Nscale Glomfjord Site Reportedly Six Months Late

Data Center Dynamics reported on 16 October 2026 that Nscale’s data centre project in Glomfjord, Norway, is reportedly running six months behind schedule, according to a Bloomberg report citing people close to the project and public records. Bloomberg’s original piece is paywalled and was not read directly for this post; the facts below follow Bloomberg via DCD.

DCD says it contacted Nscale for comment and verification of Bloomberg’s claims. Nscale has not commented in DCD’s reporting. Treat the delay, the incident list and the floating-hotel detail as reported, not company-confirmed.

What Bloomberg said, via DCD

According to DCD’s account of Bloomberg:

The Glomfjord project has faced hurdles including construction delays, local regulations, leaks in the cooling system, and a transformer fire.
Currently, 6.4MW of the planned 30MW is operational at the site.
To help speed up construction, Bloomberg reports that the company has towed in the Kalmar floating hotel to house construction workers nearby.

Those are the operational claims DCD attributes to Bloomberg. This post does not add Bloomberg wording beyond that DCD summary.

IPO context — from DCD, not a dated launch

DCD notes that Nscale is preparing to launch an initial public offering, with reports suggesting a valuation of up to $35 billion. DCD does not give an IPO date in the piece summarised here, and this post does not invent one.

DCD says Nscale published an IPO prospectus in September, which noted:

Five active data centres offering a combined 55MW
12 sites in various stages of development

Among risk factors listed in that prospectus, DCD highlights “delays in construction”, “construction site accidents”, and a long list of permit and entitlement risks (zoning, work permits, demolition, design, operational and close-out permits, plus delays from public agencies and utilities).

Investors, DCD writes, will be weighing those risks against contracted demand. Per DCD, the company cites $103bn in total contracted value, largely thanks to contracts with Anthropic and Microsoft, for which it is building out capacity.

Revenue and losses in the prospectus figures DCD cites

DCD reports that Nscale recorded $140.6 million revenue for the six months to the end of June 2026, which the company claims is up 1,252 percent year on year. Losses, DCD says, rose to $1.02bn, compared with $368.9m in the same period of 2025.

Those growth and loss figures are company claims / prospectus figures as relayed by DCD, not independent audits summarised here. The combination — fast revenue growth, large losses, and a reported construction slip at a flagship Nordic site — is exactly the tension IPO investors price when neocloud builders scale GPU capacity ahead of utilisation.

Why a six-month slip matters

Glomfjord’s planned 30MW, with only 6.4MW said to be live, is a concrete example of the gap between contracted AI cloud ambition and energised halls. Cooling leaks and a transformer fire are the kinds of site incidents the prospectus risk factors already flag; a reported six-month delay turns those boilerplate lines into a live operational story.

Nordic sites are attractive for power and climate, but local regulation and construction logistics still bind delivery. Towing in a floating hotel for workers, if Bloomberg’s account via DCD is accurate, is a visible sign that labour housing — not only transformers and chillers — is on the critical path.

Wider market colour DCD ties in

DCD links the timing to a tougher IPO climate for AI infrastructure names. Earlier this month, Australian neocloud Firmus dropped plans for an IPO, blaming “volatile market conditions,” as TSN covered in Firmus Pulls Its IPO…. Separately, Gartner’s forecast of more than $3.5 trillion of AI infrastructure that could face write-downs by 2030 and AirTrunk’s reported delay of its roughly $1.5bn Singapore REIT IPO show how delivery risk and public-market timing are colliding across the sector.

None of that proves Nscale will change its IPO plans. It does explain why a reported delay at Glomfjord, published while the company is preparing to go public, will be read closely against the prospectus risk factors DCD quotes.

What’s still unknown

Nscale’s own account of schedule, cooling leaks, the transformer fire, or the Kalmar hotel
Whether the six-month figure is measured against an original energisation date or a later internal plan
How the 6.4MW / 30MW split maps to customer commitments with Anthropic, Microsoft or others
Any IPO timetable (not stated in the DCD piece used here)

What to watch

Any Nscale statement on Glomfjord schedule or incidents
Updates to energised megawatts at the Norwegian site
How prospectus risk language on construction delays is handled in subsequent investor materials

What this means

Reportedly six months behind at Glomfjord (Bloomberg via DCD).
Hurdles cited: construction delays, local rules, cooling leaks, transformer fire.
6.4MW of planned 30MW operational (Bloomberg via DCD).
IPO context per DCD: valuation reports up to $35bn; Sept prospectus 55MW live across five sites; $103bn contracted value (Anthropic, Microsoft); company-claimed revenue jump and large losses.
Nscale has not commented (DCD asked).