From Yandex Spinoff to AI Infrastructure Leader
Nebius Group traces its origins to the international operations of Yandex, Russia's largest search engine. After geopolitical events in 2022 forced a corporate restructuring, founder Arkady Volozh separated the international technology assets from the Russian business and rebranded the company as Nebius Group. Headquartered in Amsterdam and listed on Nasdaq, Nebius retained the AI and cloud computing capabilities that Yandex had built over two decades, including deep expertise in machine learning infrastructure.
Volozh bet the company's future on AI infrastructure. Rather than competing in consumer search or advertising, Nebius focused on providing the GPU compute capacity that AI companies need for training and running large models. The bet paid off: revenue grew from $92 million to $530 million in a single year, driven by demand from AI labs, enterprises, and hyperscalers that need access to large-scale GPU clusters.
The $3 Billion Meta Deal and Nvidia Backing
Two partnerships validate Nebius's infrastructure capabilities. Meta signed a five-year agreement worth approximately $3 billion for Nebius to deliver AI infrastructure. When a company the size of Meta, which builds its own data centers, outsources AI compute to a third party, it signals that Nebius offers something Meta cannot easily replicate: speed of deployment, GPU availability, or specialized infrastructure configuration.
Nvidia participated in a $700 million private fundraise in December 2024. Nvidia investing in its own customers is common in the AI infrastructure space, but it provides Nebius with both capital and a strategic relationship that ensures access to next-generation GPU hardware. The combination of Meta as an anchor customer and Nvidia as a strategic investor gives Nebius credibility that most AI infrastructure startups lack.
Financial Performance
- •FY2025 Revenue: $529.8 million, up from $91.5 million (479% growth)
- •Q4 2025 Revenue: $227.7 million, up from $35.2 million a year earlier
- •Capital Expenditures: $4.1 billion in FY2025; $2.1 billion in Q4 alone
- •Contracted Power: More than 2 GW secured across data center locations
- •Loss from Continuing Operations: Narrowed to $29 million despite massive capex ramp
- •2026 Target: Annualized revenue run-rate of $7-9 billion by year-end
Growth Catalysts
- •Meta Revenue Ramp: $3 billion five-year contract provides revenue visibility and anchors capacity utilization; additional hyperscaler deals could follow
- •GPU Cluster Demand: AI model training requires massive GPU compute; enterprises and AI labs increasingly outsource to specialized providers rather than building their own
- •2 GW Power Pipeline: Contracted power capacity supports expansion into multiple data center locations; power access is the primary bottleneck for AI infrastructure buildout
- •Nvidia Relationship: Strategic investment ensures priority access to next-generation GPUs (H200, B200, GB200); hardware access is a competitive advantage
- •European Market Position: Nebius has strong presence in European AI infrastructure where data sovereignty requirements favor regional providers
Risks and Challenges
- •Capital Intensity: $4.1 billion in 2025 capex with the revenue base still below $1 billion; the company must continue raising capital to fund GPU purchases and data center construction
- •Customer Concentration: The Meta deal represents a significant portion of future revenue; losing or reducing this contract would create a large revenue gap
- •Hyperscaler Competition: AWS, Google Cloud, Microsoft Azure, and Oracle all offer GPU cloud services with larger customer bases and established enterprise relationships
- •GPU Depreciation Risk: GPUs depreciate as newer generations launch; ASIC alternatives for specific AI workloads could reduce GPU demand
- •Execution Risk: Scaling from $530M to a $7-9B run-rate in one year requires flawless execution on data center buildout, power procurement, and customer onboarding
Competitive Landscape
Nebius competes in the AI infrastructure-as-a-service market alongside CoreWeave, Lambda, and the hyperscalers (AWS, Google Cloud, Azure). CoreWeave is the most direct competitor, also focused on GPU cloud for AI workloads with heavy capex spending and hyperscaler contracts. Lambda targets AI researchers with developer-friendly GPU cloud. The hyperscalers have the largest infrastructure but spread attention across millions of customers and use cases.
Nebius differentiates through vertical integration: the company builds and operates its own data centers, manages GPU clusters directly, and optimizes the software stack for AI workloads. The Yandex heritage provides decades of machine learning expertise that pure infrastructure plays lack. The European presence also matters for customers with data sovereignty requirements that prevent using US-based cloud providers.
Who Is This Stock Suitable For?
Perfect For
- ✓High-growth investors who believe AI infrastructure demand will sustain triple-digit revenue growth through 2026-2027
- ✓Those who want exposure to AI compute buildout without buying Nvidia directly
- ✓Investors who see the Meta deal and Nvidia investment as validating Nebius's technology and execution capability
- ✓Speculative growth investors comfortable with pre-profit companies spending aggressively on capacity
Less Suitable For
- ✗Income or value investors (no profitability, no dividend, massive capex spending)
- ✗Risk-averse investors (extreme capital intensity, customer concentration, execution risk)
- ✗Those who believe hyperscalers will dominate GPU cloud and squeeze out specialized providers
- ✗Investors uncomfortable with the $7-9B run-rate target that requires near-perfect execution
Investment Thesis
Nebius Group is attempting one of the most aggressive growth ramps in technology history: from $530 million in 2025 revenue to a $7-9 billion annualized run-rate by end of 2026. The Meta deal, Nvidia backing, and 2+ GW of contracted power provide the foundation. CEO Arkady Volozh brings decades of large-scale infrastructure experience from building Yandex's technology platform.
The risk matches the ambition. $4.1 billion in annual capex against sub-$1 billion revenue means Nebius is spending far more than it earns, funded by external capital. If AI infrastructure demand slows, GPU costs decline faster than expected, or the Meta contract underperforms, the company faces significant financial pressure. For investors who believe the AI infrastructure buildout is just beginning and that specialized GPU cloud providers will capture a meaningful share alongside hyperscalers, Nebius offers the highest-growth exposure in the category.