On 3 June 2026, the European Commission adopted its proposal for a Cloud and AI Development Act — in full, a "Regulation establishing a framework of measures for strengthening Europe's cloud and AI ecosystem." It travels under the shorthand CADA and the reference COM(2026) 502.

This is a plain-language explainer of what the Act is, what its scope actually covers, how it is meant to close Europe's cloud and AI gap, and where it sits on the roadmap toward becoming law. The goal is to describe the proposal and its context accurately — not to argue for or against it. One thing to keep in mind throughout: CADA is a proposal. It has been tabled by the Commission, but it still has to pass through the European Parliament and the Council before any of it becomes binding.

Primary sources: the Commission's official announcement and the full proposal, annexes and impact assessment are on the EU digital-strategy library page. Timeline and legislative context are tracked on the independent eu-cloud-ai-act.com resource (Cyber Risk GmbH). Where this article states what CADA "does," it refers to Article 1 of the proposal directly.

01 Why this Act exists

CADA did not appear out of nowhere. Its intellectual starting point is the Draghi report on European competitiveness (September 2024), which argued that Europe is falling behind the United States and China on innovation, and that computing capacity is a critical component of a modern data economy. Draghi explicitly proposed an "EU Cloud and AI Development Act" to strengthen Europe's high-performance computing, AI and related infrastructure, and to harmonise cloud architecture and procurement across Member States.

From there it became a headline commitment. Executive Vice-President Henna Virkkunen — responsible for Tech Sovereignty, Security and Democracy — named the Act in her confirmation hearing, framing it around a simple promise: "The Act will allow even the smallest businesses to access advanced AI services." It was listed in the 2025 Competitiveness Compass and folded into the AI Continent Action Plan. CADA is designed to sit alongside, not replace, the rest of Europe's digital rulebook.

How CADA relates to the other rules: the EU AI Act governs how AI may be developed and used; CADA is about the infrastructure underneath it — the data centres, cloud capacity and sovereign compute that AI runs on. It also complements the Chips Act 2.0 and the EU's open-source strategy. Think of the AI Act as the rules of the road, and CADA as building the roads.

The problem it targets is concrete. Training and running modern AI models demands enormous computational resources, and by the Commission's own assessment the EU currently lags behind the US and China on available data-centre capacity. Left unaddressed, that capacity gap becomes a dependency: European businesses, researchers and public bodies increasingly rely on a small number of non-EU cloud providers for the compute their AI needs.

02 The scope, defined

Article 1 of the proposal ("Subject matter") sets out the scope precisely. CADA establishes a framework for strengthening the Union's cloud and AI ecosystem through five concrete measures:

Article 1 measureWhat it means in practice
(a) Leadership InitiativesEstablishes a Cloud Leadership Initiative and an AI Leadership Initiative (together, the "Cloud and AI Leadership Initiatives") to coordinate investment and effort.
(b) Data-centre deploymentSets a framework for the accelerated deployment of data centres across the Union — the physical capacity AI and cloud run on.
(c) Sovereign offerEnables the availability of a sovereign cloud and AI offer to safeguard the Union's public order — capacity that stays under European control for sensitive uses.
(d) Reduce dependenciesAims at reducing dependencies on critical technologies, i.e. the reliance on a handful of non-EU providers and supply chains.
(e) Public-sector adoptionFosters the adoption of cloud computing across the public sector, so governments help build demand for European capacity.

The proposal then states two general objectives — deliberately kept separate from each other:

  1. Ensure the conditions necessary for the competitiveness and innovation capacity of the Union's cloud and AI ecosystem.
  2. Improve the functioning of the single market by laying down a uniform Union legal framework that increases the EU's resilience and strategic autonomy in cloud and AI technologies.

The Commission's own communication groups the same substance into three pillars, which are the easiest way to remember what CADA is trying to do:

1. Research, development and innovation

Support the effort to build and deploy the next generation of advanced, sustainable cloud and AI technologies — so Europe is developing capability, not only consuming it.

2. Capacity

Accelerate the conditions for rolling out data centres across the EU, with particular attention to facilities that strengthen essential public functions. This is the physical build-out — the megawatts and square metres.

3. Autonomy

Introduce a common, EU-wide cloud and AI sovereignty assessment framework, paired with an adoption mechanism for the public sector. This is the part that lets a buyer say, in a consistent way across all Member States, "this offer is sovereign enough for this workload."

Scope in one line: CADA is an industrial and infrastructure Act. Its centre of gravity is capacity, competitiveness and sovereignty — not consumer protection or compliance obligations. That is a genuine shift in tone from most recent EU digital laws, which were built around rights, transparency and data governance.

03 How it aims to meet Europe's cloud and AI needs

The value of CADA is easiest to see through the problems it is trying to solve. The Commission's call for evidence spelled these out, and they map cleanly onto the measures above.

The capacity gap

AI is compute-hungry. Training needs large, centralised capacity; fine-tuning and inference lean on more distributed cloud and edge resources. Europe simply does not have enough data-centre capacity today to serve that demand at the scale it is growing. CADA's data-centre and RDI measures are aimed squarely at closing this gap — and at doing so in a way that prioritises highly sustainable solutions.

The build-out barriers

Data centres are hard to build in Europe, and CADA names the reasons: difficult access to energy, water and land; slow and fragmented permitting processes that differ between Member States; capital-intensive construction that raises barriers for new entrants; and rising energy and water consumption made worse by the heavy concentration of existing data centres in the north-west of the EU. High energy prices sit on top of all of it. A single Union framework is meant to make the permitting and investment path more predictable.

The missing sovereign offer

There is currently no competitive, EU-based cloud offer at sufficient scale to serve the most security-sensitive use cases — the ones found across government, healthcare, finance, energy and defence. CADA's sovereign-offer measure and the sovereignty assessment framework are designed to fill exactly that gap, giving public bodies and critical sectors a European-controlled option they can actually procure.

Access for the smallest players

The recurring political promise around CADA is that it should widen access, not narrow it — letting even small businesses and researchers reach advanced AI capacity that today is effectively gated behind the largest providers. This is the demand side of the same coin: build the capacity, then make it reachable.

The honest tension: "sovereignty" and "reducing dependencies" inevitably touch the large non-EU providers — Amazon, Microsoft and Google — that dominate the European cloud market today. CADA does not target any specific company or country, but its objectives are read closely in Washington, and a period of negotiation over definitions, eligibility and market access is widely expected. For most European organisations the practical takeaway is simpler: a sovereign, EU-controlled option is being built, and it will eventually be something you can choose.

04 The roadmap

This is the part worth bookmarking. CADA has moved through a fairly clear sequence of milestones, and there is an agreed target for when it should be finalised. Here is the roadmap from origin to target:

DateMilestone
17 Sep 2024 The Draghi report on EU competitiveness proposes an "EU Cloud and AI Development Act" to strengthen high-performance computing, AI and infrastructure.origin / political idea
Dec 2024 – Feb 2025 Henna Virkkunen takes office as Executive Vice-President for Tech Sovereignty, Security and Democracy and commits to propose the Act.mandate
9 Apr – 4 Jun 2025 Call for evidence and impact assessment: the Commission gathers stakeholder input on the problem and the options.consultation
20 Oct 2025 Commission Work Programme 2026 lists CADA as a legislative initiative (legal basis Article 114 TFEU), with an indicative timing of Q1 2026.planned
23 Apr 2026 "One Europe, One Market" Joint Roadmap: Parliament, Council and Commission commit to key initiatives, with CADA targeted for Q4 2027.political commitment
3 Jun 2026 The Commission adopts the proposal — COM(2026) 502, with annexes and impact assessment. This is where we are now.proposal tabled
Q4 2027 (target) Indicative target window for agreement between the co-legislators (Parliament and Council).target — not entry into force
After Q4 2027 Formal legal steps — publication in the Official Journal and entry into force — follow, and may extend beyond the target date.to be confirmed
Read the timeline carefully. "Q4 2027" comes from the Commission's roadmap and is indicative, not a formal legal schedule. It is a target for political agreement between the Parliament and the Council — it does not mean publication in the Official Journal or legal entry into force on that date. Treat these dates as strategic planning guidance, not a compliance deadline.

05 What it means for organisations

Because CADA is a proposal and mostly an infrastructure-and-investment instrument, it does not drop new day-one compliance obligations on the average business the way the GDPR or the AI Act did. Its impact is more strategic — and it lands differently depending on who you are.

It is also useful to read CADA next to what is happening on the other side of the Atlantic. Where the recent US AI Executive Order 14409 leans toward light-touch, market-driven integration, CADA leans toward sovereign control and strategic autonomy. Same technology, two different centres of gravity — which is precisely why multinationals will feel pulled in both directions.


The bottom line

CADA is Europe's attempt to build the ground its AI ambitions will stand on: more data-centre capacity, a credible sovereign cloud and AI offer, fewer dependencies, and a single framework to make it all buildable and procurable. Its scope is set out cleanly in Article 1 — five measures, two objectives, three pillars — and its roadmap runs from the Draghi report to a Q4 2027 target for agreement. For now it is a proposal moving through the legislative machine, so the right posture is to understand it, track the milestones, and plan for a European sovereign option that is being built rather than one you must comply with tomorrow.

Jasper Bernaers  ·  Cloud & AI Strategist. This is a factual explainer; the underlying documents are linked below.



More articles

US AI Executive Order 14409: what it means for Europe The light-touch, security-first counterpoint to Europe's sovereignty push. The EU's 2025 digital omnibus package How Europe's simplification drive is reshaping digital and AI regulation. How to Build a Zero Trust Modern Workplace with Microsoft 365 Sovereignty starts with control — identity-first architecture in practice.