Nebius Group stock hit an all-time high of $214.44 on Tuesday May 13, 2026, surging between 15 and 18 percent on earnings released before the opening bell that showed the Amsterdam-based AI cloud company had grown revenue by 684 percent year-over-year in the first quarter.
The stock had been trading around $180 before the report. The 52-week low was $32.88.
The numbers were big enough to move a stock that was already up over 120 percent year-to-date and that went into earnings with one of the most anticipated quarterly reports in the AI infrastructure sector.
The Q1 2026 Numbers
Nebius reported first quarter 2026 revenue of $399 million, against a consensus analyst estimate of approximately $375 to $389 million.
That beat was meaningful but not the number that sent the stock to an all-time high. The Q1 2025 comparison revenue was approximately $50 million.
A move from $50 million to $399 million in a single year is the 684 percent growth figure, and it reflects a business that has been transformed at its foundation by the contracts it signed with Meta and Microsoft in 2025.
The earnings per share figure was the more dramatic beat. Analysts expected a loss of $0.78 per share.
Nebius delivered a loss of $0.23 per share. That is a 70 percent smaller loss than the market had priced in, not because the company dramatically cut spending but because the revenue was large enough to absorb costs in ways the models had not anticipated.
The full-year 2026 guidance confirms that Tuesday’s quarter was not a one-time event.
Nebius is guiding for $3 to $3.4 billion in 2026 revenue. Full-year 2025 revenue was $530 million.
The guidance implies the company expects to grow at a pace that makes the first quarter’s performance look like the starting point rather than the peak.
The Number Everyone Is Missing
The revenue beat and the EPS beat will dominate the first wave of coverage. The number that deserves more attention is the Adjusted EBITDA line.
One year ago, Nebius’s Adjusted EBITDA was a loss of $53.7 million. In Q1 2026, it flipped to a profit of $129.5 million. That is not a marginal improvement.
That is a company that moved from spending roughly 336 percent of its revenue on operating costs to spending 132 percent, a structural efficiency improvement of the kind that operating leverage is supposed to deliver but rarely does at this speed.
The cost of goods as a percentage of revenue tells the same story more specifically.
A year ago, Nebius spent 49 cents on every dollar of revenue on the cost of goods. In Q1 2026, that number had fallen to 26 cents.
The GPU cluster infrastructure that produces Nebius’s revenue has become dramatically more efficient to operate at scale than it was at smaller scale. That is the signal that the business model is working, not just growing.
The full profitability picture is still evolving. The company is spending aggressively on capital expenditures, $16 to $20 billion in planned spending for nine new data centers, and the free cash flow burn remains real.
The direction of the margins and the speed of the improvement suggest that the path to sustained profitability is becoming clearer.
The $50 Billion Backlog That Makes This Different From A Growth Story
Most high-growth technology companies that trade at extreme valuation multiples are asking investors to trust that future revenue will materialize. Nebius has something more concrete than trust.
A contracted revenue backlog approaching $50 billion, built from three agreements that represent some of the largest AI infrastructure deals ever publicly disclosed.
Meta Platforms expanded its agreement with Nebius in March 2026 to as much as $27 billion for dedicated GPU infrastructure.
Meta’s AI ambitions, including its own large language models and the computing infrastructure to run them at scale, require exactly the kind of dedicated GPU capacity that Nebius builds and operates.
A $27 billion commitment is not a letter of intent. It is a contracted anchor for a significant portion of Nebius’s forward revenue.
Microsoft signed a five-year deal with Nebius in September 2025 worth up to $19.4 billion. Microsoft’s own AI infrastructure needs, supporting both Azure’s commercial AI services and the internal compute demands of Copilot and related products, have been growing faster than its own data centers can accommodate.
The Nebius deal fills a portion of that gap.
Nvidia invested $2 billion in Nebius in March 2026, acquiring an 8.3 percent stake.
Nvidia CEO Jensen Huang described Nebius as the infrastructure backbone for what he called the “agentic era,” the next phase of AI development where AI systems take multi-step autonomous actions rather than simply responding to prompts.
When the company that makes the chips your entire business depends on also becomes a major equity owner and says you are essential to the next phase of their industry, the signal is as clear as investment signals get.
The Eigen AI Acquisition
Two weeks before the earnings report, on May 1, Nebius announced the acquisition of Eigen AI for approximately $643 million in cash and Class A shares.
The announcement drove a 12 to 14 percent single-day stock gain on its own, the stock was already at a high before the earnings poured more fuel on the momentum.
Eigen AI specializes in making AI model deployment faster and cheaper by reducing the compute and memory requirements needed to run large models.
The specific technology, chip performance optimization for inference and post-training workloads, addresses one of the most significant cost constraints in AI deployment.
When a company wants to run a large language model in production, serving millions of queries per day, the compute cost at inference time is often the primary operating expense. Eigen’s technology reduces that cost.
Nebius is integrating Eigen into its Token Factory platform, the managed inference product that allows developers to deploy, refine and scale AI models.
The integration moves Nebius from being primarily an IaaS provider, a company that rents GPU capacity, toward being a PaaS provider, a company that offers software-defined AI services with higher margins and deeper customer relationships.
The distinction matters because pure infrastructure providers compete primarily on price and capacity.
Platform providers compete on performance, features and the software ecosystem they offer on top of the hardware.
The Missouri AI Factory
On May 12, the day before earnings, Nebius broke ground on its flagship AI factory campus in Independence, Missouri.
The Independence campus will be Nebius’s first gigawatt-scale AI factory in the United States, with up to 1.2 gigawatts of power and land already secured.
The company is building nine new data centers total as part of a $16 to $20 billion capital expenditure plan.
The scale of those numbers, one location requiring up to 1.2 gigawatts of power, reflects the physical reality of what AI infrastructure at the frontier actually requires.
A gigawatt of power is roughly what a city of 750,000 people uses. The AI factories that train and run the most powerful models consume power at urban scale.
Nebius’s decision to build its own facilities, rather than relying on colocation data centers or leasing capacity from other cloud providers, reflects its strategy of owning the infrastructure economics rather than renting them.
Owned capacity is more capital-intensive upfront and more profitable per unit at scale.
The Short Squeeze Dynamic
More than 20 percent of NBIS’s float was held short going into Tuesday’s earnings.
Short interest at that level, more than one in five shares outstanding borrowed and sold by investors betting the price would fall, creates a specific kind of volatility when an earnings report beats dramatically to the upside.
Short sellers who borrowed shares must eventually return them, which means buying shares.
When the stock rises sharply on earnings, short sellers face mounting losses and begin buying to cut their losses, which adds buying pressure on top of the buying from investors who want to own the stock because of the earnings beat.
The combination produces moves that are larger than the fundamental news alone would generate.
Tuesday’s 15 to 18 percent move is likely amplified by this dynamic in ways that are difficult to separate from the pure earnings reaction.
What The Stock Is Pricing In
At Tuesday’s all-time high of $214.44, Nebius’s market capitalization was well above $40 billion.
The $3 to $3.4 billion in full-year 2026 revenue guidance puts the stock at approximately 12 to 13 times forward revenue, a multiple that is aggressive but not insane for a company growing at this rate with this much contracted backlog.
The price-to-earnings ratio is effectively uninformative at current earnings levels given the investment phase the company is in.
The 12 analysts covering the stock have a consensus Strong Buy rating. The average price target before earnings was approximately $163 to $170, already below where the stock was trading heading into the report.
Price target upgrades in the days following the earnings release will reset the analyst consensus closer to where the market is trading.
The year-to-date gain of more than 120 percent and the one-year gain of more than 450 percent reflect a market that has been consistently repricing this company upward as each new contract, acquisition and earnings report confirms the scale of what is being built.
Tuesday’s report added another confirmation.
