Nebius Just Reported 684% Revenue Growth—and It's Building the AI Factory That Explains Why

If you want to understand how fast AI infrastructure demand is actually growing, look at Nebius. The Nvidia-backed AI cloud company reported first quarter 2026 revenue of $399 million today—a 684% increase over the same period last year—and simultaneously announced it had secured 1.2 gigawatts of power and land for a new AI factory in Pennsylvania.

The numbers are not a typo. This is what happens when AI compute demand runs ahead of supply.

The Q1 Numbers

Nebius's Q1 2026 results beat analyst expectations across the board. Revenue hit $399 million, up from approximately $50 million in Q1 2025. The company's annualized recurring revenue (ARR) reached $1.9 billion, and Nebius raised its capital expenditure guidance to $25 billion—a signal that demand shows no sign of slowing.

The company's stock surged Wednesday as investors digested results that dramatically outpaced consensus estimates. Nebius (ticker: NBIS on Nasdaq) has become one of the most watched names in AI infrastructure as pure-play cloud providers capable of serving frontier AI workloads at scale are in short supply.

The Pennsylvania AI Factory

The more significant announcement may be the infrastructure buildout. Nebius confirmed it secured up to 1.2 gigawatts of power and land in Pennsylvania for a new owned AI factory—its second gigawatt-scale location in the United States, alongside a 1.2 GW facility already underway in Independence, Missouri.

"AI factory" is the term Nebius uses deliberately. These aren't traditional data centers hosting websites and enterprise software. They're purpose-built facilities optimized for AI training and inference—packed with Nvidia GPUs, designed for the extreme power and cooling requirements of running frontier models at scale.

The Pennsylvania rollout follows a phased timeline: lights up by end of 2027 with an initial 250-300 MW, adding roughly 300 MW per year through the early 2030s until reaching the full 1.2 GW capacity.

How Nebius Built This Position

Nebius's backstory is unusual. The company spun out of Yandex—Russia's dominant internet company—in 2024 when Yandex restructured its international operations amid geopolitical pressures. The new Nebius Group, led by Arkady Volozh, retained Yandex's AI infrastructure capabilities and rebuilt as an independent, Western-listed AI cloud provider.

The Nvidia backing matters enormously. Nebius has secured priority access to Nvidia's latest GPU generations at a time when supply constraints have been the primary bottleneck for AI cloud expansion. That preferential supply relationship is a structural competitive advantage that most cloud providers can't replicate.

What 4 Gigawatts Means

To understand the scale of what Nebius is building, some context helps. One gigawatt of power capacity can support roughly 100,000 Nvidia H100 GPUs running continuously. At 4 GW total contracted capacity—which Nebius now targets by 2030—the company would be operating one of the largest AI-dedicated infrastructure networks in the world.

Nebius now has contracted power exceeding 3.5 GW across its facilities, with owned capacity representing more than 75% of that total. The owned-versus-leased ratio matters because owned capacity provides more stable long-term economics and greater control over facility design and operations.

The Demand Side of the Equation

Nebius's growth isn't happening in a vacuum. The company is benefiting from the same structural force driving all AI infrastructure investment: the gap between how much AI compute enterprises need and how much exists.

Training large language models and running inference at scale requires an enormous amount of GPU compute. The hyperscalers—Amazon, Microsoft, Google—are building as fast as they can, but demand still routinely outstrips supply. That creates an opening for well-capitalized, GPU-rich alternatives like Nebius that can serve customers who can't get what they need from AWS or Azure.

More than 1,000 business customers now spend over $1 million annually on Nebius services—a figure that has grown significantly in recent quarters, according to company disclosures.

The Competitive Context

Nebius operates in an increasingly crowded AI cloud market. CoreWeave (recently acquired by Nvidia), Lambda Labs, and emerging players like IREN (which announced a $3.4 billion Nvidia contract earlier this year) are all competing for the same enterprise AI customers. The differentiator in this market is reliable GPU access, facility quality, and the engineering talent to help customers extract maximum performance from their clusters.

Nebius's Nvidia relationship—which predates the current GPU shortage and runs deeper than typical customer-vendor arrangements—gives it a supply advantage that shapes every competitive interaction.

What's Next

Nebius's trajectory points to continued aggressive expansion through 2027 and beyond. The Pennsylvania site comes online at an important moment: by late 2027, Nvidia's next-generation GPU architectures are expected to be in volume production, and facilities with capacity to house them will be in high demand.

Watch Nebius's ARR trajectory in Q2 and Q3 as the leading indicator of whether this 684% growth rate can sustain. The company's ability to convert contracted power into deployed capacity on schedule will be the operational proof point.

The Bottom Line

Nebius's Q1 2026 results—684% revenue growth, $399 million in revenue, and a 1.2 GW Pennsylvania facility—are a snapshot of what AI infrastructure demand actually looks like when it's unconstrained by supply. The company has built a GPU-rich, rapidly-scaling position in a market where demand is running ahead of everything. If AI continues to eat enterprise compute budgets at the current pace, Nebius's next few years could be even more dramatic than its last few quarters.