OpenAI has withdrawn its plans to secure compute resources directly from a data center operated by Nscale, a UK AI cloud company located in Norway. Instead, Microsoft is taking over the substantial spare capacity that was originally intended for OpenAI.
Details of the Stargate Facility Adjustment
The specific facility at the heart of this change is the planned 230MW “Stargate Norway” site in Narvik. Previously, OpenAI had been positioned to be an initial major consumer of resources at the data center, which was part of a larger infrastructure project also branded as “Stargate.”
Sources with direct knowledge of the matter indicated that OpenAI and Nscale ultimately could not reach an agreement on an off-take deal for a substantial portion of the facility’s capacity. Microsoft has since entered discussions to acquire this necessary compute power, allowing OpenAI to pivot its strategy.
“We are moving ahead with our plans in Norway,” stated an OpenAI spokesperson speaking to CNBC. “Microsoft is an important partner in our network and we will work with them to access compute in Norway just as we already do in other parts of the world.”
Financial Strategy and Partnerships
The company explained that coordinating with Microsoft makes more financial sense for its operations, as this arrangement falls under existing contracted spending. The spokesperson referenced an earlier October announcement detailing a contract to purchase $250 billion worth of services from Azure, which is Microsoft’s cloud-computing division.
Adding to the collaboration, Nscale announced that Microsoft is expanding its agreement at the Narvik campus by deploying over 30,000 Nvidia Rubin GPUs. Previously, in March, Nscale had stated it would assist Microsoft with rolling out Nvidia’s Vera Rubin platform across sites not only in the UK and Norway but also elsewhere.
“Expanding our work with Nscale in Narvik helps ensure Microsoft customers have access to the advanced AI infrastructure they need as demand continues to grow across Europe,” said Jon Tinter, president of business development and ventures at Microsoft, in a statement issued Tuesday.
Broader Spending Context
The shift reflects OpenAI’s effort to manage expectations regarding its massive spending commitments, particularly with a potential Initial Public Offering (IPO) anticipated this year. The company had recently halted plans for the U.K. Stargate project, citing concerns over energy costs and the country’s regulatory landscape.
In addition to these changes, OpenAI previously announced in March that it was shutting down its video generation service Sora. Despite recent funding adjustments, the startup continues to attract significant investment; notably, it closed a record $122 billion funding round with a post-money valuation of $852 billion.
In February, OpenAI informed investors that by 2030, the company anticipates having total compute expenditures near $600 billion. This follows comments made by CEO Sam Altman in November, who had suggested the firm would commit to $1.4 trillion in infrastructure spending over the subsequent eight years.
