Data sharing to support the training and development of AI foundation models in the energy sector

What is this call about?

The topic funds projects that develop secure data sharing and AI foundation models (open-source wherever possible) for the energy system, tested in laboratory and real-life pilots. Projects are expected to reach technology readiness level 7-8.

What does this call fund?

Scope

The quality of AI models depends on the amount, quality and representativeness of training data. Projects should develop, train and test AI foundation models using extensive datasets to speed up the energy transition in key areas of the sector. Building on earlier work on the common European energy data space and the Data 4 Energy expert group, they are expected to show access to large relevant datasets when submitting the proposal and develop innovative, scalable data governance and sharing strategies that remain operational after the project. Strategies to explore include data space connectors, federated training that allows confidential data to be used, synthetic data to fill gaps, and an AI gym.

What projects are expected to do

• Develop foundation models for at least one energy use case, such as planning and operation of electricity grids (static power flow and dynamic electromagnetic transient modelling), forecasting, congestion management, anomaly detection, fault diagnosis, predictive maintenance, flexibility management, demand-side energy efficiency, or smart and bidirectional charging of electric vehicles.
• Test and validate the models in laboratory demonstrators and real-life pilots, using facilities from earlier AI testing and experimentation facility projects; regulatory sandboxes may be considered.
• Bring together data owners (for example utilities, grid operators and asset owners), application developers (start-ups, SMEs, hyperscalers) and model deployers (industry, grid operators, equipment manufacturers).
• Projects can benefit from the computing capacity of the Commission's AI factories, in particular the three aimed at energy applications.
• Respect ethical, safety and security principles, with transparency, explainability and accountability built in, pursue open-source development where feasible, and avoid bias so that data represent diverse populations.
• Contribute to and take part in the BRIDGE initiative.

Activities are expected to reach technology readiness level 7-8 by the end of the project, and may start at any level.

Expected outcomes

All of: effective, innovative methods to gather, share and use large energy datasets for AI training while ensuring privacy and security; advanced and, wherever possible, open-source foundation models supporting digitalisation of the energy system through better grid observability, supply and demand forecasting, storage and renewables integration, demand-side flexibility and energy efficiency; more cooperation and interoperability among energy actors for secure data exchange; and AI development methods whose results stay findable, accessible, interoperable and reusable beyond the project, open source and adaptable to technology change.

When is the deadline?

1 December 2026 (26 days)