Velatir's EUR5M seed: why Europe's AI sovereign play is the control layer

€5 million does not buy a model. It buys the layer that decides what every model inside a company is allowed to do. That is the wager behind Velatir, a Danish startup that closed a €5 million seed round this week to build what it calls the operating system for enterprise AI.

Velatir is not alone in spotting the opening. It is one of a cluster of European vendors racing to turn AI oversight into a product before the hyperscalers close the gap. What makes the round worth a private investor's attention is not the size of the cheque but the shape of the bet.

Why a control layer, not another model

Most enterprise AI spending still chases capability: bigger models, faster inference, cheaper tokens. Velatir bets the durable value sits one layer up. Its platform plugs horizontally across the tools a company already runs, from Slack and Microsoft Teams to email and the AI agents employees wire up on their own. It logs every interaction, flags sensitive data before it leaves, and routes high-risk actions to a human reviewer. The pitch is that the organisation, not the individual, should hold the controls.

The numbers behind the bet are unusual for a seed-stage company. According to EU-Startups and Tech.eu, Velatir reached 80 customers across six European countries within seven months of launch and is approaching $1 million in annual recurring revenue. The round was co-led by Spintop Ventures and Ugly Duckling Ventures, with Norrsken Evolve returning and angels including Jan Oberhauser, the founder of workflow automation company n8n, and Thomas Visti, a former chief executive of Mobile Industrial Robots. The Danish Export and Investment Fund added a match loan.

The regulatory tailwind

The European Union Artificial Intelligence Act, the bloc's first binding framework for AI, is the quiet engine of this category. As the Act's obligations phase in, organisations must be able to show what their systems actually do, not only what their policy says. Velatir's co-founder Christian Møller previously ran AI Act and DORA (the Digital Operational Resilience Act) implementation at a large European financial group, and that regulatory fluency is the company's sharpest sales asset. A governance product sells best when the buyer is already legally exposed.

We basically told people not to paste anything sensitive into AI tools and hoped for the best. Now sensitive data actually gets caught before it leaves, so we stopped saying no to everything and started letting teams use what they need.— Thomas Overgaard, Head of Compliance, a Velatir customer

That shift, from hoping to enforcing, is the product's whole reason to exist. The harder the regulation bites, the larger the buyer pool that cannot keep running on hope.

Sovereignty as the moat

Velatir's second bet is geopolitical. The platform runs entirely on European-owned and hosted infrastructure, a deliberate choice to avoid US hyperscalers. For public-sector buyers and regulated industries, that distinction is the difference between a sale and a procurement rejection. The June 2026 EU Tech Sovereignty Package, which proposes a Cloud and AI Development Act, signals where public money is heading: toward infrastructure that answers to European law rather than a US terms-of-service agreement.

The team behind the thesis

The founding team is the part a private investor should weigh hardest. Chief executive Michael Sørensen spent 14 years in Danish military intelligence, represented Denmark on NATO's Electronic Warfare Working Group, and later led security audits across Meta's global data centre portfolio. That background maps directly onto a product whose promise is trust and control. The rest of the founding team pairs regulatory implementation experience with product engineering, including a chief product officer who rebuilt the admin dashboard for Usercentrics' Cookiebot consent tool.

Can a standalone control layer survive the hyperscaler bundling?

Development scenarios