Microsoft is investing heavily in Narvik
Microsoft has entered into a lease agreement with data center operator Nscale for a facility in Narvik and plans to equip it with a total of 30,000 Nvidia Vera Rubin accelerators. This is reported by CXO Digitalpulse, supported by Microsoft's own disclosures regarding European infrastructure development.
Vera Rubin is Nvidia's next-generation GPU architecture — a successor to the Hopper series — and is specifically designed for large language models and generative AI in data center format. Microsoft's choice of this platform for Narvik signals that the facility will handle the most demanding AI workloads.
30,000 Vera Rubin accelerators in one Norwegian data center puts Narvik on the map as one of Europe's most significant AI hubs.
Part of a Massive European Offensive
The Narvik project is not an isolated measure. According to research based on Microsoft's own investment announcements, the company has spent over $20 billion on AI and cloud infrastructure in Europe over the past 16 months. Capacity is set to increase by 40 percent in the next two years, and by 2027, Microsoft will have more than doubled its European data capacity compared to 2023 levels.
Compared to other European projects, the Narvik installation is impressive:
In Portugal, Microsoft — in collaboration with Nscale, Nvidia, and Start Campus — is building a data center in Sines with 12,600 Nvidia GPUs, valued at $10 billion. In the UK, an investment of $30 billion in AI infrastructure has been announced for 2025 to 2028.
Why Narvik?
Narvik possesses several characteristics that make the city attractive for data center operations: access to clean, hydropower-based electricity, a stable cool climate that reduces cooling needs, and robust fiber infrastructure. These factors are often highlighted by industry players when considering Northern Norwegian locations.
Nscale, Microsoft's lessor in Narvik, is a European player also involved in Microsoft's projects in Portugal and the UK — indicating closer strategic cooperation between the two companies.
Microsoft's Chip Strategy: Nvidia, AMD, and Proprietary Chips
Microsoft does not only use Nvidia in its European infrastructure. According to the company's own statements, it plans to deploy “the most advanced AI accelerators from both Nvidia and AMD” in its European facilities.
In addition, Microsoft has developed two proprietary chip types:
James Sanders, an analyst at CCS Insight, has previously pointed out that proprietary chips give Microsoft “greater flexibility to acquire the computing power the company needs” — and that the connection to OpenAI makes the investment in the Maia platform particularly strategic.
What Does This Mean for Norwegian Business and Research?
An infrastructure of this size in Norway could have concrete implications for Norwegian users of the Azure platform. Greater local GPU capacity can provide lower latency, higher performance, and potentially better availability of powerful AI services for Norwegian businesses, public sector entities, and research communities.
It is important to emphasize, however, that Microsoft has not disclosed a detailed timeline or total cost for the Narvik project. Known figures and plans stem from the company's broader European investment communications and third-party reporting — not from a separate Norwegian press release.