Morgan Stanley: Nebius AI Cloud Strategy Validated, Maintain Neutral Rating

Overview

Morgan Stanley released a conference review report on Nebius Group (NBIS.US), maintaining an Equal-weight (Neutral) rating and a $144 target price based on observations from the 2026 Nebius Inflection Point Conference. The report believes the conference effectively validated Nebius's positioning as a differentiated 'New Cloud' (Neocloud), proving it has genuine enterprise customer demand beyond hyperscalers and complete AI software stack capabilities. Although the long-term growth logic is solid, given that the current valuation is relatively full, short-term net new booking targets appear aggressive, and profitability awaits verification, the institution chose to maintain the existing rating.

Core views

Nebius is successfully reshaping market perception, transitioning from a pure GPU computing broker to a full-stack AI cloud service provider. The report points out that for skeptics, this conference provided key evidence that Nebius positions itself between traditional cloud providers and small GPU specialists: it possesses more AI-native attributes and responsiveness than traditional clouds, while being more reliable and having full-stack capabilities compared to smaller vendors. Production-grade AI customers need not only chips but also reliability, orchestration, inference performance, and cost optimization tools, which are precisely Nebius's core differentiating advantages. The breadth and quality of customers were the most significant incremental benefits of this event. The company showcased a diversified customer base covering AI labs, product companies, enterprises, and digital natives, directly addressing market concerns about over-reliance on hyperscale customers. Specific cases include Recraft achieving a 6x increase in training speed, Brave generating over 16 million AI summaries daily, Shopify using it for pre-training and fine-tuning, and Mastercard adopting Agentic Search. This diversification of customer structure not only validates the authenticity of demand but also lays the foundation for future margin expansion, as cloud services directly targeting enterprises and AI-native customers typically have higher profit potential than large bare-metal contracts. Supply-demand imbalance remains the core contradiction currently, with capacity becoming a key bottleneck for growth. Feedback from multiple customers indicates that due to insufficient capacity or high prices at hyperscalers, they turned to Nebius for alternatives, but Nebius itself faces challenges in resource acquisition. The product roadmap is increasingly tied to production-grade AI and Agent workflows, emphasizing accelerated computing, storage, orchestration, serverless AI, and Agent toolchains. The report believes that if Nebius can continue to overlay platform capabilities on top of infrastructure, it will help increase revenue value per MW and enhance customer stickiness, thereby achieving a value leap from 'selling resources' to 'selling services'.

Analysis framework

The report adopts an 'event-driven + risk-return framework' analysis method. First, by dissecting specific customer cases and product updates from the Nebius Inflection Point Conference, it qualitatively verifies the execution of the company's AI cloud strategy, focusing on whether it has摆脱ed dependence on a single type of customer and the commercialization progress of its software stack. Second, at the valuation level, considering the company is in a period of high capital investment and has not yet stabilized profits, the institution abandons P/E valuation and instead uses the EV/EBITDA multiple method, discounting forward valuation based on projected 2028 EBITDA. Finally, combining bull, base, and bear case scenarios, it assesses the premium degree of the current stock price relative to long-term fundamentals, leading to the conclusion of maintaining a neutral rating.

Methodology notes

• Valuation MethodEV/EBITDA valuation

  For AI infrastructure companies with high capital expenditure, large depreciation and amortization, but unstable profitability, use Enterprise Value multiples (EV/EBITDA) rather than Price-to-Earnings (P/E) for valuation.

  The report values Nebius at 8x projected 2028 EBITDA. This method excludes the impact of massive depreciation and differences in capital structure, better reflecting the core operating cash flow generation capability of heavy-asset AI cloud companies in a steady state. It is a common paradigm for evaluating such 'New Cloud' vendors.

• Competition and Strategic FrameworkMoat / competitive advantage

  Assess competitive barriers by analyzing the company's unique position in the industry chain (between hyperscale clouds and small GPU vendors) and its full-stack capabilities.

  The report emphasizes that Nebius's moat lies in 'differentiated positioning': more flexible than big players, more full-stack than small players. This analytical approach suggests to investors that in a homogenized computing market, only vendors with added value such as software orchestration, inference optimization, and engineering support can avoid becoming mere participants in price wars.

• Industry/Sector Analysis FrameworkSupply-demand framework

  Identify 'capacity availability' as the core constraint variable at the current stage, rather than just a demand issue.

  The report repeatedly mentions 'demand > supply' and 'capacity is the limiting factor'. During the AI infrastructure construction period, the focus of analysis often shifts from 'whether there are buyers' to 'whether delivery is possible'. This framework helps investors understand why, even with strong demand, the company's short-term revenue may still be limited by the physical pace of electricity acquisition, data center construction, and GPU deployment.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Nebius Group NV (NBIS.US)

  Core coverage target, benefiting from AI cloud demand validation and full-stack strategic transformation

  Strengths

  Differentiated Neocloud positioning; genuine diversified enterprise customer base; vertically integrated full-stack AI capabilities; robust balance sheet with strategic asset equity stakes

  Weaknesses

  Aggressive short-term net new booking targets; profitability and track record not yet proven; high capital intensity

  Comparison

  More AI-native and responsive than hyperscalers; more reliable and possessing full-stack software capabilities compared to small GPU specialists

  Risks

  Normalization of AI demand or large customers building their own computing power; data center delivery difficulties exceeding expectations; rising financing costs; intensified competition compressing pricing

Key data

• Target Price$144.00Derived from 8x projected 2028 EBITDA ($11.5 billion), implying approximately 35% downside from the current stock price of $220.12
• 2028 Projected ARR~$13bnThe report expects strong demand and capacity launch to drive Annual Recurring Revenue to this level by the end of 2028
• 2028 Projected Adjusted EBITDA Margin64.0%As revenue scale expands and software stack added value increases, margins are expected to improve significantly from -12.2% in 2025
• Base Case Valuation Range$130 - $300Corresponding to 8.5x-11x 2028 EBITDA; Bull case $400 (12x), Bear case $70 (7x)
• Customer Efficiency Cases6x / 16mn+Recraft training speed increased 6x; Brave processes over 16 million AI summaries daily, validating actual platform productivity

Impact & implications

The report believes that the Nebius Inflection Point Conference eliminated some market concerns that the company was merely a 'GPU subletter', confirming its path towards evolving into a high-value AI cloud platform. The diversification of customer structure and the refinement of the software stack mean that long-term unit economics are expected to be superior to pure computing rental models. However, the current stock price already implies high growth expectations, and the company still requires significant capital expenditure in the short term to realize capacity, with the profitability inflection point yet to arrive. Therefore, although the fundamental trend is positive, upside potential may be suppressed until valuation is fully digested or earnings visibility further improves.

Risks

• Normalization of AI demand environment or large customers building their own AI computing power in the future leading to demand decline
• Data center construction and electricity acquisition difficulties higher than expected, hindering capacity launch
• Intensified market competition and macro regulatory dynamics putting pressure on pricing and return on investment
• Difficulty raising funds at attractive or improving debt costs
• Short-term net new booking targets too aggressive, with high execution risk

What to watch

• Delivery progress of newly signed power, grid-connected power, and active power
• Signing of new customer contracts and Net New Bookings
• Software stack innovation and improvements in Attach Rate
• Evidence of monetizing GPUs beyond the estimated four-year useful life
• Operational leverage, margin improvement, and free cash flow improvement driven by economies of scale