As investors look ahead to the next decade of artificial intelligence development, tech giants like NVIDIA, Alphabet, Microsoft, and Amazon often become the preferred targets. These companies have already invested hundreds of billions of dollars in the AI field, establishing competitive advantages through diversified cash-flow businesses, exceptional execution capabilities, and robust financial foundations. However, with market capitalizations each exceeding $2 trillion, achieving multi-bag returns from here would require staggering performance growth. Against this backdrop, emerging companies represented by Nebius Group (NBUS) offer investors a new balance of risk and growth potential.
Unlike hyperscale vendors offering general-purpose cloud services, Nebius focuses on building specialized infrastructure for AI training and inference workloads. Its distinctive feature lies in deploying large-scale NVIDIA GPU clusters, utilizing high-density power and liquid cooling systems, and operating a proprietary AI-optimized software stack. According to assessments by the commercial IT institution Uptime Institute, such dedicated AI cloud platforms can achieve cost savings of up to 66% compared to traditional data centers. This advantage is particularly crucial as energy consumption surges with the shift of AI workloads from training to inference.
Nebius’s third-quarter performance confirms the strong market demand for AI computing power. Revenue grew 355% year-over-year to $146 million, with nearly 90% coming from its core AI infrastructure business, which grew by 400%. Management emphasized that the company’s growth is limited only by available computing capacity, not market demand. As of the third quarter, the annualized revenue for its core business had reached $551 million. More notably, despite ongoing investment in new data centers, the adjusted profit margin for the core business still reached 19%, indicating a positive trend in profitability.
The multi-year agreements reached with Microsoft and Meta Platforms significantly enhance Nebius’s business credibility and development prospects. The five-year agreement with Microsoft is valued at $17.4 billion and can be increased to $19.4 billion; the deal with Meta is valued at $3 billion. The company notes that the scale of existing deals is still constrained by available capacity. While the revenue contribution from the Microsoft partnership is expected to be fully realized only by 2027, the deal with Meta will largely reach its annual revenue level by 2026. For a company with a market capitalization of approximately $22.3 billion, a contract backlog totaling over $20.4 billion provides a solid foundation for future growth.
With the launch of its third-generation AI cloud platform, Aether, Nebius has further refined its enterprise-grade security, compliance, and management tools. The concurrently launched Nebius Token Factory inference platform supports the large-scale deployment of open-source models, ensuring performance and pricing transparency. Against the backdrop of increasingly frequent AI inference workloads, these software innovations effectively strengthen customer partnerships and platform stickiness.
Despite facing execution risks associated with massive capital expenditure, and plans to support expansion through debt, asset-backed financing, and equity sales – which could increase liabilities or dilute equity – Nebius’s growth expectations remain compelling. The company anticipates annual revenue will reach $9-11 billion by the end of 2025, rising further to $70-90 billion in 2026. Combined with the context of its technological solutions being validated by tech giants, Nebius offers investors who can withstand short-term volatility an emerging opportunity to participate in the rapid development of the AI infrastructure sector.
