expertise. A language model can produce a wrong figure with confidence, mix up two versions of a text or invent a reference. In a field where a calculation error can cost a subsidy or distort the sizing of an installation, every result must be checked by a professional who takes responsibility for it.

This is the line we follow at IHD. AI serves the rigour of our studies and does not replace it: every figure in a deliverable remains checked, sourced and owned by an engineer.

The legal framework: the AI Act

Who is concerned

Regulation (EU) 2024/1689, known as the AI Act, was published in the Official Journal of the European Union on 12 July 2024. It lays down harmonised rules on artificial intelligence in the Union. The text can be read on EUR-Lex.

The regulation distinguishes several roles. A deployer is a natural or legal person that uses an AI system under its own authority, except where the system is used in the course of a personal non-professional activity. In our reading, a company that makes an AI assistant available to its teams for their work falls within this definition.

AI literacy

Article 4 of the regulation deals with AI literacy. The regulation defines it as the skills, knowledge and understanding that make it possible to deploy AI systems in an informed way, while weighing their opportunities and risks.

This article was rewritten by Regulation (EU) 2026/1744 of 8 July 2026, published on 24 July 2026. Providers and deployers must now take measures to promote AI literacy among their staff. The text specifies that this obligation does not require them to guarantee a specific level of literacy for each individual. We explained this change in our article on Article 4 of the AI Act.

High-risk systems in energy

Annex III of the AI Act classes as high-risk certain uses linked to critical infrastructure. It covers AI systems used as safety components in the management and operation of the supply of water, gas, heating or electricity.

Under Regulation (EU) 2026/1744, the obligations applying to these Annex III systems take effect on 2 December 2027. For systems falling under Annex I, the deadline is set at 2 August 2028.

Personal data: the GDPR applies

Using an AI tool does not suspend the General Data Protection Regulation. As soon as an AI system processes personal data, for example an occupant's name or the consumption of an identified dwelling, the GDPR applies in full.

The CNIL, the French data protection authority, sets out the principles to follow. An AI system using personal data must pursue a specified, legitimate and explicit purpose. It must rely on one of the six legal bases provided for by the GDPR: consent, legal obligation, performance of a contract, public interest task, protection of vital interests and legitimate interest. Finally, an AI system cannot be used on personal data that was collected unlawfully.

For an energy consultancy, the consequence is concrete. Before submitting a client file to an AI tool, you need to know where the data goes, who has access to it and for what use. A consumer tool used without precautions can expose confidential information.

What this means for you

For the head of a consultancy or an energy SME

Your teams probably already use AI tools, with or without a framework. The AI Act asks you to promote their literacy in these tools. The GDPR requires you to control the data that passes through them. A clear usage policy and suitable training address both.

For an installer or a works contractor

AI can ease the preparation of quotes, technical files and administrative documents. Responsibility for the content remains yours: an automatically generated document must be reviewed with the same care as a document written by hand.

For an industrial company or a property manager

Optimising processes and buildings with AI is among the potentials identified by the IEA. However, solutions that control critical energy supply installations may be high-risk within the meaning of the AI Act: choosing them deserves a regulatory review beforehand.

For a heating network or electricity operator

AI systems used as safety components in the supply of gas, heating or electricity will be subject to the high-risk requirements from 2 December 2027. This timetable leaves time to prepare, provided you start early.

What we recommend

Adopting AI in a technical energy profession means three things: knowing what the tools do well and badly, framing the data that flows through them, and keeping human verification at the centre of every deliverable.

IHD supports energy companies on all three. Our artificial intelligence training courses are designed for technical teams and start from the concrete situations of our professions. Our experience as an energy consultancy, described on our energy audit and engineering page, allows us to speak the language of your engineers.

To build a training programme suited to your teams, contact us.