AI and GenAI in smart grids – a guide for action

A new ETIP SNET paper on AI and generative AI in smart grids reviews opportunities and challenges and offers a strategic roadmap for stakeholders across the energy sector.

Over four months in the making with as broad input as possible, the 130-page paper is aimed to present a vision and a pathway for integrating AI and GenAI into the fabric of the energy system, particularly in Europe, and thereby accelerate the transition towards a sustainable and resilient future.

In many respects complementary to a major report on AI from the IEA appearing around the same time, in which a significant focus is on the growing energy demand of data centres, the ETIP SNET paper seeks to provide a comprehensive analysis of the transformative potential of AI and GenAI in smart grids.

Insights are intended for broad groups of stakeholders, including policymakers, network operators, technology developers, researchers and SMEs and startups as well as consumers.

Also of interest
EU Energy Projects Podcast: The potential of AI and GenAI in Europe with ETIP SNET

The paper, ‘Unlocking the Potential of AI and Generative AI in European Smart Grids’, starts with a comprehensive overview of EU legislation pertaining to AI and smart grids, of which the AI Act, which was adopted in 2024, and the new EU AI Continent Action Plan are the most important, with significant implications for DSOs and TSOs.

Within this context, the paper comments that developing and deploying trustworthy, grid-specific foundational models is crucial for unlocking the full potential of and GenAI in enhancing grid operations safely and effectively.

Technical foundations

Moving on to the technical foundations of AI and GenAI in smart grids, the paper highlights the need for the availability of robust, high quality data, before delving into applications and implementations in production, some early stage and still prospective but with most in development.

Just two are identified as mature at TRL 9, models for load forecasting and for renewable energy forecasting.

A lesson highlighted is that effective implementation of AI and GenAI in smart grids requires partnerships among grid operators, vendors, digital service providers and regulators, ensuring that solutions are practical, scalable and aligned with regulatory requirements.

In addition, closing the research gaps is essential to fully realise the potential of AI in driving a more efficient, resilient and sustainable energy infrastructure.

Among challenges that are identified are regulatory uncertainty, with ambiguity in the application of the AI Act and GDPR to specific energy use cases, data access and governance with siloed, insufficient or inaccessible data hindering AI model development and deployment and a lack of personnel proficient in both AI and energy systems domain knowledge.

Others include standardisation gaps, trust, cybersecurity risks and the need for significant funding for R&D, infrastructure and skills development.

A roadmap for implementation

The paper proposes a phased roadmap for implementing AI and GenAI solutions in European smart grids, with a flexible ‘shopping list’ of action points.

Phase 1 to ‘build the foundation’ is envisaged to take up to two years and is focussed on rapidly establishing the essential groundwork – regulatory clarity, initial data access, pilot projects, foundational skills and trust mechanisms.

Key technologies like open-source AI models, synthetic data generation and initial pilot projects drive implementation readiness. Simultaneously, establishing data spaces, governance frameworks, standards compliance and training programmes ensures the electricity sector is prepared legally, technically and organisationally.

Phase 2 to ‘scale and harmonise’ over two to five years is focussed on moving from pilots to broader deployment, harmonising standards and regulations across the EU and strengthening infrastructure and collaboration.

Medium-term actions are potential high impact initiatives that cannot be done without carefully consulting industry and policy stakeholders and stakeholders are invited to proactively start such explorations on these.

Then in the long term, beyond five years, phase 3 to ‘achieve full integration and optimisation’ is aimed to realise the full potential of AI for a highly optimised, resilient and sustainable energy system, establishing EU leadership and ensuring continuous adaptation.

Long term action points require developments such as the result of medium term actions or technological progress.

The paper, which is accompanied with a series of six briefing notes for the different stakeholder groups, describes the action plan as not just a list but a structured roadmap.

“Each step builds trust, capability, and the technical/regulatory infrastructure needed for the next level of integration.”