How AI Is Helping Government Companies in Denmark Cut Costs and Improve Efficiency

Denmark's government companies are at an inflection point: AI can cut costs and speed services, but pilots must mature into governed, data-ready programs to deliver those gains.

A recent EY survey finds only 26% of public-sector organisations have integrated AI even though 64% expect major cost savings and 63% expect better service delivery - obstacles include data privacy, weak infrastructure and missing strategies - and Danish researchers from Aarhus and the Danish Technological Institute stress that predictive systems must be co‑designed with stakeholders to work in practice.

Concrete pilots - from 72‑hour flood risk maps for emergency teams to automated content workflows - show quick wins, while targeted reskilling and applied courses such as the AI Essentials for Work bootcamp can help civil servants move from experimentation to measurable efficiency; read the EY survey on public-sector AI adoption and an example flood-forecasting use case in Denmark.

Bootcamp	Length	Early bird cost	Syllabus
AI Essentials for Work	15 Weeks	$3,582	AI Essentials for Work bootcamp syllabus

“The initial focus has paid off for pioneers who have developed a more effective digital and data foundation, and in some cases, data platforms that embrace cloud technologies. They have made faster progress in embedding data capabilities organisation-wide, rather than just in specific teams and departments. This helps maintain high standards of data quality and consistency, breaks down organisational silos and provides a unified approach to data governance and regulatory compliance.” - Permenthri Pillay, EY Global Government & Public Sector Digital Modernisation Leader

Denmark's national push to move pilots into scale is now backed by concrete budgets and timelines: the government's new digitalisation strategy for 2024–2027 lays out 25 initiatives and a headline investment of 800 million DKK to accelerate AI, automation and the green transition, while a targeted strategic AI budget of 55 million DKK (including a 17 million DKK regulatory sandbox to let authorities safely test systems) gives public organisations practical room to experiment and learn - details in the Responsible AI briefing.

At the EU level Denmark's Digital Decade roadmap signals broader fiscal muscle too, with a national package amounting to EUR 1.07 billion (EUR 832 million from public budgets) to shore up cloud, data and skills for the same period.

These national and EU strands sit alongside an earlier public investment fund - approximately EUR 27 million - set up to test and scale AI across municipalities and regions, which already financed signature pilots in healthcare, administration and climate adaptation; together they form a clear, time‑bound opportunity for government companies to move from one‑off demos to governed, interoperable services that save time and money while supporting Denmark's digital-by-default public sector.

Measure	Amount	Period / Notes
New digitalisation strategy initiatives	25 initiatives	2024–2027
Government investment (strategy)	800 million DKK	2024–2027
Strategic AI budget	55 million DKK	2024–2027
Regulatory Sandbox	17 million DKK	For testing AI with public authorities
Digital Decade roadmap (total)	EUR 1.07 billion	Includes EUR 832M public budgets
Public sector AI investment fund	≈ EUR 27 million	2019–2022; municipal & regional pilots

Denmark's regulatory landscape for AI is now firmly aligned with the EU's phased schedule: the EU AI Act was published in July 2024 and sets stepped obligations through 2027–2030 (see the EU AI Act implementation timeline and obligations), including Member States' duty to designate national competent authorities by 2 August 2025 and the August 2025 entry point for GPAI, governance and penalty rules.

Denmark moved early - Parliament adopted national implementation legislation on 8 May 2025 and designated the Agency for Digital Government, the Danish Data Protection Authority (Datatilsynet) and the Danish Court Administration as key supervisors - positioning Denmark as the first member state to complete national implementation and giving government companies clearer enforcement and procurement signals (Denmark early AI Act implementation legislation sets a national precedent).

For Danish government companies this means treating AI rollouts as regulated projects: GPAI disclosures, confidentiality rules, market‑surveillance duties and fines begin to bite from August 2025, with full compliance deadlines for existing GPAI providers by 2 August 2027 and national sandboxes due by August 2026, so procurement, legal and IT teams must map risk categories and build compliance into contracts and operations now.

Measure	Date / Note
Denmark national law adopted	8 May 2025 (enters into force 2 Aug 2025)
Member State authority designation deadline	2 August 2025
GPAI & governance rules apply	2 August 2025 (full compliance for pre‑existing GPAI by 2 Aug 2027)
AI regulatory sandboxes operational	By 2 August 2026

“There is no stop the clock. There is no grace period. There is no pause.” - Thomas Regnier, European Commission spokesperson

Denmark punches above its weight on AI adoption: in 2024 roughly one in four Danish companies - about 28% - reported using AI, nearly double the EU average of 13.5%, which makes the country a practical testbed for pilots that scale into live services (Invest in Denmark 2024 AI adoption report).

That strong headline masks important gaps and opportunities: the 2025 Digital Decade report highlights a persistent shortage of ICT specialists and a widening adoption gap between large firms (many of which already embed AI) and smaller companies that struggle with skills, costs and data quality (Denmark 2025 Digital Decade country report - EU Digital Strategy).

Practically, this means government companies can win big by funding targeted reskilling, creating shared data platforms and running sandboxes that let municipalities and SMEs trial GenAI and predictive tools safely - turning Denmark's early-adopter lead into broad, productivity‑boosting diffusion rather than a narrower innovation elite.

Measure	Value (2024)
Enterprises using AI (Denmark)	28%
EU average (enterprises using AI)	13.5%
Large Danish enterprises (250+ employees) using AI	63%

Børge, launched in February 2025, is a Denmark‑focused AI writing assistant built to help editors rewrite and optimise content on borger.dk and lifeindenmark.borger.dk by suggesting texts that follow borger.dk's internal writing guidelines; designed for use across approximately 40 Danish authorities covering some 1,200 pages, the tool raises efficiency and readability while framing GenAI as a collaborative aid rather than a replacement for human editors.

Early feedback shows it speeds routine content updates and supports on‑the‑job upskilling, helping teams keep citizen‑facing language clear and consistent - read the Decoding briefing on public‑sector AI for the case story and context Decoding briefing on public‑sector AI (Digital Hub Denmark).

Launch	Platforms	Coverage	Purpose
February 2025	borger.dk; lifeindenmark.borger.dk	Approximately 40 Danish authorities; 1,200 pages	Suggest texts aligned with internal writing guidelines to improve efficiency and readability

“a good, helping hand during a busy workday,”

Gefion has turned Denmark into a genuine AI infrastructure hub: built as an NVIDIA DGX SuperPOD with 1,528 NVIDIA H100 Tensor Core GPUs (191 DGX H100 units) and NVIDIA Quantum‑2 InfiniBand networking, it lets public organisations and researchers train large models and run petabyte‑scale simulations far faster than before - IO benchmarks even place its storage among the world's best so routine jobs that once took weeks can now finish in days or hours.

Hosted by Digital Realty in an AI‑ready, 100% renewable datacentre and assembled by Eviden, Gefion is run by the Danish Centre for AI Innovation (DCAI) after major backing from the Novo Nordisk Foundation and EIFO; read the DCAI announcement for technical and access details and the Novo Nordisk Foundation briefing on pilots and funding.

For Danish government companies, Gefion's combination of sovereign capacity, strong IO performance and targeted pilot slots creates a practical path to scale predictive services, speed up drug‑discovery and climate models, and lower long‑run cloud costs while keeping sensitive data local.

Specification	Detail
GPUs	1,528 NVIDIA H100 Tensor Core GPUs
DGX units	191 NVIDIA DGX H100 systems (DGX SuperPOD)
Networking	NVIDIA Quantum‑2 InfiniBand
TOP500 rank	21st
IO500 storage rank	7th (production storage), 4th (10‑node)
Hosting & energy	Digital Realty (100% renewable energy)
Funders	Novo Nordisk Foundation (~DKK 600m) & EIFO (DKK 100m)

“Gefion's ranking on the IO500 list is another testament to our commitment to pushing technological boundaries in advanced computing for AI and research,” says Nadia Carlsten, CEO of the Danish Centre for AI Innovation (DCAI), which runs and operates Gefion.

AI can meaningfully cut costs and speed services for Danish government companies by automating routine back‑office tasks, surfacing institutional knowledge and running predictive models that scale pilots into operations: EY's playbook shows GenAI streamlines grant applications, accelerates contract review and delivers automated financial commentary, while targeted assistants can improve receivables performance and reduce turnaround time (EY artificial intelligence use cases for government services).

Practical implementations - from semantic search that finds the right regulation in seconds to transport and scheduling optimisers - translate into real savings: Talbot West highlights research suggesting AI could free up to 30% of staff time and, in one example, cut invoice processing from five days to one–two days, turning a weekly bottleneck into near‑real‑time work (Talbot West analysis of AI for government efficiency).

In Denmark that matters on the ground too: 72‑hour flood risk maps used by emergency teams show how predictive models turn complex data into actionable operations, helping agencies coordinate evacuations and resources faster (72-hour flood risk forecasting and emergency mapping in Denmark).

The bottom line: well‑scoped GenAI pilots can convert repetitive hours into citizen‑facing capacity and predictive insight, often with measurable KPIs within months.

Measure / Example	Outcome	Source
Potential workforce time liberated	≈30%	Talbot West (Deloitte research)
Invoice processing improvement	From 5 days → 1–2 days	Talbot West
AR Collection Assistant impacts	30% better DSO; 40% productivity; 22% shorter receivable period	EY use cases

Denmark's skills response is no afterthought: the AI Kompetence Pagten is a bold public‑private drive that aims to upskill 1 million Danes by 2028, backed by a national AI strategy and a DKK 62.5 million framework to embed AI across schools, workplaces and public services; the pact brings together employers, unions, universities and government so upskilling is practical (Microsoft even offers a free online learning environment) and explicitly targets women, people with disabilities and those with shorter formal qualifications to widen access - see the national briefing on the pact and the coalition's aims at DSJC and the academic framing from SDU. Practical compliance help is already available too: the Agency for Digital Government's AI‑literacy guidance is explained in the Decoding briefing, giving authorities a roadmap to meet EU AI Act requirements and turn training into measurable service gains.

Measure	Value / Note
Upskilling target	1 million Danes (AI Kompetence Pagten, target by 2028)
Strategy funding	62.5 million DKK (framework through 2027)

“SDU co-founded Denmark's AI Skills Pact because we have a responsibility to enhance young people's foundational AI competencies and AI literacy.” - Peter Schneider‑Kamp, Professor, SDU

Governance in Denmark centres on “trust‑by‑design”: a citizen‑centric national strategy that pairs clear regulatory duties with practical tools so public organisations can adopt AI responsibly.

The EU AI Act and national guidance make AI‑literacy, risk assessments and documented decision‑chains part of everyday deployment, while privacy authorities push best practices for anonymisation, consent management and algorithmic transparency; see the Denmark citizen‑centric AI strategy – OECD AI Policy Observatory and the Decoding briefing on the EU AI Act and AI literacy – Digital Hub Denmark for concrete steps.

Practical implementation is supported by public‑private playbooks and nine‑step processes for responsible assistants - covering use‑case definition, data minimisation, logging and red‑teaming - so audits, human oversight and ongoing monitoring become as routine as security patching; Securiti's guide shows how to map models to data, classify risk and keep auditable logs (Responsible AI assistants guidance in Denmark - Securiti).

The upshot: treat transparency, documentation and worker training as non‑negotiable controls and AI can scale in the public sector without sacrificing trust.

“In Denmark, we've already shown how the public and private sectors can work together to responsibly implement large-scale digital solutions grounded in democratic values and complex legal frameworks. Digital IDs and digital post are just two examples. This approach has inspired not only Europe but the rest of the world, putting Denmark at the forefront of digital innovation. With this new collaboration, we're once again leading the way - this time in the responsible use of artificial intelligence,” says André Rogaczewski, CEO of Netcompany and Chairman of Danish Industry Digital.

A practical roadmap starts by treating every AI idea as a project: map use‑cases to the EU risk categories, run a gap analysis against the Decoding three compliance steps (risk assessment, an anchored compliance lead, and documented decision chains) and prioritise those with clear citizen value, not just novelty - see the Decoding briefing on the EU AI Act for the compliance checklist (Decoding briefing: AI in public sector digitalisation (EU AI Act checklist)).

Next, move from lab to live via Denmark's regulatory sandboxes and agency guides: prototype in a GDPR‑aware sandbox (the practiceguides overview notes completed sandbox projects such as Tryg Forsikring and Systematic) and use the Danish Agency for Digital Government's practical AI guides to lock down data governance and AI literacy before scaling (PracticeGuides: Artificial Intelligence 2025 - Denmark country guide, Danish Agency for Digital Government - Practical AI guides on data governance and AI literacy).

Parallel to pilots, invest in staff competence and a single coordinating compliance owner so that audits, human‑in‑the‑loop controls and documentation become routine; the payoff is tangible - sandboxed pilots that pass compliance checks can be operationalised across municipalities instead of stalling on procurement or privacy questions.

Phase	Key actions	Source
Assess & classify	Map use‑case to EU risk level; run internal audit	Decoding briefing: EU AI Act checklist (Digital Hub Denmark)
Pilot & validate	Use regulatory sandbox; conduct DPIA and red‑teaming	PracticeGuides: Artificial Intelligence 2025 - Denmark
Scale with governance	Appoint compliance lead; train staff; document decision chains	Danish Agency for Digital Government - Practical AI guides

Denmark stands at a clear crossroads: roughly one-quarter of authorities already use AI, and national funds and guidance mean pilots can scale quickly - but rights and trust cannot be an afterthought (see the Danish Institute for Human Rights summary on adoption and concerns).

Next steps for government companies are straightforward and practical: classify projects by EU risk level and run high‑risk pilots in regulated sandboxes, bake human‑rights audits into every procurement, and invest in staff capability so services remain citizen‑centred rather than surveillance‑led; the Decoding briefing offers a useful three‑step compliance checklist for teams planning rollouts.

Targeted reskilling is central - public programmes such as the AI Kompetence Pagten and applied courses like the Nucamp AI Essentials for Work bootcamp can turn experimental gains into everyday capacity.

At the same time, scrutiny from civil society - including Amnesty's report on automated welfare systems - is a reminder that technical scale must be matched by transparency, proportional data use and clear redress mechanisms to keep Denmark's digital public services both efficient and just; read the Amnesty analysis for the human‑rights perspective and the Decoding briefing for practical next steps.

Measure	Value / Note
Authorities using AI	≈25% (Danish Data Protection Agency / Danish Institute for Human Rights)
Public sector AI investment fund	≈EUR 27 million (for testing and scaling)
Upskilling target	AI Kompetence Pagten: 1 million Danes by 2028

“The way the Danish automated welfare system operates is eroding individual privacy and undermining human dignity.”