Nebius Q3 2026 Earnings Call - NEBIUS GROUP NV

Financial Performance:

• Group revenue reached $980.8 million, a 571% year-over-year increase and 68% sequential rise; Nebius AI Cloud accounted for 99% of total revenue ($973.8 million).
• Adjusted EBITDA surged to $465.8 million from a loss of $5.2 million a year ago, with margins expanding to 47.5% (Q2: 40.6%); AI Cloud margin reached 53.2%.
• Approximately $325 million in non-recurring revenue was recognized in Q3 from asset-light partnerships deploying Nebius architecture.
• Capital expenditures (CapEx) totaled $7.3 billion, driven by server/network equipment and site expansion; Group cash and equivalents stood at $7.4 billion.
• Debt and financing activities included a $775 million asset-backed facility, a $5.75 billion convertible senior notes offering, and $1.0 billion in equity issuance via the at-the-market program.

Guidance and Future Outlook:

• 2026 Revenue guidance updated to $3.0–$3.2 billion (previously $3.0–$3.4 billion) and $3.0–$3.2 billion, reflecting timing shifts of some third-party capacity to 2027.
• 2027 Revenue outlook projected to quadruple to $12–$13 billion, driven by the ramp-up of new capacity.
• Pre-tax profitability expected in Q4 2027, assuming five-year depreciation schedules and full ramp-up of owned sites.
• 2026 CapEx guidance raised to ~$27 billion (previously $20–$25 billion); 2027 CapEx expected to be $40–$45 billion for over 1 GW of new capacity.
• 2027 Funding plan anticipates 75–85% of CapEx funded via customer prepayments ($17–$20 billion) and asset-backed financing of investment-grade contracts.
• Target ARR exceeded $7 billion by February 2027, even before New Jersey capacity fully ramps; ARR guidance discontinued in favor of revenue and earnings metrics.

Business Segments and Product Lines:

• Nebius AI Cloud remains the core revenue driver, supported by a low-cost full-stack model (hardware, racks, storage, software).
• Token Factory (managed inference platform) accelerated growth in Q3/Q4, targeting multi-hundred-million-dollar ARR with beta margins by leveraging owned GPUs.
• Enterprise adoption expanded with multi-year agreements from Databricks, Datadog, Snowflake, and Palantir; AI natives continue to represent the majority of new contract value.
• Asset-light partnerships recognized $325 million in Q3 non-recurring revenue; management expects this to become a recurring revenue stream as the partner footprint grows.
• Software stack enhancements included acquisitions (EGON, Clarify, InferEyes) to support model loading, cold-start reduction, and inference optimization.

Market and Competitive Landscape:

• Pricing power demonstrated with Q3 agreements closing above $25 million per MW, up from $20–$25 million in Q2; Q4 saw the initiation of multi-billion-dollar agreements.
• Demand remains robust with customers committing to longer durations and higher upfront prepayments; 90% of Q3 contract value included prepayments.
• Contract mix strategy optimized across three horizons: spot/short-term premium, 1–3 year core contracts, and 5-year long-term agreements.
• Competition addressed via vertical integration (owning infrastructure and software), offering superior margins compared to resellers or rented GPU providers.
• Open-weight model adoption is accelerating, viewed as a positive for expanding the customer ecosystem and use cases.

Risks and Challenges:

• Site delays at third-party co-location facilities (e.g., New Jersey, Vineland) due to power generation challenges (DataOne temporary generation wind-down); delays shifted some 2026 revenue to 2027.
• Vineland delay specifically attributed to the shift from diesel to Bloom micro-turbines and the accelerated wind-down of temporary power; project remains on track with electricity on-site.
• Industry-wide power supply and grid interconnection remain primary constraints for scaling capacity.
• Revenue recognition timing variance due to non-recurring asset-light deals and capacity deployment phases may cause quarterly EBITDA fluctuations.

Management Commentary and Tone:

• Management expressed high confidence, characterizing Q3 as a "record quarter" with a "market-leading" position.
• Tone emphasized resilience and control; while acknowledging industry-wide delays, management highlighted that owned sites remain "on track and even ahead of schedule."
• CEO Arkady emphasized that while supply/demand dynamics will eventually balance, the company is building a model to thrive in both favorable and challenging environments.
• CFO Dado and IR highlighted the shift to "prudent" but confident guidance, with a focus on profitability and cash generation over growth at all costs.

Analyst Questions and Answers:

• Financing Strategy: Prepayments now cover >100% of CapEx for new contracts, providing flexibility to substitute external debt when market rates are high; management confirmed ability to tap prepayments for additional cash needs.
• Regulatory/Land Use: Political shifts (e.g., Texas) and community questions do not anticipate meaningful timeline impacts; management emphasizes active engagement with state leaders.
• Asset-Light Revenue: The $325M non-recurring revenue is a result of the first major asset-light transaction; future deployments from this model are expected to become recurring.
• Customer Mix: While AI natives dominate the $12B+ committed value since Q3, enterprise adoption (platforms and direct) is accelerating, evidenced by new enterprise logos.
• Supply Chain: Management confirmed GPU and network chip supply is secured for 2027, with no anticipated bottlenecks.

Other Key Points:

• Contract Tenure: Management highlighted a shift toward longer contracts to diversify tenure and facilitate financing.
• Power Strategy: Increased reliance on Bloom micro-turbines for on-site power generation, with plans to deploy this solution at other sites to mitigate grid delays.
• Strategic Pivot: Shift from ARR guidance to Revenue and Pre-tax Earnings as primary KPIs to reflect the scale and maturity of the business.
• Market Dynamics: Prices rising and contract values growing; customers paying more upfront to secure capacity and lock in rates.
• Platform Evolution: Token Factory now supports advanced routing, caching, and post-training, positioning Nebius as a comprehensive AI infrastructure and software provider.