Summary
- DRA02 is planned for 40MW of IT capacity and more than 21,300 square metres of technical space.
- The building will sit beside DRA01 and incorporate direct liquid cooling for high-density GPU systems.
- Polar has not disclosed the construction programme, customer commitments, or grid-delivery sequence.
Polar Data Centers has announced a 40MW addition to its Drangedal campus in Norway, where a second facility will be built beside its existing DRA01 data centre.
DRA02 is planned to provide more than 21,300 square metres of technical space, with direct liquid cooling incorporated into the building for artificial intelligence, cloud, and high-performance computing systems. Polar describes the 40MW figure as IT capacity rather than the site’s total utility demand.
The development expands a campus already marketed around Norwegian hydroelectric generation, although Polar has not published a construction start, service date, customer commitment, or detailed power-delivery plan for the new building.
Liquid cooling moves into the base build
Designing DRA02 around direct liquid cooling avoids some of the constraints created when high-density hardware is introduced into halls built mainly for air-cooled racks. Facility pipework, coolant distribution units, controls, drainage, leak detection, and maintenance access can be coordinated with the structure rather than threaded through an operating room after handover.
Polar has not identified the precise cooling system. Direct-to-chip cold plates, rear-door heat exchangers, immersion systems, and hybrid configurations impose different requirements on water temperatures, flow rates, coolant chemistry, pumping, and the amount of heat that remains in the room air.
Even where processors are cooled directly, memory, networking, power electronics, and ancillary equipment can leave a substantial residual air load. DRA02 will therefore still require an air-management strategy and heat-rejection plant sized for the combination of liquid and air cooling that its customers eventually deploy.
At 40MW of IT capacity, the building’s total electrical demand will be higher once cooling, conversion losses, lighting, security, and other supporting systems are included. The gap between the two figures affects the utility connection, substation size, energy procurement, and the volume of backup generation or stored energy required by the resilience design.
Customer hardware will also determine how quickly the facility reaches its rated capacity. A building configured for dense GPU systems may fill with fewer racks than a conventional colocation hall, concentrating power and cooling into a smaller area while increasing the consequence of losing a CDU, pump, busway section, or network fabric.
Hydroelectric generation still needs a local connection
Norway’s generation mix gives developers access to a power system dominated by hydropower, but national generation statistics do not guarantee capacity at a particular industrial site. DRA02 will still rely on local network headroom, substation infrastructure, connection agreements, and any reinforcement required before the additional load can be energised.
Data centres are competing for that capacity with electrified transport, manufacturing, hydrogen projects, and other industrial users. Utilities and public authorities are consequently examining not only the size of proposed loads but their delivery schedules, employment effects, flexibility, and ability to use energy efficiently.
Polar has not stated whether DRA02 will use spare capacity already secured for the campus or require additional network work. Shared electrical infrastructure can reduce construction time when transformers and switchgear have been sized for later phases, although every new block adds protection, control, and fault-coordination requirements.
The company has also avoided publishing a PUE target, water-use metric, or heat-recovery plan. Norway’s climate can reduce mechanical cooling demand, while warmer liquid-cooling loops may improve the prospects for free cooling and recovered heat, but annual performance will depend on server utilisation, operating temperatures, plant staging, and the final cooling architecture.
DRA02 gives Polar a defined second block at Drangedal, but the path from announcement to service will run through grid delivery, equipment procurement, construction, and integrated testing. Further disclosures on those elements will establish whether the 40MW can arrive on the same timetable as the computing systems it is intended to host.
