The CEPS Task Force on the Apply AI Strategy: What Kind of Disagreement Is This, Really?

The Strategy That Arrived at the Right Time – With the Wrong Budget

The European Commission's Apply AI Strategy represents a genuine shift in Brussels' approach to artificial intelligence. After years dominated by the AI Act's regulatory architecture, the Commission has pivoted toward deployment and scale. The strategy targets eleven sectors – from healthcare and automotive to public administration – with the explicit goal of making Europe an "AI Continent."

The framing is ambitious. The strategy promotes an "AI first" policy orientation and a "buy European" approach, particularly for public sector procurement. It establishes the Apply AI Alliance as a coordination forum and creates an AI Observatory to track sectoral impact.

Here's where the disagreement begins – and where it deserves careful disentangling.

Three Disagreements Masquerading as One

When critics say the Apply AI Strategy is insufficient, they might mean several different things:

First, a resources disagreement. As CEPS researchers have noted, the strategy "comes with a lot to do but no budget." This is a factual claim about implementation capacity. Either the European Competitiveness Fund will be linked to Apply AI priorities, or it won't. Either AI Factories and Gigafactories will receive sufficient investment, or they won't. This disagreement can be resolved with numbers.

Second, an infrastructure disagreement. The strategy assumes that compute infrastructure – data centres, GPU clusters – will serve sectoral needs. But as the same CEPS analysis observes, "simply relying on large-scale data centres and powerful GPUs isn't going to be enough. Every sector will require a different solution." Healthcare AI may need edge computing for privacy-sensitive diagnostics. Automotive AI may need real-time inference capabilities. Government AI may need sovereign cloud architectures. The question isn't whether Europe needs infrastructure – it's whether the infrastructure being built matches the use cases being prioritised.

Third, a data governance disagreement. The EU's data strategy has, to put it diplomatically, struggled to produce results. From the European Health Data Space to Catena-X in automotive, industry hasn't found effective mechanisms for data sharing that support both primary and secondary uses. The Apply AI Strategy assumes these mechanisms will materialise. Sceptics ask: why would it be different this time?

These are three distinct problems requiring three distinct solutions. Conflating them produces heat without light.

What the CEPS Task Force Actually Does

The CEPS Task Force on the Apply AI Strategy is designed to disaggregate exactly these questions. Running from June 2026 to February 2027, it structures analysis across three sector tracks: Healthcare & Pharma, Automotive & Mobility, and Government & Public Sector.

Each track applies a consistent analytical framework across five issue areas: key use cases, infrastructure requirements, data governance, flagship applications, and sovereignty and implementation roadmaps. The methodology matters here. By using parallel structure across sectors, the Task Force makes positions comparable – revealing where challenges are genuinely sector-specific and where they reflect systemic EU-level gaps.

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

An online information session for prospective participants takes place tomorrow, 15 April 2026, at 15:00 CET.

The Sovereignty Question Beneath the Surface

The Apply AI Strategy's "buy European" framing has received mixed reception. The strongest version of the argument runs something like this: European public sectors should not build critical infrastructure on foundation models controlled by non-European entities whose governance, data practices, and strategic priorities may diverge from European values and interests.

The strongest counter-argument: procurement restrictions that prioritise European origin over capability may leave European institutions with inferior tools, undermining the very competitiveness the strategy aims to enhance.

What would have to be true for both positions to be right? Perhaps this: that European AI sovereignty is worth pursuing where European alternatives are genuinely competitive, while maintaining openness where they are not yet – combined with strategic investment to close capability gaps in priority domains.

The Task Force's sector-specific approach may help here. Sovereignty requirements differ dramatically between, say, defence applications and creative industries. A blanket "buy European" policy may be both too restrictive in some domains and insufficiently protective in others.

The Implementation Burden No One Wants to Name

The CEPS Ideas Lab session on the Apply AI Strategy in March 2026 surfaced a tension that deserves more attention: the strategy's ecosystem approach is intellectually coherent but operationally demanding.

Connecting use cases with infrastructure, data, and skills requirements across eleven sectors requires coordination mechanisms that don't yet exist at scale. The Apply AI Alliance and AI Observatory are governance innovations, but governance innovations require institutional capacity to function.

The risk – and this is worth naming explicitly – is that the Apply AI Strategy becomes another entry in the EU's growing collection of ambitious frameworks that remain isolated from each other. The 2021 industrial strategy and its "transition pathways" for industry remained disconnected from the AI agenda, the skills agenda, and the Green Deal. The Apply AI Strategy could follow the same trajectory unless deliberate effort connects it to the European Competitiveness Fund, the Data Union Strategy, and sectoral regulatory frameworks.

The Question Worth Asking

The CEPS Task Force represents one model for translating strategy into practice: structured stakeholder engagement, sector-specific analysis, and deliverables designed to inform implementation rather than simply critique ambition.

Whether this model scales – whether it can produce the kind of cross-sectoral learning and coordination the Apply AI Strategy requires – remains genuinely uncertain. The Task Force operates under Chatham House Rules, which enables candid discussion but limits public accountability. Member contributions are welcomed but do not determine conclusions, which preserves analytical independence but may limit buy-in from participants whose positions aren't reflected.

The question worth asking isn't whether the Apply AI Strategy is good or bad. It's whether the EU has the institutional capacity to implement ecosystem-level strategies that require coordination across sectors, member states, and policy domains – without dedicated budget, without proven data-sharing mechanisms, and without clear infrastructure priorities.

The CEPS Task Force is, in effect, a test case. Registration closes 31 May 2026.