OpenAI Withdraws from Stargate Norway Deal Amid Microsoft Takeover

OpenAI Pivots Norway Data Center Plans, Opts for Microsoft Cloud Amidst Infrastructure Reassessment

OpenAI has shifted its strategy regarding a significant compute capacity acquisition in Norway, opting to directly leverage Microsoft’s extensive cloud infrastructure rather than securing direct access from a planned “Stargate Norway” data center. This development follows a similar adjustment in the United Kingdom, underscoring a broader recalibration of the AI pioneer’s ambitious hardware expansion plans.

The proposed 230MW facility in Narvik, Norway, initially envisioned as a cornerstone of OpenAI’s “Stargate” infrastructure initiative, was being developed by UK-based AI cloud startup Nscale. OpenAI had previously signaled its intention to be a primary offtaker for a substantial portion of this facility’s computing power. However, negotiations between OpenAI and Nscale for a definitive offtake agreement have not materialized.

A spokesperson for OpenAI confirmed to CNBC that the company is now in discussions to rent compute capacity directly from Microsoft. This strategic alignment with its key partner is seen as a more financially prudent approach, integrating seamlessly with OpenAI’s existing contracted spending with Microsoft’s Azure cloud services. “We are moving ahead with our plans in Norway,” the spokesperson stated. “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.”

This decision aligns with OpenAI’s October announcement detailing a multi-year agreement to purchase $250 billion in services from Microsoft Azure. The move also reflects a broader trend of hyperscalers consolidating their control over critical AI infrastructure. Microsoft itself announced an expansion of its collaboration with Nscale at the Narvik campus, increasing its deployment of Nvidia’s latest GPUs. This strategic pivot by OpenAI suggests a preference for greater flexibility and integration within established cloud ecosystems as the demand for raw compute power escalates.

The rationale behind this shift appears multi-faceted. OpenAI is reportedly recalibrating its substantial infrastructure spending projections amidst burgeoning market expectations, particularly as a potential initial public offering (IPO) looms. The company confirmed last week that it had paused its U.K. “Stargate” project, citing concerns over energy costs and the regulatory landscape. Furthermore, OpenAI recently announced the discontinuation of its Sora video generation service, signaling a more focused approach to cost management and resource allocation.

These adjustments come at a time when OpenAI has secured significant funding, including a record $122 billion round in March that valued the company at $852 billion post-money. However, the sheer scale of AI compute requirements is pushing the boundaries of capital expenditure. Earlier this year, OpenAI informed investors it was targeting approximately $600 billion in total compute spend by 2030, a figure that followed CEO Sam Altman’s November remarks estimating $1.4 trillion in infrastructure commitments over the next eight years.

The Narvik facility’s capacity, originally slated for OpenAI, will now be absorbed by Microsoft. This integration of advanced AI infrastructure by Microsoft at the Narvik campus is designed to meet the escalating demand from its global customer base across Europe. Jon Tinter, president of business development and ventures at Microsoft, emphasized in a statement that this expansion “helps ensure Microsoft customers have access to the advanced AI infrastructure they need as demand continues to grow across Europe.”

OpenAI’s current strategic maneuvers underscore the immense capital demands of cutting-edge AI development and the intricate dance between AI developers, chip manufacturers, and cloud providers. As the industry grapples with building out the necessary computational backbone, direct partnerships with hyperscalers like Microsoft offer a pathway to scale while potentially mitigating some of the direct capital risks associated with building and operating vast data center infrastructure. The long-term implications for the competitive landscape of AI compute will be closely watched as OpenAI continues to refine its infrastructure strategy.