Denmark hands Nine A/S contract to build shared public-sector AI layer
Denmark's Agency for Digital Government has selected Nine A/S, a subsidiary of Nasdaq Copenhagen-listed Trifork Group, to build the shared infrastructure underpinning AI assistants for the Danish public sector. The initial deployment will introduce chat-based AI on two of the country's most-visited government portals: borger.dk, the national citizen services hub, and sundhed.dk, the national health information platform. The contract positions Denmark as an early mover in state-led, centrally architected AI rollout, at a moment when most European governments are still debating governance frameworks rather than shipping production systems.
The scope is deliberately expansive. The Agency has designed the project so that the shared assistant infrastructure can be extended to additional public-sector platforms beyond the initial two sites, meaning Nine is effectively building a reusable sovereign AI layer, not a one-off chatbot. That architectural choice carries significant strategic weight: rather than allowing individual ministries to procure fragmented AI tools from competing vendors, Copenhagen is consolidating the capability into a single, governmentally controlled stack.
Sovereign infrastructure, not just automation
"We are proud to have been selected for a project with the potential to shape how citizens and businesses interact with the public sector in Denmark," said Jacob Strange, CEO of Nine, adding that AI presents "a significant opportunity to make public services more accessible and reduce administrative workloads."
The contract sits within a broader Trifork Group track record in Danish state digitisation. Nine was selected in 2025 to develop the first phase of Denmark's Digital Identity Wallet, enabling citizen digital ID and proof-of-age credentials. A separate four-year framework agreement with the National Agency for IT and Learning covers seven applications spanning shared login and data sharing across education and employment. Trifork's health division has spent three decades contributing to national digital health infrastructure, including the forthcoming Shared Public Treatment Platform, which will support digital psychiatric and clinical programmes across the country. Together, these engagements suggest the group is becoming something close to a default integrator for Danish sovereign digital infrastructure.
The convergence read-across
The strategic significance for cross-sector observers extends well beyond Denmark's borders. European governments face a compound pressure: the EU AI Act requires public-sector AI deployments to meet transparency and human-oversight requirements, while fiscal constraints make bespoke per-agency AI procurement economically unsustainable. The Danish model, centralising the AI assistant layer into shared, governmentally governed infrastructure, offers one credible answer to both pressures simultaneously.
The architecture also has implications for the European data-sovereignty debate. Trifork's own product stack reflects the design philosophy: its Netic Contain platform is explicitly cloud-agnostic and open-source, Tiris Messenger is hosted in Denmark with data kept in Europe, and Corax AI is framed as a compliance-oriented deployment layer. For the class of investor tracking which European technology companies are structurally positioned to benefit from digital-sovereignty regulation rather than threatened by it, Trifork presents an unusual profile: a publicly listed, mid-cap technology integrator with recurring government contracts across health, identity and now AI infrastructure.
The broader capital landscape in govtech AI remains fragmented. US hyperscalers are bidding for European public-sector AI contracts, while a cluster of European sovereign-stack challengers, ranging from large integrators to specialist sovereign-cloud platforms, are competing on regulatory alignment rather than raw model capability. Denmark's decision to contract a domestic specialist for the infrastructure layer, rather than licensing a foundation model directly from a US provider, is a meaningful data point in that contest.