Amsterdam-based AI infrastructure firm Nebius Group has agreed to provide Microsoft with a GPU infrastructure capability from a new data centre in Vineland, New Jersey. The deal is valued at $17.4 billion over a five-year period. With optional expansions, the total contract could reach $19.4 billion by 2031.
The announcement sent Nebius shares soaring more than 47% after hours Monday, before surging 50% in early Tuesday to $95.80.
A few months back, Nebius secured $1 billion through a private placement of convertible notes. The offering comprises two senior unsecured tranches: $500 million in 2.00% convertible notes maturing in 2029 and $500 million in 3.00% convertible notes due in 2031.
Addressing a global capacity crunch
The deal highlights the critical shortage of high-performance computing as demand for AI continues to accelerate worldwide. Microsoft has invested record sums in new infrastructure, yet its executives acknowledge that the company remains capacity-constrained and unable to meet its own growing needs. The new partnership ensures a dedicated pipeline of GPU resources at a time when even hyperscale providers struggle to keep up.
Nebius’ specialised offering
At its core, Nebius provides GPU infrastructure built on Nvidia chips, alongside cloud software architecture and in-house hardware designed for demanding workloads. Customers use its services to train, fine-tune, and run machine learning models at scale, supported by managed tools and storage. While rivals like CoreWeave already serve Microsoft, Nebius’ entry marks a broadening of the tech giant’s supplier base.
The infrastructure rollout is scheduled to begin later this year in Vineland, with multiple deployment phases planned over a two-year period.
Nebius traces its origins back to the international operations of Yandex, the Russian internet company. Spun out as an independent business in 2023, Nebius has since rebranded as a neocloud provider, focusing squarely on GPU-driven infrastructure designed for training and deploying advanced machine learning models. Its positioning as a specialist provider allows it to compete with broader cloud giants by targeting the most resource-intensive workloads.
Why did Microsoft turn to Nebius?
Microsoft already maintains a substantial agreement with CoreWeave, another GPU infrastructure specialist backed by Nvidia. The addition of Nebius signals Redmond’s strategy to diversify supply lines as competition in artificial intelligence accelerates. Securing guaranteed access to GPU capacity is now a critical priority for hyperscale companies, and Nebius’ offering provides Microsoft with both capacity and geographic diversification.
Investors behind the growth
Nebius counts Nvidia and Accel Partners among its backers, strengthening its ties to the hardware and venture ecosystems driving AI infrastructure. Until recently, the company concentrated on supplying startups and mid-sized businesses with affordable access to computing power. The Microsoft contract represents a step-change in scale, positioning Nebius as a critical supplier to one of the world’s most valuable companies.
Looking forward
For Nebius, the Microsoft contract not only provides substantial guaranteed revenue but also validates its focus on purpose-built GPU infrastructure. With demand showing no signs of slowing, the company’s transition from a Yandex spinoff to a global infrastructure partner signals a new chapter.
For Microsoft, securing Nebius’ capacity offers breathing room in a race where every teraflop matters. As competition intensifies, both companies stand to benefit, Nebius through accelerated growth and Microsoft through critical supply security.
