Data availability for AI isn’t just a volume game
A core concern raised in IDSA’s contribution is the growing tension between AI development and the availability of high-quality training data. Research already projects that, latest by 2030, training next-gen large language models could require more data than humans have ever created. That pressure has driven some companies to turn to social media data, scraped web content, and increasingly, video sources.
But these strategies all have limits. There’s another path, one that Europe could lead: smarter, more efficient AI models trained on curated, high-quality corporate data. Industrial data is often richer and more accurate than the publicly available datasets used today. The issue is trust. Companies are understandably hesitant to share sensitive data without control over how it’s used. That’s where data spaces come into play.
Data spaces: Not just infrastructure, but trust architecture
Data spaces are not platforms or data lakes. They are distributed systems with a defined governance framework that let organizations share data while keeping control. A shared language of standards and protocols ensures interoperability on a technical level, but the core principle is human: your data, your choice.
This decentralized model can support sector-specific collaboration without forcing everyone into the same system. It also builds the kind of trust companies need to share their most valuable data. If designed and scaled correctly, data spaces will unlock a huge source of industrial data ready to train AI models. And they could do it in a way that respects confidentiality, competition and legal boundaries. IDSA’s call is to scale these models now, especially in industrial settings where fragmentation is still a major hurdle.
Legal complexity is slowing progress and there’s a fix
A recurring obstacle is regulatory overlapping. Companies dealing with GDPR, Data Act, AI Act, and sector-specific laws often find themselves in a maze of policies. For instance, when a dataset contains both personal and non-personal data, it’s not clear which rights apply. This ambiguity leads to inactivity and lost opportunities.
IDSA stresses that the EU should focus on harmonizing how existing rules are implemented. Consistent guidance, practical toolkits and simplified compliance processes are urgently needed.
Don’t forget the global context
Europe’s digital economy depends on cross-border data flows. Blocking or over-regulating them doesn’t just isolate Europe, it weakens its global competitiveness. IDSA urges the EU to strike a balance: protect privacy but enable secure international data sharing.
There are already positive models. Japan’s 2025 Digital Ecosystem Partnership reflects many of the same principles as Europe’s data spaces: public-private collaboration, standards-driven governance, and sector-specific alignment. These are the types of partners the EU should engage with early. The takeaway: inclusion is a precondition for interoperability. The more global alignment is incorporated into the design, the less friction there will be later.
The path forward: tools, clarity, and collective action
To move from principles to practice, IDSA outlines several next steps:
- Provide blueprints and training to help sectors create data-sharing projects
- Develop networks of “data territories” to connect local ecosystems
- Clarify how new regulations interact, especially where overlap exists
- Accelerate technical standard development with industry participation
- Streamline compliance through “one-stop-shop” mechanisms
- Support SMEs with open-source tools and targeted regulatory guidance
Above all, collaboration needs to start earlier. Regulations should not be handed down; they should be co-developed with those who must implement them. That means involving businesses – especially small ones – from day one.
A strategy is only as good as its execution
The EU has no shortage of ambition when it comes to AI and data. What’s missing is coordination and clarity. IDSA’s contribution makes the case for data spaces not as buzzwords, but as actionable frameworks that balance innovation with trust. If the upcoming Data Union Strategy gets this right, it won’t just support AI development. It will help Europe build a digital economy rooted in sovereignty, interoperability, and shared value.
