Back to stories
Policy

EU Awards €180M Sovereign Cloud Contract to Four European Providers in Bid to Reduce Hyperscaler Dependence

Michael Ouroumis2 min read
EU Awards €180M Sovereign Cloud Contract to Four European Providers in Bid to Reduce Hyperscaler Dependence

The European Commission on April 17, 2026 awarded its long-running €180 million sovereign cloud tender to four European provider groups, closing a procurement process launched in October 2025 and giving EU institutions a continent-owned alternative for sensitive workloads — a move that quietly reshapes where European AI data will sit.

The winning bidders are Luxembourg's Post Telecom, Germany's StackIT, French Iliad-owned Scaleway, and Belgium's Proximus. Post Telecom has brought in OVHcloud and CleverCloud, while Proximus heads a group that includes Mistral, Clarence and S3NS — the Thales-controlled sovereign cloud joint venture with Google Cloud.

A framework contract, not a single buy

The award is structured as a framework contract running for up to six years, allowing EU institutions, bodies, offices and agencies to purchase sovereign cloud services from any of the four winners as needs arise. It does not, by itself, force existing Commission workloads off Amazon Web Services, Microsoft Azure or Google Cloud — but it creates a pre-approved European procurement lane that sits alongside them.

For frontier AI workloads, the inclusion of Mistral AI inside the Proximus consortium is the most directly consequential detail. It gives EU bodies a route to procure European-hosted model capacity without going through a US hyperscaler, at a time when Paris has been pushing for Mistral to be treated as critical sovereign infrastructure.

The SEAL grading system

The more durable part of the announcement may be the evaluation framework itself. Each winner was scored against the Commission's Cloud Sovereignty Framework, which measures providers across eight objectives covering strategic, legal, operational and environmental considerations, supply chain transparency, technological openness, security and compliance with EU law.

Providers are graded on Sovereignty Effectiveness Assurance Levels, or SEAL, running from SEAL-0 — indicating a complete lack of sovereignty — to SEAL-4, which requires a full EU supply chain from chips to software. SEAL-4 is effectively unreachable today for most AI workloads, since the underlying accelerators remain dominated by Nvidia, but the framework gives Brussels a ratchet it can tighten over time.

Why it matters beyond Brussels

The contract is small in revenue terms; €180 million spread across four consortia over six years is a rounding error next to the tens of billions hyperscalers are reportedly committing to European AI infrastructure. The signal, however, is not about money. It is that the Commission is now willing to define sovereignty technically — chips, jurisdiction, supply chain — rather than rhetorically, and to translate that definition into procurement that member-state agencies can copy.

Expect national governments with sovereignty anxieties of their own to point at SEAL grades the next time they are asked to justify why a ministry workload is not simply going on Azure.

Learn AI for Free — FreeAcademy.ai

Take "AI Essentials: Understanding AI in 2026" — a free course with certificate to master the skills behind this story.

More in Policy

Anthropic's Amodei Meets Wiles and Bessent at White House in Pentagon Dispute Thaw
Policy

Anthropic's Amodei Meets Wiles and Bessent at White House in Pentagon Dispute Thaw

Anthropic CEO Dario Amodei met White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent on April 17, 2026, signaling a possible thaw in the company's Pentagon supply-chain-risk standoff.

19 hours ago2 min read
AI Hiring Enters the Regulated Era as EU Deadline Looms and Landmark Lawsuit Advances
Policy

AI Hiring Enters the Regulated Era as EU Deadline Looms and Landmark Lawsuit Advances

The EU AI Act's August 2026 high-risk enforcement deadline for hiring tools and the Mobley v. Workday class action signal a new era of AI recruitment regulation.

2 days ago2 min read
Linux Kernel Formally Allows AI-Generated Code — With Humans On The Hook
Policy

Linux Kernel Formally Allows AI-Generated Code — With Humans On The Hook

After months of fierce debate, Linus Torvalds and kernel maintainers agreed on a policy that permits AI-assisted contributions like Copilot while forcing human submitters to take full legal and technical responsibility for any bugs, security flaws, or licensing issues.

4 days ago2 min read