FILE PHOTO: The logo of Nebius during the Viva Technology conference dedicated to innovation and startups at Porte de Versailles exhibition center in Paris, France, June 12, 2025. REUTERS/Benoit Tessier/File Photo

By Kritika Lamba

(Reuters) - Nebius Group's shares jumped nearly 44% to $91.75 on Tuesday, touching a record high as investors cheered its $17.4-billion deal supplying AI infrastructure to Microsoft over a five-year period.

The agreement, which could expand to $19.4 billion if Microsoft increases demand, underscores the growing need for high-performance AI data centers amid a global rush to develop advanced generative artificial-intelligence technologies.

The stock, which has more than doubled this year, surged 43.3%, while shares of rival CoreWeave rose 4.4%.

Analysts believe Nebius is now well positioned to win additional marquee customers, including other hyperscalers and frontier AI labs, as it expands its data center footprint through upcoming greenfield projects.

"This deal provides unprecedented clarity on the company's long-term revenue potential and significantly de-risks its planned capacity buildout," said Hamed Khorsand, analyst at BWS Financial.

Microsoft, which has repeatedly flagged a shortage of AI cloud infrastructure due to soaring client needs, has been turning to third-party providers to bridge the gap.

The tech major has a similar multibillion-dollar deal with CoreWeave, which also has a substantial contract with OpenAI for AI computing resources.

Founded from the spinoff of Russian tech giant Yandex's assets, Amsterdam-based Nebius will primarily supply Microsoft with dedicated GPU infrastructure from the company's new data center in Vineland, New Jersey, beginning later this year.

Founder and CEO Arkady Volozh said the Microsoft contract complements Nebius' core business and will accelerate its AI cloud growth in 2026 and beyond.

The company's core business is providing clients with full-stack AI cloud services built on Nvidia computing, giving AI developers the software and hardware tools and services needed to build and run their models.

(Reporting by Kritika Lamba in Bengaluru; Editing by Pooja Desai)