Nscale’s 30 MW Glomfjord Data Center Reported Six Months Behind Schedule
DataNews Editorial Team
Nscale’s data center under construction in Glomfjord, Norway, is reportedly running six months behind schedule and providing 6.4 MW of computing power against a stated 30 MW project capacity. The reported issues include cooling-system problems, an electrical fire linked to a high-voltage cable, and an expanded contractor workforce requiring offshore accommodation.
- The Glomfjord project remains under construction and is stated as a 30 MW total-project data center.
- The project is reported to be six months behind schedule.
- Reported computing power was 6.4 MW at the time of publication; the source does not fully define this metric.
- Cooling, high-voltage electrical infrastructure and labour accommodation were reported as delivery constraints.
- The use of a floating accommodation barge highlights workforce-housing limitations at remote data center sites.
Nscale is building a 30 MW data center in Glomfjord, Norway, but the project is reported to be six months behind schedule. At the time of publication, the facility was said to be providing 6.4 MW of computing power, compared with the project’s stated 30 MW total capacity.
The reported shortfall was attributed to a combination of technical and construction-delivery issues. These include problems with cooling systems and an electrical fire associated with a high-voltage cable that had reportedly been left outside for six months before connection. Nscale has also reportedly brought in hundreds of additional contractors in an effort to recover the schedule.
Workforce logistics have become a material construction constraint. With limited local accommodation, Nscale deployed the Accommodation Barge Kalmar to house workers. The vessel has 220 cabins, according to Chevalier Floatels. Its website also lists a separate floating-refugee-centre configuration capable of accommodating up to 680 people; that figure should not be treated as the capacity of the current contractor accommodation arrangement.
For data center operators, tenants and infrastructure suppliers, the project underscores that stated capacity for remote AI-oriented campuses is not equivalent to deliverable capacity. Site readiness depends on the execution and reliability of critical cooling and electrical systems, as well as the ability to mobilise and accommodate construction labour. These constraints could increase schedule and delivery risk across projects in similarly supply-constrained locations.
The case illustrates the execution risk in rapidly deploying AI-oriented data center infrastructure in remote locations. Announced capacity does not necessarily translate into available IT capacity when construction quality, cooling and power systems, workforce availability, and contractor housing constrain delivery.
- — / October 23, 2026 · EnglishSource →
This article was prepared by the DataNews editorial team based on the sources listed above.