Morgan Stanley has released a research report on the AI cloud business of Nebius Group (NBIS.US), providing a comprehensive analysis of its business transformation, customer base, product strategy, financial outlook, valuation, and potential risks. The firm has assigned the stock a "Neutral" rating, while viewing the overall industry as attractive, with a price target of $144.
The report references the company's 2026 industry summit, which further validated its overarching AI cloud development strategy. Beyond major hyperscale cloud providers, various enterprises and AI startups are demonstrating genuine procurement demand. The company has built a complete software technology stack, enabling it to increase revenue per megawatt of computing power, improve profit margins, and efficiently allocate resources to meet massive market demand.
The summit clearly showcased the business models and application scenarios of enterprise clients and AI startups, while fully presenting the company's comprehensive, full-stack AI capabilities. It provided tangible evidence of real customer demand and confirmed the broad market opportunities among traditional enterprises and native AI firms, making the company's software-centric AI cloud strategic path increasingly clear.
Previously, Nebius focused on securing large partnership contracts to facilitate financing and business expansion. The next phase will involve delivering computing resources to a broader range of enterprise clients, driving increases in both revenue per megawatt and profit margins, while building long-term differentiated competitive advantages based on its proprietary AI cloud software stack.
The summit also provided stronger evidence that Nebius's customer base is no longer limited to leading hyperscale cloud providers. The company is gradually transitioning into a high-value AI cloud service provider, moving beyond its initial role as a mere supplier of raw computing power.
Addressing Market Skepticism
The summit is sufficient to dispel the views of market bears and skeptics: Nebius is not merely a GPU computing power middleman. The company has strategically positioned itself between traditional large cloud vendors and emerging cloud service providers. Compared to traditional cloud vendors, its products are natively suited for AI scenarios with higher response efficiency. Compared to smaller GPU specialty providers, it offers stronger service stability and possesses full-stack technical capabilities.
The summit conveyed a clear signal: for enterprises deploying production-grade AI operations, relying solely on raw chips is insufficient. A suite of supporting capabilities is required, including stable operation and maintenance, technical support, resource scheduling, inference performance optimization, and cost control.
Client Cases as a Significant Positive
The disclosed partnership clients represent a major incremental highlight. Nebius's current clientele has significantly broadened, encompassing AI labs, AI product companies, traditional large enterprises, and digital-native firms. Notable case examples include: Recraft, which achieved a sixfold increase in model training speed using Nebius services; Brave, which processes over 16 million AI content summaries daily on the platform; e-commerce platform Shopify, which uses its computing power for model pre-training and fine-tuning; Mastercard, which utilizes Nebius's intelligent search function; and Higgsfield, which lists Nebius as its preferred infrastructure partner.
These diverse client cases also address a key investor concern: as the business expands, Nebius has the capability to sell computing power to a varied client base, not solely relying on large partnership contracts with major cloud service providers.
Optimized Client Structure Benefits Margin Improvement
Management reiterated that large contracts with major hyperscale cloud providers primarily support overall AI cloud infrastructure construction. The company prefers to engage directly with end customers. Comparatively, AI cloud business directly with enterprises and native AI firms yields significantly higher profit margins than large contracts for raw computing power alone. When combined with value-added features like software services, inference computing, intelligent routing, and intelligent agents, the profit potential expands further.
Strong Demand Meets Supply Bottleneck
Multiple clients indicated that market demand is very robust, but insufficient computing resources are the primary constraint. Shopify noted a turning point in GPU demand, with traditional large cloud providers either having insufficient capacity or pricing too high, leading them to seek alternatives. Rhoda highly praised the platform's stability, low latency, technical support, and resource availability, yet still struggles to obtain sufficient computing power. Higgsfield affirmed the platform's excellent inference performance, but stated its own business demand growth far outpaces current supply.
Feedback from several clients points to the same reality: market demand significantly exceeds the supply of computing power.
Product Roadmap Focuses on Production-Grade AI and Intelligent Agents
Nebius's product planning increasingly revolves around deployed, production-ready AI operations and intelligent agent applications. The company emphasizes that supporting stable AI operation in production environments requires a full suite of capabilities: high-performance computing, storage, resource scheduling, serverless AI, inference computing, intelligent routing, operational status monitoring, real-time scenario adaptation, and intelligent agent tools.
The current industry reality is that most enterprises can build basic intelligent agents, but maintaining these applications stably and cost-effectively is extremely difficult. This validates the core logic of Nebius's software-driven development. By embedding more platform functionalities into its computing infrastructure, the company can achieve higher-value commercialization and further strengthen customer stickiness.
Summary of Client Perspectives
Clients choosing Nebius primarily value its operational performance, service stability, delivery speed, professional technical support, low latency, and flexible, efficient service response. Shopify noted more efficient technical engagement compared to traditional large cloud vendors. Rhoda highlighted the platform's streamlined architecture coupled with comprehensive resource scheduling, storage, and managed database capabilities. Higgsfield views Nebius as its top partner for inference performance.
Common industry pain points include scarce computing resources, difficulties with pricing and cost control, complex model routing logic, and challenges in maintaining AI systems in production. These pain points directly correspond to Nebius's current key R&D focuses: inference computing, resource scheduling, intelligent routing, and intelligent agents.
Product and Platform Developments
The company has established a complete AI cloud technology stack, including NVIDIA GPU servers, networking equipment, AI-specific storage, accelerated computing, resource orchestration, serverless AI, enterprise operations platforms (MLOps), inference services, and intelligent agent tools represented by NebiusEcho.
The core of this product update is not a single new item, but a strategic pivot. Nebius is evolving from a foundational infrastructure service provider into a full-process, production-grade AI workflow service provider.
Financially, Morgan Stanley predicts explosive revenue growth for the company, from just $530 million in 2025 to an estimated $17.929 billion in 2028. Profitability is expected to improve continuously, with operating margins and adjusted EBITDA margins rising annually, losses narrowing, and a gradual move toward profitability, with primary revenue coming from North America.
Regarding valuation and investment rating, the report assigns a "Neutral" rating to the stock while viewing the overall industry as attractive. With a closing price of $211.69 on June 9th, Morgan Stanley's price target is $144. The report outlines three scenarios: an optimistic scenario with a target of $400, assuming large-scale computing power deployment and software monetization exceeding expectations; a base scenario with a target range of $130 to $300, based on steady performance improvement through computing power delivery and client expansion; and a pessimistic scenario with a $70 target, weighed down by factors like demand contraction, increased competition, and deployment delays. Among market institutions, 65% assign a Buy or equivalent rating, while 35% rate it Neutral.
Comments