Data Space for Manufacturing

Expected Outcome

Deliverables

The awarded proposals are expected to deliver:

• Solutions that allow the collection of large, high-quality data from real industrial environments, from different manufacturing systems/sectors.
• Solutions that allow access to these large datasets to train Generative AI models that respond to real-world industrial needs and challenges, preferably relying on trusted third parties hosting data and training compute on behalf of authorized AI developers.
• Agreements with specific AI factories providing information on the data sets available and the conditions for using them for training AI models.
• Interoperability and governance framework of manufacturing data spaces including from national industrial data spaces to ensure that mechanisms to aggregate data for training AI could be taken up at EU level.

Objective

Aligned with the previous WP, this action aims to support the continued uptake and further expansion of the data space for manufacturing. The primary objective is to develop legal, technical and business solutions to pool sufficient data and enable authorized AI developers to have direct or indirect data access to train generative AI models specifically for the manufacturing sector. This initiative seeks to reinforce the role of the data space for manufacturing as a major productivity enhancement and collaboration mechanism within the EU.

Scope

This initiative will support up to three data-collection projects, with around EUR 3 million co-funding from Digital Europe each, focused on manufacturing use cases to unlock advanced AI models, for example, predictive maintenance, process automation, supply chain management, product design and development and sustainability in production, and increase productivity in industrial environments such as purchases, logistics, resource planning and production halls. These projects aim to collect massive, high-quality data from real industrial environments, ensuring proper labelling where relevant, that could be used to train or finetune generative AI models for the manufacturing sector. The data collection has to be relevant for developing AI applications that can significantly benefit major EU manufacturing sectors (such as automotive, chemical, aeronautics and energy-intensive industries). The project proposal has to clearly identify the target sectors, the stakeholders, and have preliminary agreements about the intended data exchange.

The data-collection projects will develop both technical and business solutions to enable authorized AI developers to have direct or indirect data access and utilize these large datasets while fully respecting the data holders' control over their data.

Each data-collection project should propose clear use cases to ensure alignment with real-world needs and challenges. Ideally, the AI developers interested in using such data should be already identified in the proposal.

The proposal will also ensure technical and legal solutions to make the generated datasets available to users of AI Factories. This will also enable AI Factories to leverage these datasets for the development of AI applications. To this extent, the inclusion of AI Factories in the project will be considered an advantage.

Consortia are encouraged to use data intermediaries, as outlined in Chapter III of the Data Governance Act, or other appropriate mechanisms to manage the secure access to and processing of these datasets.

The initiative must also work in close partnership with the Data Spaces Support Centre to ensure alignment and interoperability with the broader ecosystem of data spaces implemented with the support of the Digital Europe Programme. Additionally, the action must coordinate with AI Factories to ensure that the datasets generated can be effectively used in conjunction with data already available facilitating in this way collaborative approaches for the development of advanced AI models.

Considering financial sustainability from the outset is crucial in the development of Common European Data Spaces to ensure their long-term viability and effectiveness. By establishing a sound financial model, developers can secure the resources necessary to adapt to evolving technological landscapes and user needs, ensuring that data spaces remain robust and beneficial over time.