A Neocloud Graduates to the Big Index
Nebius spent the past eighteen months transforming from a little-understood spin-out into one of the most closely watched names in AI infrastructure. That trajectory reached a symbolic peak this week. Nasdaq confirmed that Nebius will be added to the Nasdaq-100 index effective June 22, 2026, placing the AI cloud provider alongside the largest technology companies in the United States. The news helped push the stock up nearly 30 percent over the prior week and roughly 1,320 percent since its October 2024 listing, a run that reflects how aggressively investors have repriced anything tied to scarce AI compute.
Index inclusion is more than a vanity metric. Membership in the Nasdaq-100 forces passive funds tracking the index to buy the stock, broadening the shareholder base and lowering the cost of capital for a business that needs enormous amounts of it. For a company whose entire strategy rests on funding gigawatt-scale data center buildouts, cheaper and more reliable access to capital is a competitive weapon. The market is effectively betting that Nebius has crossed from speculative upstart to durable infrastructure provider, a transition that very few neoclouds have managed.
The Numbers Behind the Hype
The financials underpinning the rally are genuinely striking. In the first quarter of 2026, Nebius reported revenue of 399 million dollars, up 684 percent year over year, with its core AI cloud revenue rising even faster. The company swung to operating net income of 621 million dollars, a sharp reversal from a 104 million dollar loss in the same quarter a year earlier. For a sector where many players are burning cash to build capacity, demonstrating profitability at this scale of growth is what separates Nebius from the pack of also-ran GPU resellers.
The balance sheet is built for the buildout ahead. Nebius ended the first quarter with roughly 9.3 billion dollars in cash, having raised about 6.3 billion in the quarter through convertible notes and a 2 billion dollar investment from NVIDIA. Management has guided to full-year 2026 revenue of 3.0 to 3.4 billion dollars and set an annual recurring revenue target of 7 to 9 billion. Chief executive Arkady Volozh has summarized the demand backdrop bluntly, saying that "compute and cloud needs are vastly exceeding capacity" as enterprises move from AI experimentation to production deployment.
Two Hyperscaler Anchors
What truly separates Nebius from the crowded neocloud field is its customer roster. The company has signed a five-year agreement with Microsoft valued at up to 19.4 billion dollars, providing dedicated GPU capacity that includes more than 100,000 NVIDIA GPUs. On top of that sits a commitment from Meta that, between dedicated capacity and additional purchases, could reach roughly 27 billion dollars over its term. Winning anchor contracts of that magnitude from two of the largest hyperscalers on earth is a striking endorsement of an independent provider's ability to deliver at scale.
These deals matter for a structural reason that goes beyond revenue. Hyperscalers building their own data centers still cannot keep pace with their internal AI demand, so they increasingly rent capacity from specialized providers to bridge the gap. Nebius has positioned itself as exactly that overflow valve. The contracts provide the contracted, long-duration revenue that lenders and equity investors want to see before underwriting multi-billion-dollar capacity expansions. They also validate that a neocloud can meet the operational and reliability bar that companies like Microsoft and Meta demand of their own infrastructure partners.
The Capacity Race Is Really a Power Race
Behind the contracts lies a brutal physical reality: none of this works without power and the data centers to house it. Nebius currently operates seven data centers across North America, Europe, and Israel, and is targeting sixteen by the end of the year. Management plans to exit 2026 with between 800 and 1,000 MW of connected data center capacity, a dramatic jump from the roughly 170 MW it had at the end of 2025. The company is also securing additional sites that would bring total contracted power toward roughly 2.5 GW.
That is the same wall every AI infrastructure player is hitting at once. Connecting hundreds of megawatts of new load to the grid is slow, contested, and increasingly expensive, as regulators move to push more interconnection costs onto large users. Nebius has tried to get ahead of the constraint with a 2 billion dollar NVIDIA investment earmarked to scale more than 5 gigawatts of next-generation capacity over time. Execution against those power targets, not the GPU counts in its press releases, is the metric that will determine whether the company delivers on its revenue guidance.
Building Up the Stack
Nebius is not content to be a pure capacity provider. On June 10 the company completed its acquisition of Eigen AI, an inference and model optimization firm, in a move designed to climb up the value chain. Raw GPU capacity is increasingly commoditized, and the margin and the customer stickiness sit in the software layer that makes that hardware easier and cheaper to use. Folding inference optimization into what Nebius calls its token factory platform is an attempt to capture more of the value its customers generate rather than simply renting them compute by the hour.
This is a familiar playbook for infrastructure companies that want to avoid becoming undifferentiated utilities. By offering managed inference, model optimization, and a higher-level platform on top of its data centers, Nebius can charge for outcomes rather than just megawatt-hours. It also deepens the relationship with customers who would otherwise find it easy to shift workloads to whichever provider is cheapest that quarter. The Eigen AI deal is small relative to the Microsoft and Meta contracts, but it signals where management believes durable margins will eventually come from.
What Enterprise Buyers Should Read Into It
For technology leaders, Nebius is the clearest proof yet that the AI cloud market is broader than the handful of hyperscaler logos that dominate the conversation. A specialized provider can now win contracts measured in the tens of billions and connect capacity at gigawatt scale. Enterprises that have struggled to secure GPU allocation from the major clouds should treat credible neoclouds as a serious procurement option, particularly for large, predictable training and inference workloads where dedicated capacity and long-term pricing can be negotiated directly.
The cautionary note is the same one that applies across this entire sector. Nebius is valued for flawless execution against enormous capacity commitments, and any slip in power delivery, financing, or customer concentration could reset expectations quickly. Two customers account for a large share of its contracted revenue, and its growth depends on bringing online far more capacity than it has ever operated. The smart enterprise posture is to engage with providers like Nebius on their merits while diversifying across suppliers, because in a market this supply-constrained, no single source of compute, however impressive its quarter, should become a single point of failure.
