The Sovereignty Gap: Healthcare's path to scaling AI

On May 18, 2026, Danish AI Lab Corti hosted a high-level panel discussion about healthcare AI in Europe and how to ensure European strategic interdependence. The dialogue is intended to continue, and to inform politicians and decision-makers in Denmark and across the European Union.

The debate originated from a shared premise: Europe has a credible path to leadership in healthcare AI and AI more broadly — but only if policy, infrastructure, investment, and implementation begin to move in greater alignment. It is urgent for Europe to both capture opportunities and be less critically dependent on a few, mostly US, tech giants.

The conversation at Corti's headquarters in Copenhagen was open only to invited guests, including founders, researchers, policymakers, industry leaders, and media. The debate was held under Chatham House Rule, meaning information and opinions would not be attributed to specific individuals or organizations.

The conclusions and recommendations in this document are Corti's own, informed and inspired by the panel discussion. They are not necessarily representative of all participants in the debate, but a collection of topics and ideas to consider and to discuss further.

Panelists

Andreas Cleve: CEO and Founder of Corti, creating the AI layer for healthcare, fundraising in Europe, living the compliance gap daily

Kezia Wexøe-Mikkelsen: Analyst at Think Tank Europa, tracking where EU legislation is, where it is going, and where it diverges from reality

Jan Damsgaard: Professor at CBS, Denmark's national Digital Advisor, author of Digital Suverænitet (2026)

Dan Rose Johansen: CEO and Founder of Todai, ten years building and deploying AI in Danish businesses, author of Applying Artificial Intelligence: The Practitioner's Handbook, public debater in Børsen and Finans.dk

Moderator

Connie Hedegaard: Former EU Commissioner for Climate Action, former Danish Minister for Climate and Energy, journalist

Corti’s role

Corti convened this debate as a European-headquartered frontier lab building clinical AI for regulated healthcare. The questions of sovereignty, capability, and dependency are not abstract for us - they shape how we research, build, certify, and deploy every day, across borders that do not yet operate as a single market. We hosted the evening to widen the conversation beyond the people already in it, to test our own thinking against the strongest voices we could bring into one room, and to move the outcome forward for a stronger, more entrepreneurial Europe.

The current situation

Europe speaks increasingly about digital sovereignty. How to become less dependent on tech giants from the US and China. How to build our own platforms, regulate and control, protect our data, our jobs, our music. There has been a lot of fear and a lot of regulation.

Maybe we have spent too long looking backward at what we want to protect, and not long enough looking forward to what we want to build and the great potential ahead of us.

Until very recently, digitalization was, for most Europeans, a matter of productivity, cost reduction, and better services. That has changed quickly, mainly because of geopolitical shifts. What was purely commercial has become a matter of critical infrastructure, security, and strategic interdependence. The reality is that a foreign administration could, in principle, turn parts of Europe off. We have already seen examples of this, including the Chief Prosecutor of the International Criminal Court (ICC), Karim Khan, whose email access at the ICC was reportedly shut down by Microsoft following pressure and threats from US President Donald Trump.

It may not happen on a larger scale. But power exists, and that power can be abused.

Recently, France made a decision that illustrates exactly this: after five years of controversy, the country chose to move its national Health Data Hub, which contains the records of tens of millions of French citizens, off of Microsoft Azure and onto a European cloud provider, Scaleway. That decision took five years to make. The question we are here to ask is: what will it take to make the next one faster?

Despite European momentum, US and Chinese tech companies still dominate globally. According to the Australian Strategic Policy Institute's (ASPI) Critical Technology Tracker, the United States leads in 7 of 44 critical technologies, while China leads in the remaining 37. Europe, therefore, leads in none. 

Digital solutions are something Europe uses, not something Europe owns. If we do not create the conditions for European platforms to compete and win, sovereignty remains a slogan, and Europe's digital future is still outsourced by default.

This conversation is about Europe's path forward, with healthcare AI as our lens. The stakes are highest here. If Europe can get it right in healthcare, it can get it right anywhere.

Conclusion 1: Healthcare is a
European strategic AI opportunity

Healthcare may be the domain where Europe possesses genuine structural advantages: trusted public institutions, longitudinal patient datasets built over decades, deep clinical expertise, and a strong tradition of regulated interoperability. These are assets that cannot be replicated quickly elsewhere. Yet this advantage is not being converted into competitive AI infrastructure. The dominant electronic health record systems used across European hospitals are largely American. If China or the US develops superior AI models on better or larger health datasets, European patients will justifiably demand access to the resulting diagnostics, therapies, and molecules, and Europe will lose the ability to invest in and control its own health innovation ecosystem. Of course, the cost of falling behind isn't only economic; it's clinical. The countries that lead in healthcare AI will set the standard of care for their populations. 

Healthcare will serve as a proving ground for other sectors. If Europe can build and scale health AI at home, it can build and scale anything.

Critically, Europe's advantage may lie not only in building AI models but also in its deep experience implementing and adopting AI in complex healthcare environments. The true competitive edge may be the accumulated capability to turn general-purpose technology into specific, trusted, high-quality clinical outcomes. An edge that compounds with every deployment.

Recommendations

• Designate healthcare as a flagship sector for demonstrating European AI leadership, with dedicated funding and implementation targets.
• Enable secure and well-functioning cross-border health data collaboration within the European Health Data Space (EHDS) framework.
• Support clinically integrated AI deployment environments where European companies can build, validate, and scale, not just conduct research.
• Connect pharma and medtech strengths (such as Denmark's pharmaceutical cluster) to AI development to prevent the erosion of competitive advantage through superior foreign models.
• Invest in clinical innovation environments, both private and public, that can keep pace with AI development cycles.

Conclusion 2: Regulation alone will not lead to European competitiveness or digital sovereignty

Europe has invested heavily in regulation while underinvesting in industrial capacity and scaling mechanisms. The AI Act took nearly a decade to adopt and remains uncertain in its implementation. Compliance frameworks designed at the EU level are often costly and complex for startups and regional health systems to navigate, thereby accelerating market concentration toward the large global platforms they were designed to constrain. Hyperscalers can afford to be individually compliant for each of the 27 EU member states, but startups can't and shouldn’t be asked to.

There is a significant and growing gap between Europe's sovereignty ambitions and what companies, hospitals, and health systems can realistically procure and deploy today. Practitioners on the ground report that building on European alternatives can cost up to 15 times as much as comparable US-based solutions, making sovereign procurement a theoretical rather than practical choice for most organizations.

Recommendations

• Clarify procurement standards for sovereign AI and cloud solutions so that public institutions can make informed, practical purchasing decisions.
• Reduce the fragmentation between EU-level ambitions and national implementation.
• Ensure compliance frameworks are accessible to startups and regional health systems, not only to large technology companies that can easily absorb compliance costs.
• Invest in regulation-fluency as a competitive capability for European companies through certification support, regulatory affairs talent development, and shared compliance infrastructure for startups and scale-ups, so European companies can navigate complex regulation as a strength rather than being excluded by it.
• Accelerate public-private implementation partnerships that move from ambition to deployed infrastructure.

Conclusion 3: Europe needs faster decision-making and more strategic coordination

The EU's instinct toward comprehensive, process-heavy legislation creates lengthy decision-making cycles that do not keep pace with the potential pace of AI development. The gap between where Europe legislates and where member states implement is precisely where sovereignty is lost. Within individual countries, IT systems may not be interoperable with each other. Across borders, frameworks such as Germany's BSI C5 and France's HDS regulate medical data security. Still, they are non-interoperable, preventing cloud providers certified in one country from serving customers in another, even when the security standards substantially overlap. Addressing this fragmentation is not a bureaucratic improvement but a precondition for a functioning European health AI market.

Recommendations

• Strengthen coordination between governments, academia, infrastructure providers, and startups, and move toward operational alignment.
• Create European mission-oriented AI initiatives with clear industrial targets, not only research mandates.
• Use smaller, fast-moving countries like Denmark as rapid implementation testbeds for a broader EU rollout.
• Harmonize medical data security frameworks (e.g., BSI C5 and HDS) so that certification in one country is recognized across the single market.
• Advance the 28th Regime, a unified European legislative framework, as a practical tool to reduce compliance fragmentation for companies operating across borders.

Conclusion 4: The aim is strategic interdependence, not isolation

The debate surfaced an important reframing: the dominant goal of 'winning AI' by building foundation models to rival the main US or Chinese alternatives is unlikely to succeed and may be the wrong ambition. The goal is not independence but rather a strategic interdependence: the ability to participate in a global technology ecosystem on Europe's own terms, with sufficient resilience, leverage, and alternatives such that any dependence does not become a vulnerability.

The value of AI in health lies in adoption, implementation, and clinical integration: turning general-purpose technology into specific, trusted, high-quality outcomes. Europe has strong implementation capacity, clinical governance traditions, and trusted institutional frameworks. These are competitive assets for the application layer, even if the foundation model layer remains dominated by US and increasingly Chinese players. The risk of dependency is real, but mutual interdependence, managed intelligently through dual-sourcing strategies and exit-by-design contracts, is more achievable than full independence, and more stable than isolation.

Recommendations

• Resist the impulse to replicate US or Chinese business models; build on European values around data governance, patient rights, and public accountability. Compliant mediocrity is not sovereignty.
• Redefine European AI ambition around leadership in adoption, clinical integration, and sector-specific application — not only foundational model development.
• Develop national and EU-level 'dependency scores' for critical digital infrastructure, so that organizations understand and can actively manage their exposure.
• Promote exit-by-design procurement standards: public contracts with technology providers should require data portability, interoperability, dual-sourcing options, and, where appropriate, counter-purchase agreements.
• Welcome and learn from international technology. The goal is strategic resilience, not isolation.

Conclusion 5: Europe has talent and
investment, but lacks positive ambition

Several of the world's leading AI researchers are European by origin, including Yann LeCun, Chief AI Scientist at Meta (French), and Demis Hassabis, co-founder of DeepMind (British). What is needed is a market environment in which talent can stay, companies can grow, and capital can be deployed at the speed and scale required. Brain drain from Europe to the US and UK is accelerating as AI becomes the defining technology of the era.

Compounding this, women remain dramatically underrepresented in AI development and deployment, representing a vast, untapped talent pool whose absence reduces both the quality of innovation, and the diversity of applications being built. Europe is also experiencing complacency: the assumption that a high quality of life is a permanent endowment, rather than something that must be actively built and renewed. Europe’s capital markets remain too shallow and risk-averse for high-growth technology companies, with weak IPO activity and limited access to large-scale growth capital making it difficult for companies to scale beyond the early stages.

Recommendations

• The EU and governments should avoid supporting specific companies, but should support winning sectors to create conditions for clusters to form and compete.
• Complete the Capital Markets Union and Digital Single Market as urgent, time-bound policy priorities.
• Actively address the brain drain by making Europe a more attractive environment for AI talent to build and scale companies.
• Invest in inclusion: the underrepresentation of women in AI is a strategic liability, not only an equity issue. Address barriers including regulatory uncertainty, confidence gaps, and structural exclusion.
• Change the public narrative around ambition: European political culture tends toward risk aversion and the defense of the status quo. Leaders should articulate a positive, concrete vision of a better future.
• Pursue an active industrial policy that selectively and intelligently mirrors the mechanisms that the US uses to stimulate its technology economy, through defense procurement, public research investment, upskilling of labor, and strategic contracts.

Where does this leave us?

The debate at Corti did not produce a single, unified prescription for the path toward greater strategic interdependence in health AI and AI in general. That was never the intention. What it produced was something more valuable over the long term, and for the ongoing conversation: a shared map of the terrain, with honest disagreement about the right routes through it.

One panelist framed the moment as existential: Europe either becomes good at innovating and creating new companies, or it becomes a place nice to visit to see how our ancestors used to live.

Another offered cautious optimism, pointing to real movement: positions in Northern Europe shifting on sovereignty, the 28th Regime gaining ground, the Commission's tech sovereignty package taking shape.

A third challenged the framing of the debate itself, arguing that 'winning AI' is the wrong ambition. They argued that the value lies in adoption and implementation, and that Europe should stop playing someone else's game and find leverage in the position it actually holds.

Another closed on the most pointed note of all: that the most dangerous reflex in European political culture is the assumption that this is as good as it gets. That a hundred years from now, people will not look back at us thinking we were so blessed. We must dare to imagine a brighter future.

Across the panel, there was a convergence on a shared rejection of complacency and a shared belief that Europe has the agency to build something better. No one in the room believed the status quo was acceptable. No one believed the gap between European ambition and European reality is closing fast enough.

Europe is not without assets. It has talent, institutions, clinical depth, and public trust. It includes examples such as France's Health Data Hub decision, the Digital Markets Act, and Denmark's digital public services that show what is possible when ambition meets implementation. Whether those examples become a pattern rather than fragmented exceptions is the open question.

The closing note of the evening was one of cautious optimism with an embedded warning. Europe has a tendency, as one panelist put it, to sit in the garden and assume the garden will always be there. It will not. Not without tending. The choice is not between a comfortable present and a risky future. The choice is between building deliberately and being built around.

—

This report was produced by Corti (corti.ai) following the panel event held on May 18, 2026, at Corti's Copenhagen headquarters. It is informed by the panel conversation and inspired by the perspectives shared that evening, but the views and recommendations expressed are Corti's own. The event was held under the Chatham House Rule.