Denmark's First Homegrown Hyperscale for AI, Built on Green Energy and Data Sovereignty

Thylander plans Denmark's first Danish-owned hyperscale data center: 100 MW to start, expandable to 200, opening in 2026. Built for AI with dense racks, air-cooling and heat reuse.

Published on: Dec 05, 2025
Denmark's First Homegrown Hyperscale for AI, Built on Green Energy and Data Sovereignty

A new asset class for Danish real estate: AI-ready, sustainable data centers

Thylander is moving into data centers, and not in a small way. The Copenhagen firm plans to build Denmark's first Danish-owned and operated hyperscale facility: 100 MW at launch, with the option to scale to 200 MW.

Opening in 2026, the site is built for AI. Think high-density racks, GPU clusters, and colocation options from single racks to full halls. It's a direct response to what enterprises now want: local control, serious compute, and credible sustainability.

Why Denmark works for AI infrastructure

  • Abundant low-carbon electricity and a temperate climate suited for air-cooling
  • Low-latency fiber connectivity into Europe and the Nordics
  • One of Europe's most reliable power grids, with underground high-voltage transmission
  • Stable political, economic, and weather conditions

In short, the inputs are here. Now a local owner-operator wants to keep the value here too.

From local backing to global ambition

Thylander saw a gap: many organizations need data processed and stored locally, with higher standards for energy use and accountability. "Looking at data sovereignty and thinking about who actually owns data centers was the starting point for us to say that we want to do something Danish for Danish companies," says Anders Mathiesen, CEO of Thylander Data Centers, "but also for externals who think the Danish markets are valuable."

Mathiesen's view on timing is blunt. "The United States invests 15 times as much into high-risk and venture opportunities as we do in Europe. We need to start having the kind of risk tolerance and venture capital available to invest in these opportunities." Thylander believes it can stand up AI-ready capacity in about 12 months on prepared land with green energy access, local permits, and fiber in place-far faster than a new market entrant starting cold.

Sustainability built into the model

This isn't a bolt-on "green" label. The site is planned as a low-carbon, water-efficient facility. Air-cooling is viable thanks to Denmark's climate, reducing water draw versus traditional systems.

The project will repurpose the seawater cooling infrastructure from the Esbjerg power plant, cutting build time and upfront cost. Waste heat will be recycled into local district heating-a practical win for both the operator and the community.

  • Low-carbon electricity as the baseline
  • Air-cooling first, water use minimized
  • Existing seawater cooling assets repurposed
  • Heat reuse to support nearby homes and businesses

What tenants get: control, performance, and choice

Thylander's design targets AI-scale compute: GPU clusters, high-density racks, and predictable power and cooling. Customers can start with a rack and grow to a hall without moving cities or providers.

For many Danish organizations, the model enables hybrid architectures-some workloads stay on global cloud platforms, others run in a local private cloud for sovereignty or IP protection. As Mathiesen puts it, "They'll be able to say, 'I need to have some control over this.'"

Why this matters for real estate and construction teams

Data centers are no longer a niche asset. They sit at the intersection of real estate, energy, and digital infrastructure-with demand driven by AI adoption and latency-sensitive services. The value isn't just rent; it's long-term utility alignment, heat offtake, and community benefits.

  • Power strategy: confirmed megawatt availability, grid connections, and firm PPAs
  • Network: diverse fiber routes, latency objectives, and carrier-neutral options
  • Cooling: clear path to water-light or waterless operation, with contingency
  • Density: GPU-ready halls, rack kW targets, and modular scalability to 200 MW
  • Permitting and community: pre-consultation with municipalities, transparent environmental impact
  • Heat offtake: signed agreements with local utilities and practical tie-in timelines
  • Delivery risk: supply chain for switchgear, generators, transformers, and GPUs; realistic lead times
  • Reporting: auditable sustainability metrics and insurance aligned to resilience plans

If you're evaluating similar projects, treat energy and network as your core real estate. Everything else builds on that foundation.

Key milestones

The facility is planned to open in 2026. Initial capacity: 100 MW, expandable to 200 MW based on demand.

Leasing will span single racks to full halls, with configurations optimized for AI workloads. The aim: local control with global-grade performance.

Further context

For grid and energy background, see Energinet's overview of Denmark's electricity system here, and the Danish Energy Agency's resources on renewable energy here.

If your team is scoping AI-driven demand and needs fast baseline literacy on AI workloads and tooling, you can browse practical upskilling resources by job role here.

Bottom line

Thylander is betting that AI-ready, sustainable colocation is the next durable asset class in Denmark. The inputs are strong, the timetable is aggressive, and the value proposition is clear: sovereignty, scale, and smarter use of energy.

For real estate and construction leaders, this is a blueprint-pair ready-to-build land and grid access with compute-driven tenants, and design for sustainability from the first sketch.


Get Daily AI News

Your membership also unlocks:

700+ AI Courses
700+ Certifications
Personalized AI Learning Plan
6500+ AI Tools (no Ads)
Daily AI News by job industry (no Ads)
Advertisement
Stream Watch Guide