Cloud and AI Development Act
EU law setting sovereignty tiers for cloud and AI infrastructure serving sensitive public services.
On 23 July, Atos launched a sovereign cloud platform for government, defence and healthcare buyers as the Cloud and AI Development Act prepared to take effect on 4 August.
Last refreshed: 22 September 2026 · Appears in 2 active topics
Can CAIDA survive a trade war it was never designed to fight?
Timeline for Cloud and AI Development Act
Mentioned in: Bruegel puts EU AI compute at 5%
European Tech SovereigntyMentioned in: Atos launches EU-engineered sovereign cloud platform
European Tech SovereigntyPoland proposes model-weights test for state contracts
European Tech SovereigntyMistral ships self-hosted OCR 4 to Europe
European Tech SovereigntyMentioned in: Anthropic AI ban enters its second week
AI: Jobs, Power & MoneyBackground
The Cloud and AI Development Act is an EU law that sets sovereignty-tier requirements for cloud and AI infrastructure used in sensitive public services. It directs public procurement towards providers able to meet European control requirements for specified workloads. the Commission adopted it on 3 June after earlier delays.
Its core purpose is to reduce legal and operational dependence on cloud infrastructure subject to non-EU control. The Act focuses on public-sector tenders involving health, finance, judicial and energy services, rather than private-sector procurement. Its strictest tier applies to the most sensitive government data, which the Commission estimates accounts for about 1% of public services.
The Act forms part of the EU's broader digital-sovereignty policy, alongside measures supporting open-source software and European infrastructure. It can influence where sensitive public workloads run, but it does not itself create domestic AI chips or regulate the whole cloud market.
The law is shaping supplier demand
On 23 July, Atos launched a sovereign cloud platform built in the EU on open-source components for government, defence and healthcare buyers. The launch made it the fifth named provider in a market directly shaped by the Cloud and AI Development Act's procurement rules.
From 4 August, the Act gives public bodies a legal framework for choosing infrastructure under European control for sensitive work. Its phased duties begin in February 2028, so providers have time to reshape contracts before the strictest requirements arrive. The law therefore creates a defined public-sector customer base, even though it does not regulate the wider private cloud market.
Delays narrowed the law's ambition
On 27 May, the Cloud and AI Development Act missed its third planned adoption date after German resistance inside the European Commission. The delay left France without the proposed sovereignty package when it chaired the G7 Digital Ministerial two days later.
the Commission adopted the Act on 3 June, but its scope was narrowed to sensitive public-sector procurement. That compromise made adoption possible after repeated delays, while leaving private-sector cloud contracts outside its direct reach. For the Act, the political history explains both its eventual legal form and the limits of the market it can change.
Why has the EU Cloud and AI Development Act been delayed three times?
How does CAIDA relate to the USTR Section 301 investigation into EU digital rules?
Does CADA restrict private-sector cloud contracts?
When does CADA enter into force?
What is the Cloud and AI Development Act (CADA)?
What is the difference between CAIDA and the existing EU sovereign cloud procurement framework?
How does CADA relate to the Draghi report?
