What Polar announced and why it lands now
Polar Data Centres has announced DRA02, a new 40MW IT-capacity facility in Drangedal, Norway, sited directly adjacent to its flagship DRA01 campus. The building offers more than 21,300 square metres of technical space, engineered specifically for hyperscale, cloud, AI, and HPC customers. The timing is deliberate. Demand for GPU-dense capacity has outrun the supply of power-rich, low-carbon locations that can actually host it, and the Nordics have emerged as one of the few regions with abundant renewable electricity and a cool climate. DRA02 is Polar's move to convert that geographic advantage into contracted capacity.
The announcement reads as a bet that AI infrastructure requirements have shifted permanently. Andy Hayes, CEO of Polar Data Centres, put it directly: "AI is fundamentally changing infrastructure requirements, and customers need partners that can deliver capacity purpose-built for tomorrow's workloads." The phrase that matters there is purpose-built. General-purpose colocation designed for 5 to 10 kW racks cannot absorb the thermal density of modern GPU clusters. DRA02 is designed from the ground up for that density, which is what makes it relevant to enterprises whose current facilities are hitting cooling and power ceilings.
Direct liquid cooling is the density enabler
The technical centerpiece of DRA02 is direct liquid cooling for high-density GPU deployments. As accelerator power draw climbs, air cooling reaches a physical limit where moving enough air becomes impractical and inefficient. Direct liquid cooling brings coolant to the components generating heat, which lets operators pack far more compute into each rack and each square metre. For customers, that translates into higher usable density and better energy efficiency per unit of compute, both of which matter when a training cluster's economics hinge on how much work each watt performs.
This is a meaningful signal for enterprises still running air-cooled estates. The next generation of AI hardware assumes liquid cooling as the baseline, and retrofitting an existing facility to support it is expensive and disruptive. A provider that has already built liquid cooling into the fabric of a new campus removes that burden from the customer. For technology leaders weighing whether to upgrade owned facilities or place high-density workloads with a specialist, DRA02 represents the specialist option, already provisioned for the thermal profile that next-generation GPUs require.
Hydro power turns sustainability targets into infrastructure
DRA02 runs on renewable Norwegian hydroelectric energy, which converts an abstract sustainability commitment into a concrete siting decision. Norway's grid is dominated by hydro, giving data centers there a carbon profile that is difficult to match in most other regions without buying offsets or unbundled certificates. For enterprises under pressure to report credible emissions figures for their AI workloads, locating compute on a low-carbon grid is one of the few levers that produces a real reduction rather than an accounting adjustment. The power source is part of the product Polar is selling.
The cool Nordic climate reinforces the advantage. Lower ambient temperatures reduce the energy needed for cooling, which improves overall efficiency and lowers operating cost. Combined with liquid cooling and hydro generation, the result is a facility whose sustainability story is structural rather than bolted on. For CIOs whose organizations have committed to emissions targets, the ability to place GPU capacity on a genuinely low-carbon grid changes what those commitments cost to honor. Sustainability and performance stop competing when the underlying power is already clean and abundant.
The financing signals durable commitment
DRA02 follows a EUR 800 million senior secured Nordic bond, roughly $932 million, that Polar raised in early June to refinance DRA01 debt and fund two additional facilities across 2026 and 2027. That capital structure matters to prospective customers because data center commitments are long-lived, and the financial durability of the provider is part of the risk a buyer takes on. A large secured bond aimed explicitly at building out capacity tells the market that Polar has both the intent and the funding to deliver a multi-facility roadmap rather than a single speculative building.
Hayes framed the expansion as continuity: "DRA02 builds on the strong foundations established with DRA01 and demonstrates our continued commitment to sustainable, high-performance data center infrastructure in Norway." For an enterprise evaluating where to place multi-year workloads, that continuity is a procurement consideration as much as a marketing line. A provider building a second facility next to a proven first one, backed by committed capital for more, offers a more credible growth path than a newcomer. Capacity that scales alongside your demand reduces the risk of outgrowing your provider mid-contract.
Nordic sovereignty and the regulatory tailwind
The strategic case for DRA02 extends beyond power and cooling into regulation. EU data sovereignty and sustainability mandates are pushing enterprises to keep workloads and data inside jurisdictions with strong governance and clean energy, and Norway sits well within that envelope. For European organizations, and for multinationals serving European customers, a low-carbon campus on the continent's edge addresses residency and reporting requirements at the same time. That combination is why Nordic hydro sites are shifting from a niche green option toward a primary region for serious GPU capacity.
This reframes a decision many technology leaders have been deferring. The default instinct has been to place AI workloads wherever GPUs were available fastest, often in regions with carbon-heavy grids and uncertain regulatory footing. Facilities like DRA02 make the compliant, low-carbon choice viable at scale rather than a compromise on capacity. As sovereignty rules tighten, the enterprises that placed workloads in defensible jurisdictions early will spend less time retrofitting for compliance later. Region selection is becoming a governance decision that a CIO owns.
What this means for your capacity strategy
For enterprises planning where AI workloads will live over the next several years, DRA02 is a data point worth acting on. It shows that a specialist provider can now offer the full package: purpose-built liquid cooling, genuinely low-carbon power, EU-aligned jurisdiction, and the financial backing to scale. Technology leaders who have been treating Nordic capacity as experimental should reassess it as a primary option, particularly for training and HPC workloads where power cost and carbon profile dominate the economics. The gap between the sustainable choice and the performant choice has narrowed to the point of disappearing.
The broader implication is that region and cooling architecture now belong in the same conversation as chip selection. A high-density GPU deployment placed in an air-cooled, carbon-heavy facility will underperform on cost, efficiency, and compliance against the same hardware in a hydro-powered, liquid-cooled campus. For the reader's roadmap, the move is to evaluate providers on their power source, cooling design, and jurisdiction alongside raw capacity and price. DRA02 raises the bar for what a credible AI hosting partner looks like, and the standard it sets is one worth measuring incumbents against.
