Nvidia has committed roughly $2 billion to Nebius through pre-funded warrants, taking a minority stake while helping finance an aggressive buildout of AI cloud infrastructure. The deal highlights a structural shift in the artificial intelligence market: the next stage of competition is not only about better models and software, but also about who can secure enough electricity, cooling, land, and advanced chips to support industrial-scale computing.
According to the companies, the partnership is designed to support a long-term infrastructure plan that could exceed 5 gigawatts of Nvidia-accelerated computing capacity by 2030. That is an extraordinary figure in data-center terms and signals the growing reality that AI is becoming a heavy industry as much as a digital one.
A strategic bet on an AI-native cloud operator
Nebius, headquartered in Amsterdam and listed on Nasdaq under the ticker NBIS, is part of a new class of infrastructure providers often described as “neocloud” companies. Unlike legacy cloud operators that adapted existing enterprise platforms for AI workloads, Nebius has positioned itself as an AI-native provider built specifically for large-scale training, inference, and emerging agentic AI applications.
The company’s history is closely tied to the restructuring of Yandex’s international assets. In 2024, after the Dutch parent of the Russian search group divested its Russian businesses for about $5.4 billion, the remaining international operations were reorganized under the Nebius name. Since then, Nebius has been building out AI infrastructure operations across the United States, Europe, and Israel.
That positioning is important for Nvidia. By backing Nebius financially, the chipmaker is not only investing in a customer but also reinforcing a distribution and deployment partner that is expected to absorb large quantities of Nvidia hardware over many years. The relationship extends across the stack, from data-center design and GPU fleet management to inference infrastructure and future software deployment.
Hardware roadmap and ecosystem alignment
Nebius already operates substantial GPU clusters based on Nvidia’s H100 and H200 accelerators. Under the expanded partnership, the company plans to adopt future Nvidia technologies, including the Rubin architecture, Vera CPUs, and BlueField storage platforms. This makes the arrangement strategically valuable for both sides: Nebius gains preferential alignment with the industry’s dominant AI chip ecosystem, while Nvidia secures a large-scale infrastructure partner with a clear appetite for capacity growth.
Nvidia founder and CEO Jensen Huang framed the investment as part of the next phase of the AI boom. In the joint announcement, he said AI has reached another inflection point, with agentic AI driving surging compute demand and accelerating infrastructure expansion. Nebius CEO Arkady Volozh similarly presented the company as a cloud platform designed for AI from the start, rather than one retrofitted from older computing models.
The emphasis on “agentic AI” is notable because it suggests demand growth will increasingly come not only from training foundation models but also from inference-heavy workloads tied to autonomous and semi-autonomous software systems. That trend tends to favor operators that can scale quickly and manage specialized infrastructure efficiently.
Missouri project shows the scale of ambition
The companies’ long-term targets are not merely theoretical. Days before the investment was announced, Nebius received approval for a 1.2-gigawatt AI data-center campus in Independence, Missouri. The site is described as one of the largest planned AI infrastructure projects in the United States.
The Missouri development is expected to create around 1,200 construction jobs and generate an estimated $650 million in economic impact over two decades. While those numbers speak to local development benefits, they also illustrate how AI infrastructure is increasingly being treated as a major industrial and regional investment category, with implications for labor markets, utility planning, and state-level economic policy.
Missouri is only one element of a wider global expansion strategy. Nebius said it could operate as many as 16 data-center sites worldwide by the end of 2026, with contracted power capacity approaching 3 gigawatts within the next year. In practical terms, those figures reinforce the idea that access to power is becoming one of the defining constraints in AI deployment.
Huge spending plans despite modest current revenue
Nebius is spending at a pace that reflects how urgently infrastructure providers are trying to get ahead of future AI demand. The company reported capital expenditures of about $2.1 billion in the fourth quarter of 2025 alone. For 2026, it expects total spending in a range of $16 billion to $20 billion.
Those outlays are massive relative to the company’s current revenue base. Nebius reported trailing twelve-month revenue of roughly $530 million, underscoring that investors and partners are valuing the company more for its future capacity and strategic positioning than for its present financial scale. Management has pointed instead to long-term contracts and backlog as evidence that demand can support the expansion.
Among the largest disclosed commercial relationships is a multi-year agreement with Microsoft tied to a data-center project in Vineland, New Jersey, estimated at between $17 billion and $19.4 billion. Nebius also disclosed a separate $3 billion arrangement with Meta Platforms. Those contracts help explain why the company is willing to spend heavily before revenue fully catches up.
Market reaction and the bigger signal for AI
Investors welcomed Nvidia’s move. After the announcement, Nebius shares rose by roughly 13% to 16%, lifting the company’s market capitalization above $24 billion in early trading. The response reflects a pattern often seen in the market: when Nvidia backs a company directly, investors tend to interpret that as a validation of both demand visibility and strategic relevance.
Beyond the stock move, the more important message is about the nature of the AI bottleneck. The industry is no longer constrained only by software innovation. It is increasingly constrained by physical infrastructure: power generation, transmission access, cooling systems, land development, and the supply of advanced semiconductors. Building AI at scale now resembles an industrial supply-chain challenge as much as a computing challenge.
Nvidia’s investment in Nebius therefore serves two purposes. On one level, it helps a fast-growing AI cloud operator finance rapid capacity expansion. On another, it reinforces Nvidia’s broader ecosystem strategy by enabling more customers and developers to access Nvidia-based infrastructure at hyperscale. The transaction also suggests that the winners in AI may not be limited to model builders; they may also include the companies capable of assembling the physical backbone that makes those models usable in the real world.
As AI demand spreads from training into inference and agentic systems, infrastructure intensity is likely to grow rather than shrink. In that environment, access to gigawatt-scale power, deployment-ready campuses, and reliable chip supply may become just as decisive as breakthroughs in algorithms. Nvidia’s $2 billion commitment to Nebius is one of the clearest recent signs that the race to build the AI backbone is accelerating.