Nebius Group emerged from the reorganisation of Yandex’s international operations after the company divested its Russian businesses in 2024. What remains is an AI-focused cloud infrastructure platform that provides GPU compute, data labelling (through Toloka), autonomous driving technology (through Avride), and educational technology. At $56 billion market capitalisation and $227.81 per share, Nebius has quickly established itself as a pure-play bet on AI infrastructure demand.
The core business is straightforward: Nebius builds and operates GPU clusters that customers rent to train and run AI models. In an environment where demand for GPU compute far exceeds supply, that is an exceptionally well-positioned business. The company has secured significant GPU capacity from NVIDIA and is building data centres across Europe, the Middle East, and North America.
What makes Nebius distinctive is its heritage. The team that built Yandex’s search engine, cloud platform, and AI capabilities brings deep technical expertise that most competitors in the GPU-as-a-service space lack. This is not a company that recently pivoted to AI. It has been building AI infrastructure for over a decade, just under a different name.
NBIS sits in a Markup phase with strong momentum across multiple analytical layers. This is one of the more decisive markup signals we have seen in the mid-cap technology space.
The volume profile tells a story of aggressive institutional accumulation. Large block trades, consistently positive volume delta on up moves, and a steady reduction in shares available for lending all point to the same conclusion: big money is building positions.
The convergence screener flags NBIS as passing multiple gates, which is notable given how selective the screener is. Most stocks fail on at least two or three layers. NBIS passes on the majority, suggesting genuine multi-dimensional strength rather than a single-factor momentum trade.
The risk, as with any markup phase, is that the stock has already moved significantly. Chasing markup is one of the most common mistakes in position management. The framework does not tell you to buy. It tells you the conditions are favourable for those already positioned or willing to wait for a pullback entry.
Nebius scores an exceptional 93.7 on our ethical screening framework, the highest of any name in this case study series. Several factors contribute to this strong result.
The complete separation from Russian operations removes the most significant ethical concern that previously surrounded Yandex. Nebius has no business activities in Russia, no data centres there, and no revenue exposure to the Russian market. The divestiture was clean and comprehensive.
The core business of providing AI compute infrastructure is ethically neutral-to-positive. Customers use GPU clusters for medical research, climate modelling, drug discovery, and a wide range of beneficial AI applications. The data labelling business through Toloka is conducted with transparent labour practices and fair compensation models, which stands in contrast to some competitors in the data annotation space.
The autonomous driving division, Avride, operates with proper regulatory compliance and safety testing frameworks. Educational technology through TripleTen provides accessible technology education, contributing to workforce development and digital literacy.
Governance has improved markedly post-restructuring, with independent board representation and transparent reporting standards aligned with NASDAQ listing requirements.
At $56 billion, Nebius trades at a substantial premium to current revenue, reflecting the market’s expectation that AI infrastructure demand will drive rapid top-line growth for several years. The revenue multiple is elevated by any historical standard, but comparisons to traditional cloud infrastructure companies may not be appropriate given the supply-demand dynamics in GPU compute.
The bull case is compelling on paper. Global AI compute demand is growing at triple-digit rates. NVIDIA’s data centre revenue trajectory provides a proxy for the total market opportunity, and Nebius is capturing a growing share of the infrastructure layer that sits downstream of GPU sales.
The bear case centres on competition and capital intensity. AWS, Azure, and GCP all offer GPU compute services with greater scale and existing customer relationships. CoreWeave, Lambda, and other pure-play GPU cloud providers are raising billions to compete in the same space. The barriers to entry are financial, not technological, which means well-capitalised competitors can close the gap.
Quarter-over-quarter revenue growth in the core AI infrastructure segment is the primary metric. Anything below 50% sequential growth would raise questions about whether demand is translating into booked revenue at the expected rate.
Nebius is building GPU clusters across multiple geographies. Track construction timelines and capacity utilisation rates. Delays in bringing new capacity online directly impact revenue potential.
Access to next-generation GPUs (Blackwell, Rubin) is critical. Any disruption in Nebius’s allocation from NVIDIA would constrain growth. Monitor the partnership dynamics closely.
Early-stage AI infrastructure companies often have concentrated customer bases. Watch for diversification across industries and geographies. Heavy reliance on a small number of large AI labs would increase revenue risk.
Track NBIS on the ticker page and in our daily Alpha Insights.