Back to Portfolio
NBIS

Nebius Group N.V.

$165.34 41.8B market cap April 2026
Nebius Group N.V. NBIS BUFFETT / MUNGER / KLARMAN SUMMARY
1 SNAPSHOT
Price$165.34
Market Cap41.8B
2 BUSINESS

Nebius Group is executing one of the most extraordinary corporate pivots in recent memory, leveraging decades of Yandex engineering expertise to build a rapidly scaling AI cloud business. The revenue trajectory from $21M (2023) to guided $3.0-3.4B (2026) is remarkable, and mega-deals with Microsoft ($17-19B) and Meta ($3B) provide contracted revenue visibility. However, at $165 per share and 79x trailing revenue, the market is pricing in near-perfect execution of a $16-20B capex plan that requires significant additional financing. Hidden assets (ClickHouse $4.2B, Avride $2.2B) provide downside protection but not enough to justify current premium. This is a business worth admiring from a distance until a meaningful correction brings the price closer to intrinsic value.

3 MOAT NARROW

Ex-Yandex engineering talent, multi-billion long-term contracts with Microsoft/Meta creating switching costs, European AI sovereignty positioning, NVIDIA strategic partnership with early access to next-gen platforms

4 MANAGEMENT
CEO: Arkady Volozh

Good - aggressive capex backed by contracted revenue, ATM equity program authorized but unused showing dilution sensitivity, NVIDIA partnership validates strategic direction

5 ECONOMICS
-112.5% Op Margin
-5% ROIC
2.2% ROE
1377.83x P/E
-3.68B FCF
26.2% Debt/EBITDA
6 VALUATION
FCF Yield-8.8%
DCF Range53 - 180

Overvalued by 44% vs probability-weighted fair value of $115

7 MUNGER INVERSION
Kill Event Severity P() E[Loss]
Execution risk on $16-20B capex plan -- data center construction delays directly impact revenue HIGH - -
Capital raising and dilution risk -- needs $6-8B in additional 2026 funding beyond existing resources MED - -
8 KLARMAN LENS
Downside Case

Execution risk on $16-20B capex plan -- data center construction delays directly impact revenue

Why Market Right

Capex execution delays in data center construction, permitting, power procurement; GPU pricing compression if supply catches demand; Equity dilution from ATM program or additional offerings; Macro downturn reducing enterprise AI spending

Catalysts

Q1 2026 earnings (April 29) showing massive revenue ramp from Meta full deployment; Additional mega-deals beyond Microsoft and Meta; ClickHouse IPO or secondary sale crystallizing $4.2B+ stake value; European AI sovereignty regulations favoring non-US cloud providers; Avride autonomous driving commercial milestones

9 VERDICT WAIT
B Quality Moderate - $3.7B cash cushion and $7.2B in subsidiary assets, but $16-20B 2026 capex plan requires massive additional financing. Asset-backed debt secured by Microsoft/Meta creditworthiness mitigates risk.
Strong Buy$75
Buy$95
Fair Value$180

Monitor for 40-50% pullback to $80-100 range. Set alerts at $95 (accumulate) and $75 (strong buy). Watch Q1 2026 earnings April 29 for execution proof.

🧠 ULTRATHINK Deep Philosophical Analysis

Nebius Group (NBIS) - Deep Philosophical Analysis

An exercise in thinking clearly about a business that defies conventional value frameworks

The Core Question: What Are We Actually Buying?

When you buy Nebius at $165, you are buying a promissory note on the future of AI infrastructure. You are betting that the world needs dramatically more GPU compute than exists today, that Nebius can build and deliver it faster than almost anyone else, and that the contracts already signed with Microsoft and Meta represent the beginning rather than the peak.

This is not a business you can evaluate with traditional value metrics. Trailing P/E of 1,378x is meaningless. FCF is deeply negative. The balance sheet is being deliberately leveraged to fund the build-out. Every conventional Buffett screen would reject this stock immediately.

And yet, there is something genuinely interesting here. Arkady Volozh built Yandex into Russia's dominant technology company over twenty years, navigating everything from the 2008 financial crisis to Western sanctions. When the Russia-Ukraine war made his life's work untenable, he did not retire. He pivoted -- at age 60 -- into the most capital-intensive, fast-moving sector in technology. That takes either extraordinary conviction or extraordinary hubris. Possibly both.

Moat Meditation: The Paradox of Scarcity-Dependent Moats

The honest assessment of Nebius's moat today is that it is narrow and primarily scarcity-dependent. The company is profitable at the unit level (69% gross margin) because GPU compute is scarce and pricing is strong. Every GPU cluster they bring online sells out instantly. The question is: will it always be so?

History suggests caution. Scarcity-dependent moats are inherently fragile. In the early 2000s, fiber optic bandwidth was scarce, and companies that built fiber networks commanded enormous premiums. Then supply caught up to demand, prices collapsed, and companies like WorldCom and Global Crossing went bankrupt. The solar panel industry went through a similar boom-bust cycle.

However, there is a critical difference with AI compute: the demand is not driven by speculative capacity building but by real enterprise workloads. Microsoft is not paying $17 billion over five years for speculative GPU access -- it needs the compute for Azure AI services that generate real revenue. Meta is not paying $3 billion for GPUs it might use someday -- it is running production AI workloads that power its advertising and recommendation systems.

The deeper moat lies in what Nebius is building on top of the hardware. The Ether platform, Token Factory, Talking Factory -- these are software layers that create stickiness. If Nebius succeeds in building a genuine AI cloud platform (not just GPU hosting), the moat widens considerably. The ex-Yandex team has the engineering depth to do this. Whether they can execute while simultaneously scaling from 9 to 18+ data centers across multiple continents is the open question.

The Owner's Mindset: Would Buffett Touch This?

Absolutely not. And for good reason.

Buffett would look at this and see: a company burning billions in cash, requiring external financing to fund growth, operating in a technology sector prone to rapid disruption, with a corporate history intertwined with Russian sanctions, and trading at 79x trailing revenue. He would politely pass and go back to reading annual reports of insurance companies.

But Buffett himself would acknowledge that his approach is not the only valid one. He has spoken admiringly of builders like Jeff Bezos who invested aggressively through periods of cash burn to build extraordinary businesses. Amazon in 2002-2005 looked nothing like a Buffett stock either.

The question for the investor is: which framework applies? Is Nebius more like Amazon in 2004 -- a visionary company investing through a cash-burn phase to build something durable? Or is it more like Webvan -- a company scaling too fast in a capital-intensive business where the economics ultimately do not work?

The contracted revenue visibility (Microsoft, Meta, NVIDIA relationship) tilts the answer toward the Amazon analogy. These are not speculative contracts -- they are backed by the most creditworthy technology companies on earth. But the sheer scale of the capex plan ($16-20B in a single year for a company with $530M in trailing revenue) is unprecedented and carries genuine execution risk.

Risk Inversion: What Could Destroy This Business?

The kill scenarios, in order of probability:

  1. GPU demand normalization (30% probability over 3 years). If AI training and inference workloads become more efficient, or if custom silicon from hyperscalers reduces demand for NVIDIA GPUs, Nebius's pricing power collapses. This is the single biggest structural risk.

  2. Capital market closure (15% probability). If credit markets tighten or equity markets crash, Nebius cannot fund its $16-20B capex plan. The company is fundamentally dependent on external capital markets remaining open and favorable.

  3. Execution failure (20% probability). Building data centers at this pace across multiple countries involves construction risk, permitting risk, power procurement risk, and supply chain risk. Any significant delay cascades through the revenue model.

  4. Customer concentration blow-up (10% probability). If Microsoft or Meta renegotiate terms, delay payments, or reduce scope, the impact on Nebius would be severe.

  5. Competitive destruction (15% probability over 5 years). CoreWeave, hyperscaler internal builds, or a new entrant could commoditize AI cloud infrastructure.

None of these are existential certainties, but collectively they represent meaningful risk. A value investor demands compensation for these risks in the form of a discount to fair value. At $165, no such discount exists.

Valuation Philosophy: The Price of Dreams

At the current price, you are paying for the dream. The probability-weighted fair value of approximately $115 per share means the market is embedding a 44% premium for momentum, excitement, and the hope of further mega-deals.

The hidden assets provide some comfort. ClickHouse at $4.2B, Avride at $2.2B, Toloka and TripleTen at perhaps $0.8B combined -- these create a $7.2B asset floor, or roughly $28 per share on a diluted basis. In a severe downside scenario, these assets prevent total loss. But paying $165 for $28 in visible asset floor is not a value proposition.

The right price for a value investor to engage with this story is in the $75-95 range. At $75, you are paying roughly 5x 2026 estimated revenue with a 33% margin of safety to the base case fair value. At $95, you are paying 6.5x with a 17% margin of safety. These are prices where the risk-reward becomes attractive even for conservative investors.

The Patient Investor's Path

The most likely path to an attractive entry point is a broad market correction combined with a specific execution stumble. AI infrastructure stocks have been among the most volatile names in the market, and a 40-50% drawdown from current levels is well within historical norms for this type of stock.

The Q1 2026 earnings report on April 29 will be important. If revenue significantly exceeds expectations (reflecting Meta full deployment), the stock could push higher short-term, making the waiting game longer. If there is any execution shortfall -- even minor -- the stock could correct sharply given the stretched valuation.

The wise course: add NBIS to the watchlist, set price alerts at $95 and $75, and occupy your time with investments that offer better risk-adjusted returns today. The business is real. The growth is real. The talent is real. The price, however, is not yet real for a disciplined value investor.

Time is on the side of the patient.

"Price is what you pay. Value is what you get. In the case of Nebius, what you get is extraordinary. The question is entirely about what you pay."

Executive Summary

Nebius Group is the former Yandex NV, restructured after selling all Russian assets in mid-2024 for $5.4 billion. The company relisted on NASDAQ in October 2024 at approximately $24 per share under the leadership of founder Arkady Volozh, pivoting entirely to AI cloud infrastructure. In 18 months, the stock has risen 588% to $165, powered by mega-deals with Microsoft ($17-19B), Meta ($3B), and a $2B strategic investment from NVIDIA. The company guides for $3.0-3.4B revenue in 2026 (up from $530M in 2025) with $7-9B annualized run-rate revenue by year-end.

Verdict: WAIT -- Exceptional business trajectory but priced far beyond reasonable value investor entry points. Current valuation embeds near-perfect execution of a capital-intensive growth plan requiring $16-20B in 2026 capex alone.

Phase 1: Risk Assessment

1.1 Execution Risk (HIGH)

The single largest risk is the audacity of the capacity build-out plan. Nebius is guiding $16-20B in capex for 2026 -- roughly 5x its projected full-year revenue. The company must:

  • Bring 800MW-1GW of data center capacity online during 2026
  • Deliver remaining Microsoft tranches on schedule (majority in H2 2026)
  • Scale from 9 data centers to potentially 18+ across US and Europe
  • Manage construction timelines, permitting, power procurement simultaneously

Any delay in capacity delivery directly translates to revenue shortfalls. The CFO explicitly stated that 2025 revenue came in at the midpoint of guidance rather than the high end "simply due to the exact timing of when capacity comes online."

1.2 Capital Structure Risk (HIGH)

Nebius ended 2025 with $3.0B cash and $4.9B total debt (D/E 1.70x). To fund $16-20B in 2026 capex, the company needs an additional $6-8B beyond existing resources. Funding sources include:

  • Corporate debt and asset-backed financing (secured against Microsoft/Meta creditworthiness)
  • ATM equity program (25M shares authorized, not yet used)
  • NVIDIA $2B investment (completed March 2026)
  • Potential monetization of ClickHouse stake (~$4.2B at $15B valuation)

Dilution is a real risk. If Nebius issues the full 25M ATM shares at $165, that adds $4.1B but dilutes existing shareholders by ~10%. The pre-funded warrants from NVIDIA add 21M shares. Total diluted share count could reach 300M+ from 253M currently.

1.3 Concentration Risk (HIGH)

The Microsoft and Meta contracts collectively represent the vast majority of 2026 revenue guidance. The CFO stated that Microsoft will only reach full annual run-rate in 2027, and Meta was fully deployed by February 2026. If either relationship deteriorates or contracts are renegotiated, the impact would be severe.

1.4 Competitive Risk (MODERATE-HIGH)

Nebius competes against hyperscalers (AWS, Azure, GCP) who have orders-of-magnitude more resources, as well as other AI infrastructure players (CoreWeave, Lambda Labs, Crusoe Energy). CoreWeave, in particular, is a direct competitor with its own NVIDIA partnership and is scaling rapidly. The hyperscalers are also building massive GPU capacity internally.

1.5 Russia Legacy / Governance Risk (LOW-MODERATE)

Volozh renounced Russian citizenship in February 2026. EU sanctions were lifted in March 2024. The corporate structure is Dutch-incorporated with Israeli operations. While the Russia stigma is fading, some institutional investors may still avoid the name. Insider ownership at 3.8% is relatively low for a founder-led company.

1.6 Technology Cycle Risk (MODERATE)

Nebius is heavily dependent on NVIDIA GPUs (Blackwell, Vera Rubin platforms). Any disruption in NVIDIA supply, or a shift in the GPU competitive landscape (AMD, custom ASICs from hyperscalers), could impair Nebius's proposition. The 5-year GPU depreciation schedule assumes sustained demand that may not materialize if AI training paradigms shift.

Phase 2: Financial Analysis

2.1 Revenue Trajectory

Period Revenue YoY Growth
FY2023 $21M N/A (pre-pivot)
FY2024 $118M +462%
FY2025 $530M +350%
FY2026E $3.0-3.4B +470-540%
FY2026 ARR $7-9B Target

Revenue growth is explosive. Q4 2025 revenue of $228M represented 547% YoY growth and 56% sequential growth. The core AI cloud business grew 830% YoY in Q4 2025. This is among the fastest revenue ramps in technology history.

2.2 Profitability

  • Gross Margin: 68.6% (FY2025) -- strong for infrastructure, reflecting pricing power
  • Adjusted EBITDA Margin: 24% in Q4 2025 (core AI cloud), targeting ~40% group for FY2026
  • Operating Margin: -112.5% (FY2025) -- deeply negative due to D&A ($404M), R&D ($177M), SBC ($83M)
  • Net Income: $102M (FY2025) -- positive due to non-operating items
  • Medium-term EBIT target: 20-30% (management guidance)

The unit economics are promising. Gross margins near 69% with EBITDA margins expanding quarter-over-quarter suggest the core business is profitable at scale. However, the massive depreciation burden from $4B+ annual GPU capex will suppress reported earnings for years.

2.3 Balance Sheet

Metric FY2025 FY2024
Cash $3.68B $2.45B
Total Debt $4.89B $0.05B
Equity $4.61B $3.25B
D/E 1.70x 0.09x
Total Assets $12.4B $3.5B

The balance sheet transformed dramatically in 2025. Debt increased from near-zero to $4.9B as Nebius leveraged up to fund GPU purchases. The $3.7B cash position provides a cushion, but with $16-20B in planned 2026 capex, the company will need significant additional capital.

2.4 Cash Flow

  • Operating CF (FY2025): $385M (boosted by $834M in Q4 from upfront customer payments)
  • CapEx (FY2025): $4.07B
  • FCF (FY2025): -$3.68B
  • Financing CF (FY2025): +$5.13B (debt issuance, equity, NVIDIA investment)

FCF will remain deeply negative through at least 2027 as the company invests aggressively. The operating cash flow dynamic is favorable -- long-term contracts generate upfront payments that help fund the build-out.

2.5 Hidden Assets (Sum of Parts)

Asset Estimated Value
Core AI Cloud Business ~$30-35B (at current market cap less below)
ClickHouse stake (~28%) ~$4.2B (at $15B valuation)
Avride (~83% stake) ~$2.2B (at $2.7B valuation)
Toloka AI ~$0.5B (estimated)
TripleTen ~$0.3B (estimated)
Total subsidiary value ~$7.2B

The non-core assets provide meaningful downside protection. The ClickHouse stake alone at $4.2B represents ~10% of the market cap and could be monetized to fund capex.

Phase 3: Moat Assessment

3.1 Moat Type: Emerging / Narrow (Widening Potential)

Engineering Talent (MODERATE MOAT): The ex-Yandex team has deep expertise in distributed systems, ML infrastructure, and large-scale operations. They built Russia's dominant search engine and cloud platform from scratch. This talent base is difficult to replicate quickly. The NVIDIA partnership validates their technical depth.

Customer Lock-in via Long-term Contracts (GROWING MOAT): Multi-billion dollar, multi-year contracts with Microsoft and Meta create revenue visibility and switching costs. Once workloads are deployed on Nebius infrastructure, migration is costly and risky. Contract durations are increasing (50% longer average for new customers).

European AI Sovereignty (EMERGING MOAT): Nebius is positioning as a non-US, non-Chinese AI cloud provider with data centers in Finland, Israel, UK, and continental Europe. As European governments and enterprises seek AI infrastructure outside US hyperscaler control, Nebius has first-mover advantage.

Full-Stack Integration (EMERGING MOAT): The partnership with NVIDIA gives Nebius early access to next-gen platforms (Vera Rubin). Combined with proprietary cloud software (Ether platform, Nebius Talking Factory, Token Factory), Nebius is building a differentiated stack rather than just renting GPUs.

Moat Width: NARROW -- Currently dependent on capacity scarcity in the AI GPU market. If the supply-demand imbalance normalizes, pricing power and margins could compress. The moat widens only if Nebius successfully builds software differentiation and customer stickiness.

Moat Trend: WIDENING -- Each mega-deal, each data center, and each software release makes the moat incrementally wider. But it is still early days.

Phase 4: Synthesis and Valuation

4.1 Valuation Metrics

Metric Value
Market Cap $41.8B
EV (approx) $43.1B (MCap + $4.9B debt - $3.7B cash)
P/S (TTM, FY2025) 79x
P/S (FY2026E, $3.2B mid) 13.1x
P/S (FY2026 ARR, $8B mid) 5.2x
EV/Revenue (FY2026E) 13.5x
Forward P/E 68.5x
P/B 9.1x
FCF Yield Deeply negative

4.2 Valuation Framework

Bull Case (ARR of $9B realized, 40% EBITDA margin stabilizes):

  • 2027 Revenue: ~$8-9B
  • 2027 EBITDA: ~$3.2-3.6B
  • At 15x EV/EBITDA (mature infrastructure): EV = $48-54B
  • Less net debt, plus subsidiary value: Equity value ~$50-58B
  • Per share (300M diluted): $167-193
  • Implies: Current price is roughly fair for the bull case

Base Case (ARR of $7B, some execution delays):

  • 2027 Revenue: ~$6-7B
  • 2027 EBITDA: ~$2.4-2.8B
  • At 12x EV/EBITDA: EV = $29-34B
  • Equity value ~$31-38B
  • Per share: $103-127
  • Implies: 23-38% downside from current price

Bear Case (GPU demand softens, contract delays, dilution):

  • 2027 Revenue: ~$4-5B
  • 2027 EBITDA: ~$1.2-1.5B
  • At 10x EV/EBITDA: EV = $12-15B
  • Add subsidiary value $7B, less net debt: Equity value ~$15-19B
  • Per share (320M diluted): $47-59
  • Implies: 64-72% downside

4.3 Fair Value Range

Scenario Probability Fair Value/Share
Bull 25% $180
Base 50% $115
Bear 25% $53
Probability-Weighted $115

4.4 Entry Prices

  • Strong Buy: $75 (at ~5x 2026E revenue, 33% margin of safety to base case)
  • Accumulate: $95 (at ~6.5x 2026E revenue, 17% margin of safety to base case)
  • Current Price: $165.34 (at 13x 2026E revenue -- pricing in bull case)
  • Gap to Accumulate: -42.5%

4.5 Catalysts

Positive:

  • Q1 2026 earnings (April 29) could show massive revenue ramp from Meta deployment
  • Additional mega-deals beyond Microsoft and Meta
  • ClickHouse IPO or secondary sale crystallizing $4B+ value
  • Avride autonomous driving milestones
  • European AI sovereignty regulations favoring non-US providers

Negative:

  • Capex execution delays (construction, permitting, power)
  • GPU pricing compression if supply catches up to demand
  • Equity dilution from ATM program or secondary offerings
  • Macro downturn reducing enterprise AI spending
  • CoreWeave IPO success drawing capital away from NBIS

Management Assessment

Arkady Volozh (CEO, Founder):

  • Built Yandex from scratch into a $30B+ company
  • Successfully navigated EU sanctions and Russia exit
  • Renounced Russian citizenship February 2026
  • Visionary founder with proven execution at scale
  • Low insider ownership (3.8%) is a concern

Dado Alonso (CFO):

  • Articulate and disciplined on capital allocation
  • Clear about dilution sensitivity
  • Transparent about execution timelines

Capital Allocation: Above average -- the team is deploying capital aggressively but with clear ROI visibility from contracted revenue. The decision to not use the ATM program yet shows discipline.

Investment Thesis

Nebius Group is executing one of the most extraordinary corporate pivots in recent memory. Volozh and his team have leveraged decades of engineering expertise from Yandex to build a rapidly scaling AI cloud business that has attracted the world's largest technology companies as customers. The revenue trajectory -- from $21M in 2023 to a guided $3.0-3.4B in 2026 -- is remarkable.

However, the current stock price of $165 already embeds near-perfect execution of an extremely capital-intensive growth plan. The company needs to raise $6-8B in additional capital in 2026, creating meaningful dilution risk. At 79x trailing sales and 13x forward sales, the market is pricing in the bull case scenario with no margin of safety.

For value investors, the right approach is to admire the business from a distance and wait for a meaningful pullback. A 40-50% correction (to $80-100 range) is not implausible given the stock's extreme volatility and the myriad execution risks. The hidden asset value (ClickHouse, Avride) provides some floor, but not enough to justify the current premium.

This is a WAIT at current levels. The business quality is improving rapidly, but the price has outrun the fundamentals.

Disclaimer: This analysis is for educational purposes only and does not constitute investment advice. Always do your own research.

=== VERDICT: NBIS | WAIT | SB:$75.00 | Acc:$95.00 | Current:$165.34 ===