CEPS Launches Task Force on the Apply AI Strategy: What the EU's Implementation Gap Actually Looks Like

In Brief

The Centre for European Policy Studies (CEPS) has announced a Task Force dedicated to the European Commission's Apply AI Strategy, running from June 2026 to February 2027. The initiative brings together senior executives, policymakers, and researchers across three sector tracks: Healthcare & Pharma, Automotive & Mobility, and Government & Public Sector. The Task Force aims to address what CEPS Director of Research Andrea Renda has identified as significant implementation challenges: no dedicated budget, unresolved infrastructure questions, and a history of EU data-sharing initiatives that have struggled to deliver results.

The gap between EU AI strategy and EU AI implementation is precisely the kind of problem that benefits from structured disagreement. For those tracking how Europe translates policy ambition into operational reality, Human x AI Europe in Vienna on May 19 offers a space to work through these questions directly.

The Strategy-Implementation Divide

When the European Commission launched its Apply AI Strategy in October 2025, it marked what many observers called a decisive turn in EU technology policy. The strategy builds on the Draghi Report's call for renewed industrial competitiveness and the Competitiveness Compass's vision for a "CERN for AI." It targets AI deployment across eleven strategic sectors, from healthcare and pharmaceuticals to automotive, mobility, and public administration.

The question worth asking: is this a strategy disagreement, an implementation disagreement, or a resources disagreement?

CEPS appears to believe it's primarily the latter two. As Andrea Renda noted in a February 2026 analysis, the Apply AI Strategy "comes with a lot to do but no budget." The strategy adopts an ecosystem approach, connecting use cases with infrastructure, data governance, and skills needs. It avoids one-size-fits-all solutions in favor of tailored, sector-specific flagships. These are defensible design choices.

But defensible design choices without implementation mechanisms remain, at best, aspirational documents.

What the Task Force Actually Does

The CEPS Task Force on the Apply AI Strategy represents an attempt to convert policy language into operational specifics. The structure deserves examination because it reveals what CEPS believes the strategy is missing.

The Task Force operates across three sector tracks:

Track 1: Healthcare & Pharma covers AI-assisted diagnostics, drug discovery, precision medicine, the European Health Data Space (EHDS), and regulatory pathways under both the AI Act and the Medical Devices Regulation.

Track 2: Automotive, Mobility & Transport addresses software-defined vehicles, autonomous driving, agentic AI for manufacturing and supply chains, connected vehicle data ecosystems, and harmonized type-approval frameworks.

Track 3: Government & Public Sector examines workflow automation, citizen-facing AI services, public procurement of AI, sovereign AI infrastructure, and AI Act compliance for high-risk public sector applications.

Each track applies a consistent five-pillar analytical framework: key use cases, infrastructure requirements, data governance, flagship applications, and sovereignty and implementation roadmaps. The consistency matters because it enables cross-sectoral comparison. When healthcare's infrastructure requirements differ from automotive's, that difference becomes visible and discussable.

The Hard Questions the Strategy Hasn't Answered

CEPS is direct about the challenge: the Apply AI Strategy carries a heavy implementation burden, no dedicated budget, and unresolved questions about infrastructure, data sharing, and policy coherence.

Consider the infrastructure question. The strategy references AI Factories and Gigafactories, but as Renda has asked: will these initiatives actually serve industrial sectors' needs? Simply relying on large-scale data centers and powerful GPUs (Graphics Processing Units, the specialized chips that power AI computations) isn't going to be enough. Every sector requires different compute and software architectures.

Then there's the data-sharing problem. The EU's data strategy has struggled to produce results. From the European Health Data Space to Catena-X in the automotive sector, industry hasn't found effective ways to share data that support both primary and secondary uses. The Apply AI Strategy assumes data sharing will work this time. What would have to be true for that assumption to hold?

The skills question presents another tension. The strategy emphasizes predicting required skills for each sector. But prediction assumes a stable target. If AI capabilities continue shifting rapidly, prediction becomes less useful than adaptation. Renda has argued the strategy should be "humans first" rather than "AI first," focusing on AI solutions that augment humans rather than replace them. This is a values disagreement embedded in what appears to be a technical planning exercise.

Timeline and Participation Structure

The Task Force runs from June 2026 to February 2027. Nine hybrid working sessions will run June through December 2026, followed by a drafting and member review phase. The Final Report launches in February 2027.

Deliverables include a Final Report with cross-sectoral analysis and recommendations, three sector policy briefs (Healthcare, Automotive, Government), and an established stakeholder network connecting European Digital Innovation Hubs (EDIHs), industry, and policymakers.

An information session for prospective participants was held on April 15, 2026. Registration remains open until May 31, 2026. Organizations may join one, two, or all three tracks, and may register up to two representatives who can rotate across sessions.

The Task Force operates under Chatham House Rules, which means participants can use information from discussions but cannot attribute statements to specific individuals or organizations. This structure encourages candor about implementation obstacles that might otherwise remain unspoken in public forums.

What This Reveals About EU AI Governance

The existence of this Task Force tells a story about the current state of EU AI policy. The Commission has produced a strategy. The strategy has identifiable gaps. A think tank is now organizing a structured process to address those gaps through multistakeholder engagement.

This is neither a failure nor a success. It's a recognition that policy documents require translation into operational reality, and that translation requires the people who will actually implement the policy to participate in defining what implementation means.

The Apply AI Strategy links to the broader EU legislative agenda, including the AI Act and the Cloud and AI Development Act. It anchors investment in a common, open-source pan-European frontier AI model. These connections matter because they determine whether the strategy operates as an isolated initiative or as part of a coherent policy architecture.

The potential links between the Apply AI Strategy and the European Competitiveness Fund remain unclear. Without coherence across Commission strategies, the risk is a repeat of 2021, when the industrial strategy and "transition pathways" for industry remained isolated from the AI agenda, the skills agenda, and the Green Deal.

The Question Worth Tracking

The CEPS Task Force represents an experiment in structured policy translation. The experiment's success depends on whether the process produces actionable recommendations that actually influence the Apply AI Alliance process and subsequent implementation.

The strongest version of the skeptical argument would note that think tank task forces have limited power to compel government action, that multistakeholder processes often produce consensus documents that satisfy no one, and that the fundamental resource constraints (no dedicated budget) remain unchanged regardless of how well the implementation challenges are analyzed.

The strongest version of the optimistic argument would note that implementation often fails not from lack of resources but from lack of clarity about what implementation actually requires, that structured engagement surfaces obstacles early enough to address them, and that the Task Force's sector-specific focus avoids the abstraction that makes many EU initiatives difficult to operationalize.

Both arguments have merit. The question is which one better describes what will happen between now and February 2027.