Your portfolio starts here.
Sign in to see your own allocation, returns, holdings and portfolio research signals. Your holdings remain private to your account.
Continue with Google →Company research on Nebius Group (NBIS), including the business, investment thesis, risks, and valuation.
Sign in to see your own allocation, returns, holdings and portfolio research signals. Your holdings remain private to your account.
Continue with Google →Company exposure across accounts and funds
Share of portfolio value by account
Signals that deserve a closer look
All accounts · September reference prices
| Company | Portfolio weight | Average cost / share | Current price | Return on cost | Research zone |
|---|
Only the designated FolioFacts administrator should sign in. Public user accounts are not available.
Continue with Google →Showing holdings in selected accounts.
Enter a holding and its account details.
Start with an existing export, or use our template.
Only the holdings you confirm are saved.
| Company | Account | Weight | Return | Shares | Cost basis | Value | Price date | Actions |
|---|
Delete your saved portfolio to remove its positions from this site’s active storage.
Ratings, opportunity scores, and fair values are independent research opinions based on stated assumptions—not personalized investment recommendations. Read our Disclaimer.
Search the US ticker directory by company or symbol. Tickers not found require your confirmation before adding.
| Watchlist rank | Company | Research tier | Zone | Price used | Quality | Model opportunity | Return to fair value | Fair value | Actions |
|---|
Watchlist ranks compare displayed Opportunity Scores across the full list. Filters keep those ranks; equal scores share a rank. Opportunity Scores remain research-model snapshots.
Explore full company reports, business fundamentals, and valuation analysis. Search by name, ticker, or tracking tier.
Price snapshot: 2026-11-03 · Research: 2026-10-21 (comparison refresh; fundamental review preserved from 2026-10-14) · Saved analysis, not live market data
Ratings and fair values are independent, dated research opinions—not personalized investment advice. Disclaimer
The reference price is below the probability-weighted fair value.
Dashed marker: reference price $151.64. Scenario values are present fair values, not dated price targets.
2029 revenue $12B; 2.5× exit EV/sales; $12B net debt; 350M diluted shares
Weaker utilization and pricing, unfavorable financing, more dilution2029 revenue $21B; 4.5× exit EV/sales; $9B net debt; 335M diluted shares
Company achieves rapid growth while remaining capital-intensive2029 revenue $32B; 5.5× exit EV/sales; $5B net debt; 325M diluted shares
Large efficient AI cloud provider; high utilization, strong prices, differentiated platformFolioFacts zones calculated from the saved fair and bear values at the dated reference price. Live zones can change with research.
Nebius has three important advantages working in its favor.
Reported FY totals, current FY quarters and cumulative YTD are grouped separately.
Scale within group: $540M
Scale within group: $600M
Scale within group: $1,000M
Completed FY, standalone current-year quarters and cumulative YTD. No TTM figures.
Each row covers a completed fiscal year. Scale within this group: $3,800M.
Each row covers an independent quarter; derived cash flow is labeled. Scale within this group: $3,600M.
YTD totals overlap the quarters above and must not be added to them. Scale within this group: $3,800M.
Each annual, quarter and YTD group uses a separate scale. SBC / FCF is a same-period comparison, not a literal employee cash payment. N/M means FCF is zero or negative (or SBC is negative); missing is not zero. Issuer FCF definitions may vary.
Only completed fiscal years and currently open fiscal-year quarters/YTD appear in charts. Standalone quarters and cumulative YTD are separate; missing is never zero. Legacy TTM rows remain in archival source text. The full financial profile below preserves definitions, accounting caveats, and period-end balance details.
This score uses the saved price and research fundamentals. Creator Watchlist calculates the current score and rank using verified market prices.
Watch the evidence behind the thesis, not just the ticker.
2026 guidance met or exceeded; strong 2027 outlook
Material shortfall or delayed recognition
Affects: Revenue forecasts and multiples800MW–1GW connected target delivered on schedule
Significant permitting or commissioning delays
Affects: Timing and execution riskPrepayments and lending reduce equity financing
Dilution or refinancing requirements accelerate
Affects: Per-share valuationStrong pricing persists across broader installed fleet
Falling revenue per active MW
Affects: Margins and terminal valuationMeasurable high-margin recurring revenue
Remains a small undifferentiated GPU service
Affects: Moat and Quality ScoreLonger economically useful GPU lives and attractive renewals
Faster hardware obsolescence
Affects: ROIC and free cash flowMore independent enterprise customers
Growing reliance on a few large contracts
Affects: Revenue durabilityNebius gains durable share
Hyperscaler insourcing or price competition
Affects: Long-term growth and multiplesNebius has three important advantages working in its favor.
First, AI infrastructure demand is moving faster than traditional data-center development cycles. A company capable of securing electricity, constructing GPU clusters, and putting them into service quickly can monetize scarce capacity before competitors finish building their facilities.
Second, Nebius has been securing contracts from some of the world's largest technology companies. Microsoft and Meta provide revenue visibility, financing credibility, and evidence that Nebius can operate infrastructure at a level acceptable to highly sophisticated customers.
Third, Nebius is attempting to become more than a supplier of rented NVIDIA GPUs. Its managed inference, orchestration, networking, storage, and developer tools could create differentiated services and customer relationships that survive beyond individual GPU generations.
However, none of these advantages guarantees an attractive shareholder return.
If Nebius keeps reinvesting nearly all available capital in hardware that depreciates rapidly, future revenue could grow spectacularly while free cash flow and returns on invested capital remain disappointing.
The optimistic interpretation is that investors underestimate how much revenue Nebius can generate from contracted and upcoming infrastructure, and how quickly the company can transition into higher-margin software and inference services.
The pessimistic interpretation is that investors are valuing Nebius like a scalable cloud software company before it has proven the economics of owning, financing, and continuously replacing its expensive infrastructure.
I lean toward the second concern at today's price. The business opportunity is credible, but its capitalization already reflects years of extremely successful execution.
Nebius emerged from the restructuring of Yandex's former international parent. Founder and CEO Arkady Volozh is now building an AI-focused cloud infrastructure company headquartered in Amsterdam, with operations and development centers across multiple regions.
Its commercial activities can be divided into the following groups.
| Business | What it does | Economic importance |
|---|---|---|
| Nebius AI Cloud | GPU clusters, networking, storage, compute orchestration, training and inference infrastructure | Dominant revenue and value driver |
| Dedicated AI infrastructure | Longer-term capacity contracted to large technology customers | Provides revenue visibility and supports project financing |
| Token Factory and AI services | Managed inference and related developer services | Potential to improve differentiation and margins |
| Asset-light partnerships | Operates or monetizes partner-owned capacity through Nebius technology | May allow growth without funding every dollar of hardware itself |
| Avride | Autonomous vehicles and delivery robots | Early-stage optionality, currently a group-level investment burden |
| TripleTen | Technology training and reskilling | Smaller, non-core business |
| Equity investments | Stakes including ClickHouse and Toloka | Additional value, but not the main investment thesis |
In Q2 2026, AI cloud revenue was approximately $574.9 million, or 98% of total group revenue. Nebius should therefore be valued primarily as an AI infrastructure and cloud platform, not as a diversified technology holding company.
Customers contract AI computing capacity — hyperscalers, AI labs, developers, and enterprises.
Nebius secures power, buildings, GPUs, and financing, using upfront investment, customer prepayments, debt, and equity.
Nebius provides training, inference, networking, orchestration, storage, and related compute and software services.
Revenue is earned over the contract term, less electricity, operations, equipment depreciation, interest, reinvestment, and dilution.
The final step is the most important. Customer cash received upfront and high adjusted EBITDA do not necessarily represent distributable profits.
| Customer | Announced agreement | Investment implications |
|---|---|---|
| Microsoft | Multiyear AI infrastructure agreement announced September 2025, with previously disclosed value up to approximately $17.4 billion | Establishes hyperscaler credibility, but creates large delivery commitments |
| Meta | $12 billion dedicated capacity commitment over five years beginning in 2027 | Supports next-generation infrastructure deployment |
| Meta — additional capacity | Up to $15 billion in additional capacity purchases under a backstop arrangement | Provides a utilization floor for selected capacity; not the same as $15 billion of guaranteed incremental revenue from customers |
The distinction in the Meta agreement matters. The company intends to sell some capacity to third-party customers, with Meta purchasing remaining eligible capacity. Therefore, describing the entire $27 billion as a simple fixed Meta revenue commitment would overstate the certainty and structure of the deal.
Recent strategic expansion also includes a September 2026 partnership with Palantir for sovereign AI and an October 1 acquisition of Inferize to improve production inference efficiency. These strengthen the platform strategy, but they do not yet establish a large, separately measurable source of recurring high-margin revenue.
Nebius already has real customers and significant operating scale. The unanswered question is whether its cloud technology can eventually become economically more valuable than the underlying GPUs.
The opportunity consists of several overlapping markets: AI training infrastructure, production inference, GPU cloud services, and associated data-center power and networking capacity.
I would avoid applying a single enormous generative-AI TAM to Nebius. Spending on AI models, applications, hardware, electricity, and services cannot all be treated as directly addressable Nebius revenue.
The most relevant demand drivers are the growing compute requirements of large models, the movement of AI systems into production, and the need for specialized GPU clusters with high-bandwidth networking. Inference may ultimately become a more durable demand source than sporadic frontier-model training runs, but it is also vulnerable to price competition and major efficiency improvements.
A consistent, independently comparable market-share dataset for specialized AI cloud infrastructure is not available. Reported revenue, contracted capacity, backlog, and installed GPU power use different definitions across companies.
Rather than invent market-share estimates, the following provides a more useful scale comparison.
| Company / Platform | Primary Position | Relative Scale | Growth Direction | Investment Relevance |
|---|---|---|---|---|
| Nebius | AI-native cloud and hyperscaler capacity | Rapidly scaling; $582M Q2 revenue | Exceptional | Capital and execution constraints |
| CoreWeave | Specialist GPU cloud | Much larger; $2.58B Q2 revenue | Strong | Leverage and financing burden |
| AWS, Azure, Google Cloud | Global hyperscalers | Vastly larger diversified platforms | Strong | Different economics and internal priorities |
| Oracle Cloud Infrastructure | Enterprise and hyperscaler cloud infrastructure | Substantial established platform | Expanding AI | Diversified exposure |
| Lambda and other private neoclouds | GPU-focused specialist providers | Private financial visibility limited | Growing | Access to capital, power, and customers |
CoreWeave's Q2 2026 revenue was approximately 4.4 times Nebius's revenue for the same quarter. That is a meaningful measure of current commercial scale, although their contract structures, hardware age, utilization, and accounting presentations differ.
Nebius has been targeting approximately 5 GW of contracted power by the end of 2026 and 800 MW to 1 GW of connected power by that date. Those are materially different operating milestones.
Management cautioned that connected power can take additional months to generate revenue. The 2026 buildout should therefore not be valued as though all committed power were already productive.
The addressable market is large enough to support much more growth. The limiting factors are likely to be electricity, execution, funding, and competition rather than a lack of potential workloads.
| Competitor | Market Share | Positioning | Advantages vs. Company | Disadvantages vs. Company | Current Direction |
|---|---|---|---|---|---|
| CoreWeave | Not comparable | Closest public AI-cloud competitor | Greater current scale, operating history, and infrastructure utilization | Nebius has potentially less entrenched debt burden and growing proprietary cloud capabilities | Scaling |
| AWS | Not comparable | General-purpose and AI cloud | Capital, global reach, enterprise distribution, ecosystem | Nebius offers more specialized GPU infrastructure and responsiveness | Expanding |
| Microsoft Azure | Not comparable | Hyperscaler cloud and AI services | Enterprise lock-in, capital, software integration | Independent infrastructure supply where Azure needs additional capacity | Expanding |
| Google Cloud | Not comparable | Cloud, TPUs, AI platforms | Vertical integration of hardware, models, and services | Nebius targets NVIDIA-centric workloads and customer choice | Expanding |
| OCI | Not comparable | Cloud infrastructure | Data-center expertise, hyperscaler resources | Nebius has AI-native platform focus | Expanding |
| Lambda and other specialists | Not comparable | Dedicated GPUs, AI cloud | Specialized offerings and targeted customer relationships | Nebius has scale of contracted deployments and integrated service stack | Growing |
I would classify Nebius's moat as developing, not yet durable, with a strengthening direction.
Its strongest potential advantages are execution capability, early access to power and sites, engineering expertise in dense GPU clusters, and the operational knowledge required to make extremely large AI systems run reliably.
Power access can provide a temporary competitive advantage. But the value of that advantage depends on how scarce power remains in the locations customers need.
Nebius's software may eventually create switching costs through integrated storage, networking, deployment, inference, and developer workflows. Today, however, most of its economic value remains closely tied to compute availability.
There is also a structural concern: Nebius is both a supplier to and a competitor of hyperscalers. Those companies may use external providers while capacity is scarce, then gradually bring more workloads onto their own infrastructure when their economics improve.
Moat direction: Strengthening, but not yet durable.
Nebius has credible emerging advantages, but I would not yet assign it the kind of durable moat premium associated with mature cloud platforms.
Standardized period history (reviewed 2026-11-07): Completed issuer fiscal years only, plus actually disclosed standalone quarters and cumulative YTD for the latest open FY. Currency USD, flows in millions, with fiscal dates defined by the issuer. Values marked derived come from disclosed period differences/sums, not TTM. Missing = N/D (not zero). Q1, Q2 and H1 FY2026 results are preserved from the existing filing-grounded report, with standardized fiscal-year labels. FCF excludes major capital investments and must not be confused with operating inflows from customer prepayments.
| Period | Revenue | Growth | Gross Margin | Operating Margin | FCF | FCF Margin | SBC | Diluted Shares | Cash | Debt |
|---|---|---|---|---|---|---|---|---|---|---|
| FY2024 | $91.5M | — | 52.2% | −436.7% | −$1,077.4M | −1,177.5% | $54.5M | Not specified | $2,434.7M | $6.1M |
| FY2025 | $529.8M | 479% | 68.6% | −112.5% | −$3,664.1M | −691.6% | $83.2M | Not specified | $3,678.1M | $4,127.7M |
| Q1 FY2026 | $399.0M | 684% | 74.0% | −32.1% | −$214.9M | −53.9% | $35.3M | Not specified | Not specified | Not specified |
| Q2 FY2026 | $582.3M | 454% | 77.1% | −30.2% | −$3,411.3M | −585.8% | $102.5M | 280.4M | $8,042.1M | $8,545.7M |
| H1 FY2026 | $981.3M | 529% | 75.8% | −31.0% | −$3,626.2M | −369.5% | $137.8M | Not specified | $8,042.1M | $8,545.7M |
Gross margins shown before depreciation and amortization because Nebius reports those costs separately. They are not directly comparable with peers that include infrastructure depreciation in cost of revenue. Q2 diluted weighted-average shares: 280.4 million; June 30 issued and outstanding shares: 271.9 million.
| Period | Revenue (USD millions) |
|---|---|
| Q3 2025 | $145.2 |
| Q4 2025 | $227.7 |
| Q1 2026 | $399.0 |
| Q2 2026 | $582.3 |
Nebius generated approximately $4.50 billion in operating cash flow during the first half of 2026, but that included a $4.40 billion positive change in deferred revenue, principally associated with customer funding.
That is commercially valuable. Receiving money before providing service lowers financing requirements.
However, that money comes with future obligations. Nebius still needs to deploy and operate the infrastructure for which customers have paid. Its $5.98 billion deferred revenue liability at June 30 is a reminder that received cash and earned revenue are different.
Free cash flow remained negative by approximately $3.63 billion in the first half, even after the benefit of those customer payments.
Nebius reported $259.7 million in Q2 depreciation and amortization against revenue of $582.3 million.
That is a substantial expense, not merely accounting noise. GPU hardware has finite economic usefulness. Rapid advances in performance per watt and per dollar can force service providers to upgrade hardware well before their physical assets stop working.
If all group depreciation and amortization were allocated against revenue and directly reported cost of revenue, the resulting gross contribution would be approximately 32.5% rather than 77.1%. This is an illustrative calculation, not the company's GAAP gross margin, but it shows why ignoring depreciation is misleading.
There is genuine financial support for the expansion: customer prepayments, hyperscaler contracts, equity proceeds, and GPU-backed lending.
There is also a large and growing capital requirement.
At June 30, Nebius reported $8.55 billion in debt and $8.04 billion in unrestricted cash. Those numbers do not incorporate the company's entire subsequent financing activity.
On August 24, Nebius closed approximately $5.75 billion in new convertible debt, while separately agreeing to exchange $800 million of older convertible debt for approximately 15.8 million Class A shares. The debt exchange lowers obligations but dilutes existing owners. The new notes provide funding while potentially creating additional dilution or refinancing requirements later.
During Q2, the company had already sold approximately 12.7 million shares through its at-the-market program, raising about $2.8 billion gross. Its earlier NVIDIA financing also introduced pre-funded warrants representing additional potential shares.
This is central to per-share valuation: a rapidly growing company can finance expansion successfully while ownership of its future profits becomes spread across more shares.
Financial direction: Improving operating scale and adjusted EBITDA, but high financing and free-cash-flow risk.
| Metric | Value |
|---|---|
| Price Snapshot | $151.64 — 2026-11-03 regular-session close, America/New_York |
| Market Capitalization | ~$50.3B |
| Enterprise Value | ~$52.4B |
| EV / trailing revenue | ~39.3× |
| EV / FY2026 midpoint revenue | ~16.4× |
| FY2026 revenue guidance | $3.0B–$3.4B |
| FY2026 year-end ARR guidance | $7B–$9B |
| FY2026 capex guidance | $20B–$25B |
Quoted market capitalization and enterprise value are vendor figures and may not fully incorporate recent post-quarter convertible exchanges, share issuance, and financing. They should be treated as indicative rather than as a reconciled pro-forma capitalization.
Nebius's June annualized run-rate revenue was approximately $3 billion. Management's target of $7–9 billion by year-end is promising, but annualized exit revenue is not the same as full-year recognized revenue. The market is pricing a steep future revenue ramp.
Only the price row is refreshed here. Other quoted valuation statistics retain their earlier source dates and definitions.
| Company | Relevance | Public / Private | EV / Trailing Revenue | EV / FY2026 Revenue | Growth / Margin / Risk Context |
|---|---|---|---|---|---|
| Nebius | Direct AI-cloud operator | Public | ~39× | ~16× guided midpoint | Smaller base, extremely fast growth, large capex |
| CoreWeave | Closest public operating peer | Public | Roughly high-single-digit multiple | Not specified in approved analysis | Greater revenue scale; heavy debt and interest burden |
| Lambda | Direct operating peer | Private | N/A — private | N/A — private | Specialized GPU cloud, limited public financial disclosure |
| AWS / Azure / Google Cloud | Secondary economic anchors | Diversified public parents | N/A independently traded | N/A independently traded | Mature distribution, infrastructure scale, diversified earnings |
| OCI | Secondary cloud infrastructure anchor | Diversified public parent | No clean standalone comparable multiple | No clean standalone comparable multiple | Hyperscaler infrastructure with different business mix |
CoreWeave's Q2 revenue of $2.58 billion dwarfs Nebius's $582 million, though both are capital-intensive and neither provides a clean, mature free-cash-flow valuation anchor.
A simple comparison of revenue multiples would suggest Nebius is relatively expensive. But CoreWeave has a substantially different capital structure and more mature operating footprint, so I would not apply its multiple mechanically.
The more relevant question is: How much revenue, at what margins, can Nebius produce using the capital shareholders and creditors are contributing?
I use a three-year scenario framework that estimates the business's enterprise value at the end of 2029, subtracts estimated net debt, divides by projected diluted shares, and discounts the result back three years at 12%.
This is an independent model, not company guidance. The scenario forecasts intentionally use broad ranges to reflect uncertainty in demand, pricing, hardware economics, and dilution.
| Scenario | Key Assumptions | Fair Value | Probability | Probability Rationale | Return From Price Snapshot |
|---|---|---|---|---|---|
| Bear | 2029 revenue $12B; 2.5× exit EV/sales; $12B net debt; 350M diluted shares | $36.61 | 25% | Weaker utilization and pricing, unfavorable financing, more dilution | -75.9% |
| Base | 2029 revenue $21B; 4.5× exit EV/sales; $9B net debt; 335M diluted shares | $181.66 | 50% | Company achieves rapid growth while remaining capital-intensive | +19.8% |
| Bull | 2029 revenue $32B; 5.5× exit EV/sales; $5B net debt; 325M diluted shares | $374.51 | 25% | Large efficient AI cloud provider; high utilization, strong prices, differentiated platform | +147.0% |
| Probability-Weighted | Weighted scenario outcomes | $193.61 | 100% | 25% bear / 50% base / 25% bull | +27.7% |
The base case assumes FY2029 revenue of $21 billion, an approximately 87% compound annual growth rate from the midpoint of 2026 revenue guidance. This is aggressive, but materially below the most optimistic interpretation of Nebius's infrastructure expansion.
The bull case assumes the company becomes a large, efficient AI cloud provider with strong utilization, attractive prices, and meaningful platform differentiation. The bear case assumes weaker utilization and pricing, unfavorable financing requirements, and more dilution.
These are valuation outcomes, not forecast share-price paths or guaranteed investment returns. The model is highly sensitive to exit multiples, net debt, and share count.
The stock becomes more interesting below approximately $155 under the current methodology, although even there I would want to reassess fundamentals rather than buy automatically.
Founder and CEO Arkady Volozh brings substantial experience building large-scale technology operations. Nebius's ability to win contracts with Meta and Microsoft, secure financing, and grow revenue so quickly supports a favorable assessment of management's execution ability.
Management has also been unusually ambitious. FY2026 capital expenditure guidance of $20–25 billion represents a staggering expansion compared with 2025 revenue of approximately $530 million.
The question is whether management is investing during a temporary supply shortage or establishing infrastructure capable of generating attractive cash returns for many years.
There is an important strategic tension:
Management's emphasis on prepayments, asset-backed lending, and less capital-intensive services is encouraging. Repeated equity and convertible issuance demonstrates that financing needs remain significant.
Strong commercial execution; capital-allocation outcomes remain unproven.
Analyzed call: Q2 2026 — 2026-08-12 · Research snapshot: 2026-10-21 (comparison refresh; fundamental review preserved from 2026-10-14)
The August 12 Q2 earnings call contained several exchanges that directly test the investment thesis. These are the material issues raised across the call, including management answers and unresolved concerns.
| Analyst / Firm | Core Question | Management Response | Research Read-Through |
|---|---|---|---|
| Ryan Lantz / Morgan Stanley | Will the unresolved Vineland hearing delay capacity? | Management maintained the construction timeline and described its community-engagement work. | Local approvals remain a real execution risk |
| Investor-submitted questions | Why is Nebius winning billion-dollar-plus deals? | Management cited proof-of-concept deployments, capability, scale and customer experience. | Encouraging evidence of demand; still need repeatable economics |
| Alex Duval / Goldman Sachs | How quickly can contracted power become operating capacity? | Management described a multiyear development process and alternative approaches to supplying power. | Contracted GW should not be confused with deployed GW |
| Investor-submitted strategy question | Why leave capacity uncontracted? | Management said leaving capacity uncontracted preserves flexibility to capture higher short-term pricing. | Creates potential upside and utilization risk |
| Arsenije Matovic / Wolfe Research | How will the company fund expansion without excessive dilution? | Management identified customer prepayments, asset-backed debt and other financing instruments as funding sources. | Central thesis uncertainty, especially after subsequent convert issuance |
| Tyler Radke / Citi | How will 2027 capacity be split between multiyear and premium short-term contracts? | Management said it intends to optimize pricing, payment terms and contract duration. | Peak pricing may not be sustainable across the fleet |
| James Kisner / Water Tower Research | Could competitors, including xAI, pressure pricing? | Management emphasized market growth and its specialized positioning. | Competitive pricing risk remains unresolved |
| Rob Oliver / Baird | How differentiated is Token Factory for inference? | Management described model availability, performance and developer capabilities. | Technology is promising; standalone economics need proof |
| Rob Oliver / Baird | How economically attractive is the asset-light model? | Management reported early partner interest but said the asset-light offering is still early. | Do not yet capitalize this as a mature high-margin business |
| Ryan Lantz / Morgan Stanley | What should investors expect for 2027 growth and guidance? | Management deferred formal 2027 guidance until later in 2026. | Investors are extrapolating beyond formal guidance |
| Brett Knoblauch / Cantor | Is 800 MW–1 GW connected by year-end still achievable? | Management maintained the connected-capacity guidance and said monetization may follow through H1 2027. | Timing gap between power connection and revenue |
| Stefan Slowinski / BNP Paribas | Does premium pricing reflect Vera Rubin availability? | Management described upcoming next-generation GPU deployments. | Pricing and hardware rollout both carry technology-cycle risk |
| Investor-submitted pricing questions | Are auction prices and short-term deals representative? | Management described strong initial pricing and take-or-pay structures. | Small premium deals should not set fleet-wide forecasts |
| Investor-submitted financing and inference questions | Can 2027 capex be funded, and how fast is inference adoption progressing? | Management emphasized prepayments, financing options and growing inference-product adoption. | Both need quantified, sustained results |
The most important issue was not whether Nebius can find customers. Management's disclosures provide substantial evidence of demand.
The bigger questions concerned delivery timing, capital availability, customer pricing, and whether newly introduced business models can become meaningful recurring contributors.
Management was persuasive on the opportunity and demonstrated operational progress, but the call did not eliminate financing uncertainty or provide a fully underwritten bridge from infrastructure capex to long-term per-share free cash flow.
| Source | Type | Date / Period | What It Supports |
|---|---|---|---|
| Nebius Q2 2026 earnings release and financial statements | Regulatory filing / company results | 2026-08-12 | Revenue, earnings, cash flow, balance sheet, SBC and shares |
| Nebius Q2 2026 shareholder letter | Company IR | August 2026 | Operating strategy, capacity, contracts and economics |
| Nebius FY2025 results | Regulatory filing / company results | FY2025 | Historical financials |
| Q2 2026 earnings call transcript | Transcript | 2026-08-12 | Management guidance, analyst questions and risks |
| Nebius convertible notes closing | Company IR | 2026-08-24 | Financing and dilution |
| Nebius–Meta agreement | Company IR | 2026-03-16 | Customer commitments |
| Nebius–Microsoft agreement | Company IR | 2025-09-08 | Contract economics and credibility |
| CoreWeave Q2 2026 results | Regulatory filing / company results | Q2 2026 | Direct competitor comparison |
| Nebius market statistics | Market data | 2026-10-14 | Intraday valuation snapshot |
| Nebius–Palantir partnership | Company IR | September 2026 | Strategic development |
| Inferize acquisition | Company IR | 2026-10-01 | Managed inference strategy |
| NBIS common regular-close snapshot | Market data | 2026-10-20 16:00 EDT | Refreshed price, scenario returns and normalized Opportunity Score / shared rank |
Company: Nebius Group
Ticker: NBIS
Exchange: NASDAQ
Sector / Industry: Technology / AI Cloud Infrastructure
Template version: 2.7
Research date: 2026-10-21 (comparison refresh; fundamental review preserved from 2026-10-14)
Valuation / score price snapshot: $151.64 — 2026-11-03 regular-session close, America/New_York
Tracking tier: On the Radar
Owned: No
Primary operating period reviewed: Q2 2026 / H1 2026
Historical period reviewed: completed FY2024-FY2025; Q1 FY2026, Q2 FY2026, H1 FY2026
Latest investor presentation reviewed: Q2 2026 shareholder letter — August 2026
Latest earnings call reviewed: Q2 2026 — 2026-08-12
Latest post-earnings management commentary reviewed: September 2026 Palantir partnership, October 1, 2026 Inferize acquisition, and August 24 financing announcements
Price snapshot date: 2026-11-03
Price snapshot time / timezone: Regular-session close, America/New_York (including exchange half-days)
Competitor / analyst-Q&A audit: 2026-10-28; original operating, price and valuation snapshots retained.
Company: Nebius Group
Ticker: NBIS
Exchange: NASDAQ
Sector / Industry: Technology / AI Cloud Infrastructure
Template version: 2.7
Research date: 2026-10-21 (comparison refresh; fundamental review preserved from 2026-10-14)
Valuation / score price snapshot: $151.64 — 2026-11-03 regular-session close, America/New_York
Tracking tier: On the Radar
Owned: No
Primary operating period reviewed: Q2 2026 / H1 2026
Historical period reviewed: completed FY2024-FY2025; Q1 FY2026, Q2 FY2026, H1 FY2026
Latest investor presentation reviewed: Q2 2026 shareholder letter — August 2026
Latest earnings call reviewed: Q2 2026 — 2026-08-12
Latest post-earnings management commentary reviewed: September 2026 Palantir partnership, October 1, 2026 Inferize acquisition, and August 24 financing announcements
Price snapshot date: 2026-11-03
Price snapshot time / timezone: Regular-session close, America/New_York (including exchange half-days)
Competitor / analyst-Q&A audit: 2026-10-28; original operating, price and valuation snapshots retained.
Current Review: 2026-10-28
Previous Review: 2026-10-21 (comparison refresh; fundamental review preserved from 2026-10-14)
Update Trigger: Competitor relevance and analyst-Q&A correction
| Area | Previous View | Current View | Investment Impact |
|---|---|---|---|
| Competition | Preceding operating and comparable-company classifications | Company-specific rivals reviewed; relevant comparisons retained | Corrects the competitive comparison; does not by itself re-underwrite the valuation |
| Analyst questions | Prior source-linked Q&A summaries | Complete questions and sentence-form management answers; all 14 existing rows retained | Improves readability while preserving the recorded exchange meaning |
| Valuation and scores | Approved operating/price snapshots, scenarios, scores and decisions | Preserved | No price refresh, rerank or change to valuation assumptions in this correction pass |
Thesis: No new fundamental thesis determination is made in this correction pass. The prior investment decision is preserved; the corrected competitive scope should govern the next substantive review.
| Metric | Current View |
|---|---|
| Price Snapshot | $151.64 — 2026-11-03 regular-session close, America/New_York |
| Rating / Zone Status | Buy |
| Bear / Base / Bull Fair Value | $36.61 / $181.66 / $374.51 |
| Scenario Probabilities | 25% / 50% / 25% |
| Probability-Weighted Fair Value | $193.61 |
| Probability-Weighted Expected Return | +27.7% |
| Bear-Case Return | -75.9% |
| Strong Buy Zone | ≤ $40.27 |
| Buy Zone | >$40.27–$154.89 |
| Avoid Chasing Above | > $215.12 |
| Quality Score / Tier | 60.7 / Tier C |
| Opportunity Score / Rank | 59.2 / #26 of 76 |
| Risk Level | High |
| Tracking / Ownership | On the Radar / Not Owned |
Decision rationale: Nebius is rapidly becoming a substantial AI infrastructure provider, but its capitalization already reflects years of extremely successful execution. At approximately $183 per share, the modeled probability-weighted upside is only about 5.6%, with severe downside if the business or financing falls short.
The unusually low Strong Buy threshold results from the repository's conservative formula requiring substantial protection against the modeled bear case. It is not a prediction that Nebius will trade at $40.
Nebius has three important advantages working in its favor.
First, AI infrastructure demand is moving faster than traditional data-center development cycles. A company capable of securing electricity, constructing GPU clusters, and putting them into service quickly can monetize scarce capacity before competitors finish building their facilities.
Second, Nebius has been securing contracts from some of the world's largest technology companies. Microsoft and Meta provide revenue visibility, financing credibility, and evidence that Nebius can operate infrastructure at a level acceptable to highly sophisticated customers.
Third, Nebius is attempting to become more than a supplier of rented NVIDIA GPUs. Its managed inference, orchestration, networking, storage, and developer tools could create differentiated services and customer relationships that survive beyond individual GPU generations.
However, none of these advantages guarantees an attractive shareholder return.
If Nebius keeps reinvesting nearly all available capital in hardware that depreciates rapidly, future revenue could grow spectacularly while free cash flow and returns on invested capital remain disappointing.
The optimistic interpretation is that investors underestimate how much revenue Nebius can generate from contracted and upcoming infrastructure, and how quickly the company can transition into higher-margin software and inference services.
The pessimistic interpretation is that investors are valuing Nebius like a scalable cloud software company before it has proven the economics of owning, financing, and continuously replacing its expensive infrastructure.
I lean toward the second concern at today's price. The business opportunity is credible, but its capitalization already reflects years of extremely successful execution.
Nebius emerged from the restructuring of Yandex's former international parent. Founder and CEO Arkady Volozh is now building an AI-focused cloud infrastructure company headquartered in Amsterdam, with operations and development centers across multiple regions.
Its commercial activities can be divided into the following groups.
| Business | What it does | Economic importance |
|---|---|---|
| Nebius AI Cloud | GPU clusters, networking, storage, compute orchestration, training and inference infrastructure | Dominant revenue and value driver |
| Dedicated AI infrastructure | Longer-term capacity contracted to large technology customers | Provides revenue visibility and supports project financing |
| Token Factory and AI services | Managed inference and related developer services | Potential to improve differentiation and margins |
| Asset-light partnerships | Operates or monetizes partner-owned capacity through Nebius technology | May allow growth without funding every dollar of hardware itself |
| Avride | Autonomous vehicles and delivery robots | Early-stage optionality, currently a group-level investment burden |
| TripleTen | Technology training and reskilling | Smaller, non-core business |
| Equity investments | Stakes including ClickHouse and Toloka | Additional value, but not the main investment thesis |
In Q2 2026, AI cloud revenue was approximately $574.9 million, or 98% of total group revenue. Nebius should therefore be valued primarily as an AI infrastructure and cloud platform, not as a diversified technology holding company.
Customers contract AI computing capacity — hyperscalers, AI labs, developers, and enterprises.
Nebius secures power, buildings, GPUs, and financing, using upfront investment, customer prepayments, debt, and equity.
Nebius provides training, inference, networking, orchestration, storage, and related compute and software services.
Revenue is earned over the contract term, less electricity, operations, equipment depreciation, interest, reinvestment, and dilution.
The final step is the most important. Customer cash received upfront and high adjusted EBITDA do not necessarily represent distributable profits.
| Customer | Announced agreement | Investment implications |
|---|---|---|
| Microsoft | Multiyear AI infrastructure agreement announced September 2025, with previously disclosed value up to approximately $17.4 billion | Establishes hyperscaler credibility, but creates large delivery commitments |
| Meta | $12 billion dedicated capacity commitment over five years beginning in 2027 | Supports next-generation infrastructure deployment |
| Meta — additional capacity | Up to $15 billion in additional capacity purchases under a backstop arrangement | Provides a utilization floor for selected capacity; not the same as $15 billion of guaranteed incremental revenue from customers |
The distinction in the Meta agreement matters. The company intends to sell some capacity to third-party customers, with Meta purchasing remaining eligible capacity. Therefore, describing the entire $27 billion as a simple fixed Meta revenue commitment would overstate the certainty and structure of the deal.
Recent strategic expansion also includes a September 2026 partnership with Palantir for sovereign AI and an October 1 acquisition of Inferize to improve production inference efficiency. These strengthen the platform strategy, but they do not yet establish a large, separately measurable source of recurring high-margin revenue.
Nebius already has real customers and significant operating scale. The unanswered question is whether its cloud technology can eventually become economically more valuable than the underlying GPUs.
The opportunity consists of several overlapping markets: AI training infrastructure, production inference, GPU cloud services, and associated data-center power and networking capacity.
I would avoid applying a single enormous generative-AI TAM to Nebius. Spending on AI models, applications, hardware, electricity, and services cannot all be treated as directly addressable Nebius revenue.
The most relevant demand drivers are the growing compute requirements of large models, the movement of AI systems into production, and the need for specialized GPU clusters with high-bandwidth networking. Inference may ultimately become a more durable demand source than sporadic frontier-model training runs, but it is also vulnerable to price competition and major efficiency improvements.
A consistent, independently comparable market-share dataset for specialized AI cloud infrastructure is not available. Reported revenue, contracted capacity, backlog, and installed GPU power use different definitions across companies.
Rather than invent market-share estimates, the following provides a more useful scale comparison.
| Company / Platform | Primary Position | Relative Scale | Growth Direction | Investment Relevance |
|---|---|---|---|---|
| Nebius | AI-native cloud and hyperscaler capacity | Rapidly scaling; $582M Q2 revenue | Exceptional | Capital and execution constraints |
| CoreWeave | Specialist GPU cloud | Much larger; $2.58B Q2 revenue | Strong | Leverage and financing burden |
| AWS, Azure, Google Cloud | Global hyperscalers | Vastly larger diversified platforms | Strong | Different economics and internal priorities |
| Oracle Cloud Infrastructure | Enterprise and hyperscaler cloud infrastructure | Substantial established platform | Expanding AI | Diversified exposure |
| Lambda and other private neoclouds | GPU-focused specialist providers | Private financial visibility limited | Growing | Access to capital, power, and customers |
CoreWeave's Q2 2026 revenue was approximately 4.4 times Nebius's revenue for the same quarter. That is a meaningful measure of current commercial scale, although their contract structures, hardware age, utilization, and accounting presentations differ.
Nebius has been targeting approximately 5 GW of contracted power by the end of 2026 and 800 MW to 1 GW of connected power by that date. Those are materially different operating milestones.
Management cautioned that connected power can take additional months to generate revenue. The 2026 buildout should therefore not be valued as though all committed power were already productive.
The addressable market is large enough to support much more growth. The limiting factors are likely to be electricity, execution, funding, and competition rather than a lack of potential workloads.
Nebius's disclosed contract economics are attractive on the surface.
Management reported that roughly 70% of Q2 signed deals included upfront prepayments and indicated that, for some deals, those payments could cover 50%–60% of the associated capital expenditure. It also highlighted selected contracts exceeding $20 million of annual contract value per megawatt and estimated hardware payback of under two years.
These disclosures make it easier to understand why lenders are willing to finance portions of the infrastructure.
But the figures need careful interpretation. A payback estimate is not the same as an audited return on invested capital. It may not capture every financing expense, infrastructure replacement cycle, overhead charge, or customer-renewal risk. Selected high-value contracts do not establish economics across the installed fleet.
The most decision-useful customer measures would be:
| Metric | Why it matters | Current visibility |
|---|---|---|
| Revenue per active MW | Shows capacity monetization | Selected deal disclosures; no standardized fleet-wide metric |
| Utilization | Determines fixed-asset productivity | Management reports strong demand; comprehensive utilization history unavailable |
| Customer prepayments | Reduces external financing requirements | Material and growing |
| Renewal economics | Tests durability after the initial contract | Not yet established over multiple GPU cycles |
| Inference revenue and margin | Tests software-platform differentiation | Insufficiently separated |
| Customer concentration | Measures reliance on major counterparties | Large strategic contracts create significant exposure |
Initial contract economics support the growth thesis. Sustainable fleet-wide capital returns and renewals remain unproven.
| Competitor | Market Share | Positioning | Advantages vs. Company | Disadvantages vs. Company | Current Direction |
|---|---|---|---|---|---|
| CoreWeave | Not comparable | Closest public AI-cloud competitor | Greater current scale, operating history, and infrastructure utilization | Nebius has potentially less entrenched debt burden and growing proprietary cloud capabilities | Scaling |
| AWS | Not comparable | General-purpose and AI cloud | Capital, global reach, enterprise distribution, ecosystem | Nebius offers more specialized GPU infrastructure and responsiveness | Expanding |
| Microsoft Azure | Not comparable | Hyperscaler cloud and AI services | Enterprise lock-in, capital, software integration | Independent infrastructure supply where Azure needs additional capacity | Expanding |
| Google Cloud | Not comparable | Cloud, TPUs, AI platforms | Vertical integration of hardware, models, and services | Nebius targets NVIDIA-centric workloads and customer choice | Expanding |
| OCI | Not comparable | Cloud infrastructure | Data-center expertise, hyperscaler resources | Nebius has AI-native platform focus | Expanding |
| Lambda and other specialists | Not comparable | Dedicated GPUs, AI cloud | Specialized offerings and targeted customer relationships | Nebius has scale of contracted deployments and integrated service stack | Growing |
I would classify Nebius's moat as developing, not yet durable, with a strengthening direction.
Its strongest potential advantages are execution capability, early access to power and sites, engineering expertise in dense GPU clusters, and the operational knowledge required to make extremely large AI systems run reliably.
Power access can provide a temporary competitive advantage. But the value of that advantage depends on how scarce power remains in the locations customers need.
Nebius's software may eventually create switching costs through integrated storage, networking, deployment, inference, and developer workflows. Today, however, most of its economic value remains closely tied to compute availability.
There is also a structural concern: Nebius is both a supplier to and a competitor of hyperscalers. Those companies may use external providers while capacity is scarce, then gradually bring more workloads onto their own infrastructure when their economics improve.
Moat direction: Strengthening, but not yet durable.
Nebius has credible emerging advantages, but I would not yet assign it the kind of durable moat premium associated with mature cloud platforms.
Standardized period history (reviewed 2026-11-07): Completed issuer fiscal years only, plus actually disclosed standalone quarters and cumulative YTD for the latest open FY. Currency USD, flows in millions, with fiscal dates defined by the issuer. Values marked derived come from disclosed period differences/sums, not TTM. Missing = N/D (not zero). Q1, Q2 and H1 FY2026 results are preserved from the existing filing-grounded report, with standardized fiscal-year labels. FCF excludes major capital investments and must not be confused with operating inflows from customer prepayments.
| Period | Revenue | Growth | Gross Margin | Operating Margin | FCF | FCF Margin | SBC | Diluted Shares | Cash | Debt |
|---|---|---|---|---|---|---|---|---|---|---|
| FY2024 | $91.5M | — | 52.2% | −436.7% | −$1,077.4M | −1,177.5% | $54.5M | Not specified | $2,434.7M | $6.1M |
| FY2025 | $529.8M | 479% | 68.6% | −112.5% | −$3,664.1M | −691.6% | $83.2M | Not specified | $3,678.1M | $4,127.7M |
| Q1 FY2026 | $399.0M | 684% | 74.0% | −32.1% | −$214.9M | −53.9% | $35.3M | Not specified | Not specified | Not specified |
| Q2 FY2026 | $582.3M | 454% | 77.1% | −30.2% | −$3,411.3M | −585.8% | $102.5M | 280.4M | $8,042.1M | $8,545.7M |
| H1 FY2026 | $981.3M | 529% | 75.8% | −31.0% | −$3,626.2M | −369.5% | $137.8M | Not specified | $8,042.1M | $8,545.7M |
Gross margins shown before depreciation and amortization because Nebius reports those costs separately. They are not directly comparable with peers that include infrastructure depreciation in cost of revenue. Q2 diluted weighted-average shares: 280.4 million; June 30 issued and outstanding shares: 271.9 million.
| Period | Revenue (USD millions) |
|---|---|
| Q3 2025 | $145.2 |
| Q4 2025 | $227.7 |
| Q1 2026 | $399.0 |
| Q2 2026 | $582.3 |
Nebius generated approximately $4.50 billion in operating cash flow during the first half of 2026, but that included a $4.40 billion positive change in deferred revenue, principally associated with customer funding.
That is commercially valuable. Receiving money before providing service lowers financing requirements.
However, that money comes with future obligations. Nebius still needs to deploy and operate the infrastructure for which customers have paid. Its $5.98 billion deferred revenue liability at June 30 is a reminder that received cash and earned revenue are different.
Free cash flow remained negative by approximately $3.63 billion in the first half, even after the benefit of those customer payments.
Nebius reported $259.7 million in Q2 depreciation and amortization against revenue of $582.3 million.
That is a substantial expense, not merely accounting noise. GPU hardware has finite economic usefulness. Rapid advances in performance per watt and per dollar can force service providers to upgrade hardware well before their physical assets stop working.
If all group depreciation and amortization were allocated against revenue and directly reported cost of revenue, the resulting gross contribution would be approximately 32.5% rather than 77.1%. This is an illustrative calculation, not the company's GAAP gross margin, but it shows why ignoring depreciation is misleading.
There is genuine financial support for the expansion: customer prepayments, hyperscaler contracts, equity proceeds, and GPU-backed lending.
There is also a large and growing capital requirement.
At June 30, Nebius reported $8.55 billion in debt and $8.04 billion in unrestricted cash. Those numbers do not incorporate the company's entire subsequent financing activity.
On August 24, Nebius closed approximately $5.75 billion in new convertible debt, while separately agreeing to exchange $800 million of older convertible debt for approximately 15.8 million Class A shares. The debt exchange lowers obligations but dilutes existing owners. The new notes provide funding while potentially creating additional dilution or refinancing requirements later.
During Q2, the company had already sold approximately 12.7 million shares through its at-the-market program, raising about $2.8 billion gross. Its earlier NVIDIA financing also introduced pre-funded warrants representing additional potential shares.
This is central to per-share valuation: a rapidly growing company can finance expansion successfully while ownership of its future profits becomes spread across more shares.
Financial direction: Improving operating scale and adjusted EBITDA, but high financing and free-cash-flow risk.
Founder and CEO Arkady Volozh brings substantial experience building large-scale technology operations. Nebius's ability to win contracts with Meta and Microsoft, secure financing, and grow revenue so quickly supports a favorable assessment of management's execution ability.
Management has also been unusually ambitious. FY2026 capital expenditure guidance of $20–25 billion represents a staggering expansion compared with 2025 revenue of approximately $530 million.
The question is whether management is investing during a temporary supply shortage or establishing infrastructure capable of generating attractive cash returns for many years.
There is an important strategic tension:
Management's emphasis on prepayments, asset-backed lending, and less capital-intensive services is encouraging. Repeated equity and convertible issuance demonstrates that financing needs remain significant.
Strong commercial execution; capital-allocation outcomes remain unproven.
| Metric | Value |
|---|---|
| Price Snapshot | $151.64 — 2026-11-03 regular-session close, America/New_York |
| Market Capitalization | ~$50.3B |
| Enterprise Value | ~$52.4B |
| EV / trailing revenue | ~39.3× |
| EV / FY2026 midpoint revenue | ~16.4× |
| FY2026 revenue guidance | $3.0B–$3.4B |
| FY2026 year-end ARR guidance | $7B–$9B |
| FY2026 capex guidance | $20B–$25B |
Quoted market capitalization and enterprise value are vendor figures and may not fully incorporate recent post-quarter convertible exchanges, share issuance, and financing. They should be treated as indicative rather than as a reconciled pro-forma capitalization.
Nebius's June annualized run-rate revenue was approximately $3 billion. Management's target of $7–9 billion by year-end is promising, but annualized exit revenue is not the same as full-year recognized revenue. The market is pricing a steep future revenue ramp.
Only the price row is refreshed here. Other quoted valuation statistics retain their earlier source dates and definitions.
| Company | Relevance | Public / Private | EV / Trailing Revenue | EV / FY2026 Revenue | Growth / Margin / Risk Context |
|---|---|---|---|---|---|
| Nebius | Direct AI-cloud operator | Public | ~39× | ~16× guided midpoint | Smaller base, extremely fast growth, large capex |
| CoreWeave | Closest public operating peer | Public | Roughly high-single-digit multiple | Not specified in approved analysis | Greater revenue scale; heavy debt and interest burden |
| Lambda | Direct operating peer | Private | N/A — private | N/A — private | Specialized GPU cloud, limited public financial disclosure |
| AWS / Azure / Google Cloud | Secondary economic anchors | Diversified public parents | N/A independently traded | N/A independently traded | Mature distribution, infrastructure scale, diversified earnings |
| OCI | Secondary cloud infrastructure anchor | Diversified public parent | No clean standalone comparable multiple | No clean standalone comparable multiple | Hyperscaler infrastructure with different business mix |
CoreWeave's Q2 revenue of $2.58 billion dwarfs Nebius's $582 million, though both are capital-intensive and neither provides a clean, mature free-cash-flow valuation anchor.
A simple comparison of revenue multiples would suggest Nebius is relatively expensive. But CoreWeave has a substantially different capital structure and more mature operating footprint, so I would not apply its multiple mechanically.
The more relevant question is: How much revenue, at what margins, can Nebius produce using the capital shareholders and creditors are contributing?
I use a three-year scenario framework that estimates the business's enterprise value at the end of 2029, subtracts estimated net debt, divides by projected diluted shares, and discounts the result back three years at 12%.
This is an independent model, not company guidance. The scenario forecasts intentionally use broad ranges to reflect uncertainty in demand, pricing, hardware economics, and dilution.
| Scenario | Key Assumptions | Fair Value | Probability | Probability Rationale | Return From Price Snapshot |
|---|---|---|---|---|---|
| Bear | 2029 revenue $12B; 2.5× exit EV/sales; $12B net debt; 350M diluted shares | $36.61 | 25% | Weaker utilization and pricing, unfavorable financing, more dilution | -75.9% |
| Base | 2029 revenue $21B; 4.5× exit EV/sales; $9B net debt; 335M diluted shares | $181.66 | 50% | Company achieves rapid growth while remaining capital-intensive | +19.8% |
| Bull | 2029 revenue $32B; 5.5× exit EV/sales; $5B net debt; 325M diluted shares | $374.51 | 25% | Large efficient AI cloud provider; high utilization, strong prices, differentiated platform | +147.0% |
| Probability-Weighted | Weighted scenario outcomes | $193.61 | 100% | 25% bear / 50% base / 25% bull | +27.7% |
The base case assumes FY2029 revenue of $21 billion, an approximately 87% compound annual growth rate from the midpoint of 2026 revenue guidance. This is aggressive, but materially below the most optimistic interpretation of Nebius's infrastructure expansion.
The bull case assumes the company becomes a large, efficient AI cloud provider with strong utilization, attractive prices, and meaningful platform differentiation. The bear case assumes weaker utilization and pricing, unfavorable financing requirements, and more dilution.
These are valuation outcomes, not forecast share-price paths or guaranteed investment returns. The model is highly sensitive to exit multiples, net debt, and share count.
The stock becomes more interesting below approximately $155 under the current methodology, although even there I would want to reassess fundamentals rather than buy automatically.
The August 12 Q2 earnings call contained several exchanges that directly test the investment thesis. These are the material issues raised across the call, including management answers and unresolved concerns.
| Analyst / Firm | Core Question | Management Response | Research Read-Through |
|---|---|---|---|
| Ryan Lantz / Morgan Stanley | Will the unresolved Vineland hearing delay capacity? | Management maintained the construction timeline and described its community-engagement work. | Local approvals remain a real execution risk |
| Investor-submitted questions | Why is Nebius winning billion-dollar-plus deals? | Management cited proof-of-concept deployments, capability, scale and customer experience. | Encouraging evidence of demand; still need repeatable economics |
| Alex Duval / Goldman Sachs | How quickly can contracted power become operating capacity? | Management described a multiyear development process and alternative approaches to supplying power. | Contracted GW should not be confused with deployed GW |
| Investor-submitted strategy question | Why leave capacity uncontracted? | Management said leaving capacity uncontracted preserves flexibility to capture higher short-term pricing. | Creates potential upside and utilization risk |
| Arsenije Matovic / Wolfe Research | How will the company fund expansion without excessive dilution? | Management identified customer prepayments, asset-backed debt and other financing instruments as funding sources. | Central thesis uncertainty, especially after subsequent convert issuance |
| Tyler Radke / Citi | How will 2027 capacity be split between multiyear and premium short-term contracts? | Management said it intends to optimize pricing, payment terms and contract duration. | Peak pricing may not be sustainable across the fleet |
| James Kisner / Water Tower Research | Could competitors, including xAI, pressure pricing? | Management emphasized market growth and its specialized positioning. | Competitive pricing risk remains unresolved |
| Rob Oliver / Baird | How differentiated is Token Factory for inference? | Management described model availability, performance and developer capabilities. | Technology is promising; standalone economics need proof |
| Rob Oliver / Baird | How economically attractive is the asset-light model? | Management reported early partner interest but said the asset-light offering is still early. | Do not yet capitalize this as a mature high-margin business |
| Ryan Lantz / Morgan Stanley | What should investors expect for 2027 growth and guidance? | Management deferred formal 2027 guidance until later in 2026. | Investors are extrapolating beyond formal guidance |
| Brett Knoblauch / Cantor | Is 800 MW–1 GW connected by year-end still achievable? | Management maintained the connected-capacity guidance and said monetization may follow through H1 2027. | Timing gap between power connection and revenue |
| Stefan Slowinski / BNP Paribas | Does premium pricing reflect Vera Rubin availability? | Management described upcoming next-generation GPU deployments. | Pricing and hardware rollout both carry technology-cycle risk |
| Investor-submitted pricing questions | Are auction prices and short-term deals representative? | Management described strong initial pricing and take-or-pay structures. | Small premium deals should not set fleet-wide forecasts |
| Investor-submitted financing and inference questions | Can 2027 capex be funded, and how fast is inference adoption progressing? | Management emphasized prepayments, financing options and growing inference-product adoption. | Both need quantified, sustained results |
The most important issue was not whether Nebius can find customers. Management's disclosures provide substantial evidence of demand.
The bigger questions concerned delivery timing, capital availability, customer pricing, and whether newly introduced business models can become meaningful recurring contributors.
Management was persuasive on the opportunity and demonstrated operational progress, but the call did not eliminate financing uncertainty or provide a fully underwritten bridge from infrastructure capex to long-term per-share free cash flow.
| Driver | Current State | Positive Trigger / Catalyst | Negative Trigger / Falsifier | What It Changes |
|---|---|---|---|---|
| Revenue growth | Exceptional | 2026 guidance met or exceeded; strong 2027 outlook | Material shortfall or delayed recognition | Revenue forecasts and multiples |
| Power and capacity | Major buildout | 800MW–1GW connected target delivered on schedule | Significant permitting or commissioning delays | Timing and execution risk |
| Capital efficiency | Large customer advances | Prepayments and lending reduce equity financing | Dilution or refinancing requirements accelerate | Per-share valuation |
| Pricing | Selected premium contracts | Strong pricing persists across broader installed fleet | Falling revenue per active MW | Margins and terminal valuation |
| Inference services | Growing offering | Measurable high-margin recurring revenue | Remains a small undifferentiated GPU service | Moat and Quality Score |
| Depreciation | Significant | Longer economically useful GPU lives and attractive renewals | Faster hardware obsolescence | ROIC and free cash flow |
| Customer diversification | Hyperscalers plus AI cloud | More independent enterprise customers | Growing reliance on a few large contracts | Revenue durability |
| Competition | Intensifying | Nebius gains durable share | Hyperscaler insourcing or price competition | Long-term growth and multiples |
The next expected quarterly earnings date is approximately November 10, 2026, although that should be confirmed through Nebius investor relations. The two disclosures I would prioritize are the updated 2027 outlook and the complete funding plan for the remaining capacity buildout.
| Component | Score | Weight | Weighted Contribution | Evidence |
|---|---|---|---|---|
| Business durability and economics | 62 | 25% | 15.5 | Large contracted demand, but unproven sustainable ROIC |
| Moat and competitive position | 48 | 25% | 12.0 | Growing execution capability; limited proven durable differentiation |
| Growth runway and durability | 91 | 20% | 18.2 | Exceptional demand and substantial infrastructure opportunity |
| Financial quality and balance sheet | 42 | 15% | 6.3 | Heavy capex, financing needs, dilution |
| Management, governance and capital allocation | 58 | 15% | 8.7 | Strong scaling execution; long-term capital outcomes uncertain |
| Quality Score | 60.7 | 100% | 60.7 | Tier C |
Although the Quality Score is above the speculative threshold, Nebius remains a high-risk investment because of the unusually wide range of plausible financial outcomes.
| Component | Raw Input | Normalized Score | Weight | Weighted Contribution |
|---|---|---|---|---|
| Quality | 60.7 | 60.7 | 30% | 18.2 |
| Probability-weighted upside | +27.7% | 100.0 | 35% | 35.0 |
| Bear-case downside | -75.9% | 0.0 | 20% | 0.0 |
| Risk / valuation confidence | 40 | 40.0 | 15% | 6.0 |
| Opportunity Score | 59.2 | 100% | 59.2 |
| Timestamp | Price | Rating | Zone Status | Buy Zone | Strong Buy Zone | Avoid Chasing Above | Base FV | P/W FV | P/W Return | Quality | Opportunity / Rank | Change / Rationale |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026-10-14 14:02 ET | $183.41 | Hold / Watch | Hold / Watch | >$40.27 to $154.89 | <=$40.27 | >$215.12 | $181.66 | $193.61 | +5.6% | 60.7 / Tier C | 45.6 / Unranked | Initial approved analysis; preserved scenario assumptions, rating, scores, and zones |
| 2026-10-21 (America/New_York) | $171.45 | Hold / Watch | Hold / Watch | >$40.27–$154.89 | ≤ $40.27 | > $215.12 | $181.66 | $193.61 | +12.9% | 60.7 | 50.8 / #34 | Common close and shared-universe rank; prior fundamental assumptions and zones preserved |
Final investment decision: Nebius has demonstrated a level of commercial traction that separates it from many speculative AI infrastructure stories. The Microsoft and Meta agreements, rising revenue, and improving adjusted EBITDA all suggest that a substantial business is being built.
Nevertheless, buying a rapidly scaling AI infrastructure company requires a different valuation mindset from buying a capital-light software company.
The assets are expensive, they depreciate, they require large amounts of financing, and their long-term cash returns depend on continued demand and pricing. Today's revenue growth is evidence of commercial success, but not yet proof of durable per-share value creation.
At approximately $183 per share, my rating is Hold / Watch. I would want either a substantially more attractive entry valuation or stronger evidence of capital efficiency before raising that assessment.
The single most important thesis question for the next several quarters is:
Can Nebius turn its rapidly increasing contracted and deployed capacity into sustainably attractive free cash flow after equipment replacement and financing costs, without excessive shareholder dilution?
If that answer becomes demonstrably positive, the quality and valuation of the company could improve substantially.
| Source | Type | Date / Period | What It Supports |
|---|---|---|---|
| Nebius Q2 2026 earnings release and financial statements | Regulatory filing / company results | 2026-08-12 | Revenue, earnings, cash flow, balance sheet, SBC and shares |
| Nebius Q2 2026 shareholder letter | Company IR | August 2026 | Operating strategy, capacity, contracts and economics |
| Nebius FY2025 results | Regulatory filing / company results | FY2025 | Historical financials |
| Q2 2026 earnings call transcript | Transcript | 2026-08-12 | Management guidance, analyst questions and risks |
| Nebius convertible notes closing | Company IR | 2026-08-24 | Financing and dilution |
| Nebius–Meta agreement | Company IR | 2026-03-16 | Customer commitments |
| Nebius–Microsoft agreement | Company IR | 2025-09-08 | Contract economics and credibility |
| CoreWeave Q2 2026 results | Regulatory filing / company results | Q2 2026 | Direct competitor comparison |
| Nebius market statistics | Market data | 2026-10-14 | Intraday valuation snapshot |
| Nebius–Palantir partnership | Company IR | September 2026 | Strategic development |
| Inferize acquisition | Company IR | 2026-10-01 | Managed inference strategy |
| NBIS common regular-close snapshot | Market data | 2026-10-20 16:00 EDT | Refreshed price, scenario returns and normalized Opportunity Score / shared rank |
This page presents the saved NBIS analysis. Ratings, zones, and opportunity scores shown above are recalculated at its dated reference price using FolioFacts rules; the live Creator Watchlist uses current market prices and can differ. Missing financial data is not estimated.
This publication reflects my personal investment research, thoughts, and opinions only. It is intended strictly for informational and educational purposes and does not constitute personalized investment, legal, tax, or financial advice. Any ratings or views expressed are general analytical opinions, not recommendations tailored to any reader or offers or solicitations to buy or sell any security or financial instrument.
I am not a registered investment adviser, broker-dealer, or financial planner. This content does not take into account your specific investment objectives, financial situation, or risk tolerance. All investments involve substantial risk, including the potential loss of all principal capital. Past performance is never indicative of future results.
While I endeavor to provide accurate information, all data and analysis are provided "as is" without any warranty of accuracy, completeness, or timeliness. You must conduct your own independent due diligence and/or consult with a licensed professional before making any financial decisions. I may hold long or short positions in any of the securities or assets discussed herein at any time without notice.
Quality is price independent. Opportunity combines that quality with scenario-based upside, bear-case protection, and confidence. Ownership and cost basis do not affect these scores. Watchlist rank compares the scored companies in that Watchlist.
Model scores, valuation estimates, and price zones are based on assumptions, not guarantees of future performance or individualized advice. Read our Disclaimer.
25% business durability
25% moat and competitive position
20% growth runway
15% financial quality
15% management and capital allocation
30% Quality Score
35% normalized probability-weighted upside
20% normalized bear-case return
15% risk and valuation confidence
At least 50% upside to P/W fair value, with price no more than 10% above bear-case fair value.
At least 25% upside to P/W fair value, above the Strong Buy boundary.
Modeled return from approximately −10% to below +25%.
More than 10% downside to P/W fair value.
Risk-adjusted view. This framework balances expected upside, downside risk and valuation confidence. Zones may be less reliable for companies priced mainly on future potential rather than current fundamentals.
These prompt a review of the research and Opportunity Score. Watchlist ranks are calculated automatically from the scores currently displayed.