NVIDIA Invests 2 Billion Dollars in Nebius as the Neocloud Race for AI Infrastructure Intensifies
NVIDIA announced a 2 billion dollar strategic investment in Amsterdam-based Nebius Group, granting the AI cloud provider early access to next-generation Rubin GPUs and Vera CPUs as both companies target over 5 gigawatts of deployed capacity by 2030.
Overview
NVIDIA announced on March 11 that it will invest 2 billion dollars in Nebius Group N.V. (NASDAQ: NBIS), a Dutch AI cloud provider, in a strategic partnership aimed at building hyperscale infrastructure for the agentic AI era. The deal grants Nebius early access to NVIDIA’s next-generation Rubin accelerators, Vera CPUs, and BlueField storage systems, and targets the deployment of more than 5 gigawatts of NVIDIA computing capacity by the end of 2030.
Shares of Nebius surged approximately 16 percent following the announcement, reflecting investor enthusiasm for the neocloud provider’s growing strategic position.
What We Know
The partnership deepens a relationship that spans the full AI technology stack. Under the agreement, NVIDIA and Nebius will collaborate on AI factory design and support, including access to partner design materials, technical review processes, and early hardware samples. The companies will also work together on building inference and agentic AI software stacks using NVIDIA’s latest technologies.
Nebius will deploy multiple generations of NVIDIA infrastructure, starting with early adoption of the Rubin platform, Vera CPUs, and BlueField storage systems. According to SiliconANGLE, Rubin GPUs run inference workloads 10 times more cost-efficiently than current Blackwell cards, and systems pairing Rubin with Vera deliver 10 times better performance per watt than previous-generation hardware. Each rack-scale unit contains 72 Rubin GPUs, 32 Vera CPUs, and over 1,000 additional chips.
Jensen Huang, NVIDIA’s chief executive, stated that “Nebius is building an AI cloud designed for the agentic era, fully integrated from silicon to software and powered by NVIDIA’s next-generation accelerated compute.” Nebius CEO Arkady Volozh said the companies are “extending that throughout the stack” from gigawatt-scale AI factories to inference and software.
Nebius currently operates data centers across the United States, the United Kingdom, France, Finland, and Iceland. A 300-megawatt campus is under construction in New Jersey, and the company has separately secured approval to build a 1.2-gigawatt AI factory near Independence, Missouri.
What We Don’t Know
The precise terms of NVIDIA’s equity stake in Nebius, including the share price and resulting ownership percentage, have not been disclosed in the press releases reviewed. It also remains unclear how Nebius’s deployment timeline for the next-generation Rubin and Vera hardware will align with NVIDIA’s production schedule, which has not been publicly detailed.
The competitive dynamics between Nebius and CoreWeave, in which NVIDIA made a similar 2 billion dollar investment and both companies have pledged to reach 5 gigawatts of capacity by 2030, raise questions about how NVIDIA will allocate its next-generation hardware between the two providers. NVIDIA also placed a separate 6.3 billion dollar infrastructure order with CoreWeave.
Analysis
The investment marks NVIDIA’s latest move in a pattern of multi-billion-dollar strategic bets on neocloud operators. In addition to Nebius and CoreWeave, NVIDIA has recently invested 2 billion dollars each in chipmakers Lumentum and Coherent, and previously took a 2 billion dollar stake in chip design firm Synopsys.
Nebius distinguishes itself from rivals through its full-stack engineering approach. Originally spun off from Yandex’s European operations and renamed in 2024 after selling Yandex’s Russian business for 5.2 billion dollars, the company was purpose-built for AI workloads rather than adapted from general-purpose cloud infrastructure. This heritage, combined with prior multi-billion-dollar deals with Microsoft and Meta, positions Nebius as one of the most aggressively scaling neocloud providers in the market.
The 5-gigawatt deployment target by 2030 represents a significant commitment at a time when the data center industry is already straining under energy demands driven by AI workloads. Whether Nebius, CoreWeave, and other neocloud operators can secure the power, land, and permitting required to meet these ambitions remains one of the sector’s defining open questions.