Machinepower

Machinepower Index / Dossiers / Denmark

Rank 29 of 50 · Tier 3

Rank 23–30 across the scoring rules tested. How this is tested

Denmark

Diffusion Advantage — For nations that cannot build the frontier but can use it faster than anyone: import the models and win on deployment, across the state and the economy, with public consent.

Score4.69
Potential6.10
Bottleneck cost−1.41
Weakest cell: Firm Power Capacity at 2.6 Binding layer: Watts Direction: Stagnating Edition: Q3 2026 · updated October 2026

How the Score, Potential and bottleneck cost are worked out: methodology.

The three layers

WattsMaterial power · holds the weakest cell
4.2 / 10

Energy, grids, data centres and silicon — the physical capacity to run models at scale.

Compute tracked by Epoch AI: Denmark controls 3,000 H100-equivalents, wherever the sites are. Hosted compute is shown, not scored.

WeightsIntelligence power
5.8 / 10

Models, alignment capability, standards influence and the talent that produces them.

WillPolitical power
8.3 / 10

The capacity to wire AI into the state and the economy while holding public consent.

Where the score comes from

LayerSub-dimensionBasisScore
WattsPlanning & PermittingAssessed4.6
Firm Power CapacityDataset2.6
Sovereign ComputeDataset4.7
Hardware & Silicon AccessDataset4.7
WeightsFrontier Model OutputDataset3.0
Alignment CapabilityAssessed6.2
Regulatory StandingDataset9.4
Talent DensityDataset4.6
WillPolicy CapacityDataset9.3
Public Sector AdoptionDataset8.5
Workforce TransitionDataset8.9
Public TrustAssessed6.3

Assessed marks a measurement we judged rather than took from a dataset: 3 of the 12 here.

Sources

Energinet2026
Planning & permitting (assessed)
4.6/10, assessed

About 60 GW of connection requests against a 7 GW peak led the grid operator to pause new large connections in March 2026.

View source →
Cullen International2025–2026
Alignment (assessed)
6.2/10, assessed

One of three EU states with an AI Act law in force; no safety statements or safety body.

View source →
Ember2025
Electricity Generation
34 TWh (2025)

Feeds firm power.

View source →
TOP500.orgJune 2026
High-Performance Compute Capacity
67 PFlop/s across 1 system

Largest: Gefion.

View source →
Stanford AI Index 2026April 2026
Top AI authors and inventors
3,900 (Zeki Data)

Talent is a log scale on the count.

View source →
Oxford InsightsMay 2026
Government AI Readiness Index 2025
#11 of 195 · 74.8/100

Feeds governance and public-sector adoption; policy capacity is listed separately.

View source →
University of Melbourne; KPMGJanuary 2025
Public trust (assessed)
51% trust AI at work

Ipsos does not cover it. Mapped to the Ipsos scale through the 28 countries both surveys cover.

View source →
IMF2023 data, published June 2024
AI Preparedness Index: human capital and labour market policies
0.177 of a possible 0.25 · 8.9/10 in the Index

Feeds workforce, rescaled across 173 economies.

View source →
Ember2023
Firm generating capacity
5 GW (2023)

Fossil, nuclear, hydro and other dispatchable plant; averaged with generation for firm power.

View source →
GitHub Innovation Graph; Oxford Insights2026
Developer base
0.5 million developers

Second source for talent, averaged with Stanford's count.

View source →
Oxford InsightsMay 2026
Policy capacity
92/100

Feeds policy capacity.

View source →

Bottlenecks

These hold Denmark below its Potential. Its two lowest scores cost it the most: firm power capacity at 2.6 (part of Watts) and frontier model output at 3.0 (part of Weights).

Watts (Material)Firm Power Capacity

Assessment

Denmark runs one of the world's most digital states and owns a national AI supercomputer, Gefion, paid for by the Novo Nordisk Foundation. Its grid has run out: new large connections have been paused since March 2026.

The art of the possible

Gefion gives Danish researchers compute most small states lack, and a digital government gives them somewhere to use it. How grid capacity is shared out decides whether more follows.

Strategic play — Diffusion Advantage

Spend scarce power on use, not hosting: give research and public services first claim on the grid, and stop competing for campuses it cannot carry.

Leverage — the nearest position. No chokepoint. The nearest key is health data: decades of national registries, now beside Novo Nordisk's compute, which drug-discovery models need.

Open Denmark in the live Index → — every source behind every number, and the comparison tool.