Nebius Group N.V.
NBIS is building real AI infrastructure demand, but at $219.13 the stock already discounts years of execution while the quantified base case remains $178.53, leverage and capex are extreme, and valuation is disconnected from current earnings.
NBIS is not a broken story. It is an expensive one. The company is clearly tapping into a real AI infrastructure shortage, and the latest quarter gave bulls tangible evidence: Q2 revenue of $582M, ARR of $3.0B exiting Q2, 41% adjusted EBITDA margin on management's figures, four landmark deals averaging more than $1B each, and a statement from the CEO that the company could sell its entire 2027 capacity today. Those are not promotional crumbs. They are signs of genuine product-market fit in a supply-constrained part of the AI stack.
The issue is what that is worth at $219.13. The stock carries a $57.7B market cap against FY2025 revenue of $0.5B, reported FCF of $-3.4B, and total debt of $10.1B. Even if FY26 lands at the high end of the $3.0B-$3.4B revenue guide, the equity already discounts a huge amount of future capacity coming online, staying filled, and remaining attractively priced. That is a lot to assume in an industry where supply can show up quickly, customer concentration can matter, and financing conditions shape outcomes. With the 10-year Treasury at 4.69%, this is not a forgiving backdrop for a company planning $20B-$25B of capex in a single year.
The short case does not depend on demand collapsing. It depends on valuation compressing from extreme levels as fundamentals mature. A stock on 45.28x sales does not need bad news to fall. It only needs less-good news, slower monetization of backlog, a heavier financing mix, or evidence that premium Blackwell pricing does not fully persist as more capacity enters the market. The recent tape reinforces that fragility: the shares are still 46.8% above the 200-day average, short interest is 29.80%, and the one-week move was -21.09%.
Conviction is only medium because the business is scaling fast enough to make old numbers stale, and because 2027 guidance could force another leg up if pricing and capacity remain this tight. The case depends on three things: monetizing ARR into reported revenue, funding the build without equity impairment, and proving that today's pricing power survives into a larger supply environment. At this price, the burden of proof sits with the bull case, not the bear case.
2027 guidance formalizes the current demand narrative, Blackwell pricing sustains the 15% auction premium, and prepayments plus structured debt fund growth without major dilution.
NBIS executes FY26 guidance broadly as promised, but valuation compresses toward a more defensible level as investors weigh $20B-$25B capex and peer-multiple reality.
AI infrastructure demand stays solid but financing costs, dilution risk, or a delay in turning ARR $3.0B into clean cash earnings forces a sharp de-rating from 45.28x sales.
Quant call (SELL) diverges from analyst consensus (Strong Buy: 6 buy / 2 hold / 0 sell) — resolve explicitly.
The four most dangerous words in investing are: 'this time it's different.'
Forward earnings power is improving fast, but the stock is still being valued on a 2027-2028 dream rather than on current profitability, with trailing valuation at 842.81x P/E against management's FY26 target of ~40% adjusted EBITDA margin that still sits below the line of massive capex and financing needs. The moat today is access to power, GPU capacity, and willingness to build at speed, but in 3-5 years that moat only holds if NBIS secures and monetizes the planned 5GW contracted power base before larger hyperscalers and rival AI clouds close the gap. AI is a strong secular tailwind here because demand for accelerated compute is clearly real, but this is also the part of tech where oversupply, pricing resets, and capital market tightening can destroy equity value quickly when a business is spending $20B-$25B a year. Rising long rates at 4.69% matter because this model is unusually dependent on external and structured financing, so the secular story is attractive while the equity at this price is not.
- Valuation is stretched: NBIS trades at 842.81x P/E and 45.28x P/S versus peer averages of 22.9x and 2.6x.
- Capital intensity is huge: management still guides to $20B-$25B FY26 capex against FY2025 revenue of just $0.5B and reported FCF of $-3.4B.
- Operating momentum is real but early: Q2 revenue reached $582M, ARR exited at $3.0B from $1.9B in Q1, and management reaffirmed FY26 group revenue of $3.0B-$3.4B with ~40% adjusted EBITDA margin.
- Demand is not the problem: management said it could sell "our entire 2027 capacity" today, ARR rose to $3.0B from $1.9B in one quarter, and Q2 capacity again sold out.
- Pricing is improving, not weakening: the first Blackwell auction cleared at 15% above the prior high price, and management discussed short-term deals at $40M-$50M per MW.
- Execution improved in recent quarters: NBIS beat EPS estimates in 5 of the last 7 reported quarters before the latest Q2 beat streak extension, including +70.0% on 2026-05-13 and +83.3% on 2026-08-12.
- Customer financing support is meaningful: 70% of Q2 landmark deals included upfront prepayments, and customer prepayments are expected to exceed $9B in 2026, which can lower outside capital needs.
- The stock is priced on future perfection: 842.81x P/E, 45.28x P/S, and a quant base value of $178.53, or -18.5% from the current $219.13.
- Cash burn and leverage remain severe: FCF is $-3.4B, total debt is $10.1B, debt/EBITDA is 20.32x, and debt/equity is 0.97 before the biggest capex phase.
- Guidance still implies a funding treadmill: FY26 capex of $20B-$25B dwarfs current revenue scale, and even with structured financing the business depends on sustained access to debt and customer prepayments.
- Trading setup is fragile: short interest is 29.80% of shares, the stock is 46.8% above its 200-day average, and options gamma is negative with zero-gamma at $232.45, which can amplify downside if momentum breaks.
| Metric | Value | Context |
|---|---|---|
| Current price | $219.13 | Reference price for the rating and scenario work |
| Price target | $178.53 | Ground-truth target, -18.5% vs current, Medium confidence |
| Market cap | $57.7B | Large equity value relative to current reported earnings and cash flow |
| FY2025 revenue | $0.5B | -97.1% YoY, though current quarterly run rate is far above this stale annual base |
| FY2025 EPS | $0.11 | +85.9% YoY, but still supports an extreme 842.81x P/E |
| Operating margin | -304.8% | Improved from -375.1% prior FY, but still deeply negative on reported FY2025 numbers |
| Net margin | -550.6% | Shows how far reported profitability still is from equity-value expectations |
| Valuation | P/E 842.81x | P/S 45.28x | P/B 5.93x | Versus peers at 22.9x P/E and 2.6x P/S |
| Balance sheet leverage | Debt/Equity 0.97x | Debt/EBITDA 20.32x | Total debt $10.1B | High leverage before the largest capex phase |
| Liquidity and cash flow | Current ratio 4.03x | FCF $-3.4B | Near-term liquidity looks adequate, but cash burn remains heavy |
| FY26 guidance | Revenue $3.0B-$3.4B | Adj. EBITDA margin ~40% | Capex $20B-$25B | Reaffirmed on the 2026 Q2 call |
| Price action | 52-week range $63.26-$299.86 | Current price is -26.9% vs high and 46.8% above the 200-day average of $149.25 |
The biggest risk to the SELL call is that FY26 guidance of $3.0B-$3.4B revenue and ~40% adjusted EBITDA margin, plus evidence of 2027 capacity already saleable, could justify a much higher earnings base faster than the current valuation framework assumes.
Next catalyst is Q3 earnings on November 10, 2026, with the key watch items being formal 2027 guidance, customer prepayments versus the >$9B expectation, and whether pricing strength converts into reported revenue rather than just backlog.
The shareholder base has some institutional sponsorship, with the top 3 institutions holding 37.0%, but insider ownership is only 1.6%, which limits insider alignment as a support signal. Short interest is very high at 29.80%, or 60.2M shares, with 3.4 days to cover, so the stock can squeeze hard on positive AI headlines but that same positioning also reflects serious skepticism about valuation and funding. Options positioning is in a negative gamma regime with zero-gamma at $232.45, close enough to spot to keep near-term trading unstable after a -21.09% one-week move.
FY2025 reported revenue was $0.5B, but Q2 2026 alone reached $582M, showing the reported base is stale relative to the current run rate.
ARR rose from $1.9B in Q1 to $3.0B in Q2 as newly added capacity sold through.
Geographic revenue split was not provided. Operating footprint references Finland, Kansas City, Iceland, Vineland, and a US build-to-suit facility, suggesting a growing transatlantic infrastructure base.
Q2 beat and FY26 guide reaffirmed, with 2027 pricing/capacity upside building.
Management's operating trajectory has improved, but the record is mixed enough to keep credibility at moderate, not high. On the 2024 Q4 call, Arkady Volozh said "our projected December, 2025 analyzed run rate revenue of $750 million to $1 billion is well within reach" and the company has now reached $3.0B ARR exiting Q2 2026, so the direction of demand and capacity ramp clearly exceeded that earlier benchmark. The 2024 Q4 team also guided to 100 MW in operations by end-2025 and a heavily 2H-weighted 2025 revenue curve; current results support that the business did ramp materially after that period, but reported FY2025 revenue of only $0.5B landed at the bottom of the old $500M-$700M range and FY2025 profitability was still poor, with -304.8% operating margin and $-3.4B FCF. Recent earnings delivery has been better, with beats of +29.9%, +70.0%, and +83.3% in three of the last four reports, but there was also a major miss of -70.7% on 2026-02-12. The strategy has not changed materially, it is still a speed-and-capacity buildout into AI infrastructure, but the burden of proof remains on converting backlog and ARR into durable cash earnings without balance-sheet strain.
We continue to expect annualized run rate revenue of $7 billion to $9 billion, group revenue of between $3 billion and $3.4 billion, group adjusted EBITDA margin of approximately 40%, and capital expenditures of $20 billion to $25 billion.
The dynamics we expect across both capacity and pricing make us extremely excited about 2027, although we will provide a formal guidance later this year.
But most importantly, we could sell today our entire 2027 capacity on these terms if we wanted to.
In Q2, we grew Group revenue by 454% to $582 million, up 46% from last quarter.
The auction result was a price 15% higher than we've ever seen before and 20% higher than our pipeline for Blackwells.
The moat is currently based on scarce AI compute supply, access to power, and speed of deployment. That advantage is real while capacity is sold out and Blackwell pricing is clearing 15% above prior highs, but it is only moderately durable because larger cloud platforms and funded AI infrastructure rivals can replicate hardware over time. The more durable edge would be long-dated customer commitments tied to power contracts, prepayments, and operating know-how across a 5GW power pipeline, but that moat is still being built rather than proven.
Capital allocation is aggressively expansionary. The company reported $-3.4B FCF, $5.7B capex in the deep-dive data, no buybacks, and no dividend, which is appropriate for the stage but leaves equity holders fully exposed to buildout execution. Management added $775M of asset-backed debt in July and expects >$9B in customer prepayments for 2026, which looks disciplined in the sense that it matches funding sources to infrastructure assets, but the absolute scale of planned $20B-$25B FY26 capex keeps financing risk elevated.
Rated SELL, Medium conviction. Nebius has a real business in a hot part of AI, and the recent quarter gave that story more substance than many skeptics expected. The problem is the stock still asks investors to pay today for a large chunk of 2027 and beyond, while the company is carrying extreme valuation, heavy leverage, and a $20B-$25B annual capex plan. A move materially closer to the $178.53 base value, or hard evidence that 2027 earnings power can absorb this capital intensity without balance-sheet stress, would change the view. Until then, the better call is to avoid chasing the story at this price.
≣ Full research note tap to expand
Executive Summary
SELL, Medium conviction. Rated SELL, Medium conviction. Nebius is showing genuine commercial momentum, with Q2 revenue of $582M, ARR at $3.0B, and FY26 guidance reaffirmed, but the stock still trades at 842.81x P/E and 45.28x sales with $-3.4B FCF, $10.1B debt, and a planned $20B-$25B FY26 capex bill. The business may grow into a much larger earnings base, but the stock is still priced for near-flawless delivery despite being 46.8% above its 200-day average and only 26.9% below its high after a speculative AI run.
Investment Thesis
NBIS is not a broken story. It is an expensive one. The company is clearly tapping into a real AI infrastructure shortage, and the latest quarter gave bulls tangible evidence: Q2 revenue of $582M, ARR of $3.0B exiting Q2, 41% adjusted EBITDA margin on management's figures, four landmark deals averaging more than $1B each, and a statement from the CEO that the company could sell its entire 2027 capacity today. Those are not promotional crumbs. They are signs of genuine product-market fit in a supply-constrained part of the AI stack.
The issue is what that is worth at $219.13. The stock carries a $57.7B market cap against FY2025 revenue of $0.5B, reported FCF of $-3.4B, and total debt of $10.1B. Even if FY26 lands at the high end of the $3.0B-$3.4B revenue guide, the equity already discounts a huge amount of future capacity coming online, staying filled, and remaining attractively priced. That is a lot to assume in an industry where supply can show up quickly, customer concentration can matter, and financing conditions shape outcomes. With the 10-year Treasury at 4.69%, this is not a forgiving backdrop for a company planning $20B-$25B of capex in a single year.
The short case does not depend on demand collapsing. It depends on valuation compressing from extreme levels as fundamentals mature. A stock on 45.28x sales does not need bad news to fall. It only needs less-good news, slower monetization of backlog, a heavier financing mix, or evidence that premium Blackwell pricing does not fully persist as more capacity enters the market. The recent tape reinforces that fragility: the shares are still 46.8% above the 200-day average, short interest is 29.80%, and the one-week move was -21.09%.
Conviction is only medium because the business is scaling fast enough to make old numbers stale, and because 2027 guidance could force another leg up if pricing and capacity remain this tight. The case depends on three things: monetizing ARR into reported revenue, funding the build without equity impairment, and proving that today's pricing power survives into a larger supply environment. At this price, the burden of proof sits with the bull case, not the bear case.
Key Metrics
| Metric | Value | Context | ||
|---|---|---|---|---|
| Current price | $219.13 | Reference price for the rating and scenario work | ||
| Price target | $178.53 | Ground-truth target, -18.5% vs current, Medium confidence | ||
| Market cap | $57.7B | Large equity value relative to current reported earnings and cash flow | ||
| FY2025 revenue | $0.5B | -97.1% YoY, though current quarterly run rate is far above this stale annual base | ||
| FY2025 EPS | $0.11 | +85.9% YoY, but still supports an extreme 842.81x P/E | ||
| Operating margin | -304.8% | Improved from -375.1% prior FY, but still deeply negative on reported FY2025 numbers | ||
| Net margin | -550.6% | Shows how far reported profitability still is from equity-value expectations | ||
| Valuation | P/E 842.81x | P/S 45.28x | P/B 5.93x | Versus peers at 22.9x P/E and 2.6x P/S |
| Balance sheet leverage | Debt/Equity 0.97x | Debt/EBITDA 20.32x | Total debt $10.1B | High leverage before the largest capex phase |
| Liquidity and cash flow | Current ratio 4.03x | FCF $-3.4B | Near-term liquidity looks adequate, but cash burn remains heavy | |
| FY26 guidance | Revenue $3.0B-$3.4B | Adj. EBITDA margin ~40% | Capex $20B-$25B | Reaffirmed on the 2026 Q2 call |
| Price action | 52-week range $63.26-$299.86 | Current price is -26.9% vs high and 46.8% above the 200-day average of $149.25 |
Financial Strength
The balance sheet is not distressed in the near term, as the 4.03 current ratio and access to asset-backed financing suggest liquidity is available, but the capital structure is still aggressive for a business with this level of reported profitability. $10.1B of debt, 20.32x debt/EBITDA, and $-3.4B FCF are manageable only if utilization, pricing, and customer prepayments continue to scale on schedule. Margin trajectory is improving on management's adjusted view, especially with the Q2 41% adjusted EBITDA margin, but reported annual margins remain deeply negative, which means investors are underwriting a bridge from infrastructure buildout to normalized earnings that has not yet been fully crossed.
Competitive Position
The moat is currently based on scarce AI compute supply, access to power, and speed of deployment. That advantage is real while capacity is sold out and Blackwell pricing is clearing 15% above prior highs, but it is only moderately durable because larger cloud platforms and funded AI infrastructure rivals can replicate hardware over time. The more durable edge would be long-dated customer commitments tied to power contracts, prepayments, and operating know-how across a 5GW power pipeline, but that moat is still being built rather than proven.
Management & Guidance
Management reaffirmed all major FY26 targets, and the specific guide remains ambitious but internally coherent: $7B-$9B annualized run-rate revenue, $3.0B-$3.4B FY26 group revenue, ~40% group adjusted EBITDA margin, $20B-$25B capex, 800MW-1GW connected power by year-end, and 5GW contracted power by year-end. Credibility improved after a strong Q2, especially with ARR moving from $1.9B to $3.0B and premium pricing holding, but guidance was only reaffirmed, not raised, and management also said the new landmark and premium deals will have no material effect on 2026 revenue guidance. That is a prudent signal rather than a red flag, but it means the debate shifts to 2027 conversion and financing rather than another near-term raise in estimates.
Segment Analysis
- Group revenue ($0.5B FY2025): FY2025 reported revenue was $0.5B, but Q2 2026 alone reached $582M, showing the reported base is stale relative to the current run rate.
- Annual recurring / run-rate revenue ($3.0B ARR exiting Q2 2026): ARR rose from $1.9B in Q1 to $3.0B in Q2 as newly added capacity sold through.
- Geography (Not disclosed): Geographic revenue split was not provided. Operating footprint references Finland, Kansas City, Iceland, Vineland, and a US build-to-suit facility, suggesting a growing transatlantic infrastructure base.
Capital Allocation
Capital allocation is aggressively expansionary. The company reported $-3.4B FCF, $5.7B capex in the deep-dive data, no buybacks, and no dividend, which is appropriate for the stage but leaves equity holders fully exposed to buildout execution. Management added $775M of asset-backed debt in July and expects >$9B in customer prepayments for 2026, which looks disciplined in the sense that it matches funding sources to infrastructure assets, but the absolute scale of planned $20B-$25B FY26 capex keeps financing risk elevated.
Management Execution & Track Record
Management's operating trajectory has improved, but the record is mixed enough to keep credibility at moderate, not high. On the 2024 Q4 call, Arkady Volozh said "our projected December, 2025 analyzed run rate revenue of $750 million to $1 billion is well within reach" and the company has now reached $3.0B ARR exiting Q2 2026, so the direction of demand and capacity ramp clearly exceeded that earlier benchmark. The 2024 Q4 team also guided to 100 MW in operations by end-2025 and a heavily 2H-weighted 2025 revenue curve; current results support that the business did ramp materially after that period, but reported FY2025 revenue of only $0.5B landed at the bottom of the old $500M-$700M range and FY2025 profitability was still poor, with -304.8% operating margin and $-3.4B FCF. Recent earnings delivery has been better, with beats of +29.9%, +70.0%, and +83.3% in three of the last four reports, but there was also a major miss of -70.7% on 2026-02-12. The strategy has not changed materially, it is still a speed-and-capacity buildout into AI infrastructure, but the burden of proof remains on converting backlog and ARR into durable cash earnings without balance-sheet strain.
Positioning & Flows
The shareholder base has some institutional sponsorship, with the top 3 institutions holding 37.0%, but insider ownership is only 1.6%, which limits insider alignment as a support signal. Short interest is very high at 29.80%, or 60.2M shares, with 3.4 days to cover, so the stock can squeeze hard on positive AI headlines but that same positioning also reflects serious skepticism about valuation and funding. Options positioning is in a negative gamma regime with zero-gamma at $232.45, close enough to spot to keep near-term trading unstable after a -21.09% one-week move.
Bull Case
- Demand is not the problem: management said it could sell "our entire 2027 capacity" today, ARR rose to $3.0B from $1.9B in one quarter, and Q2 capacity again sold out.
- Pricing is improving, not weakening: the first Blackwell auction cleared at 15% above the prior high price, and management discussed short-term deals at $40M-$50M per MW.
- Execution improved in recent quarters: NBIS beat EPS estimates in 5 of the last 7 reported quarters before the latest Q2 beat streak extension, including +70.0% on 2026-05-13 and +83.3% on 2026-08-12.
- Customer financing support is meaningful: 70% of Q2 landmark deals included upfront prepayments, and customer prepayments are expected to exceed $9B in 2026, which can lower outside capital needs.
Bear Case
- The stock is priced on future perfection: 842.81x P/E, 45.28x P/S, and a quant base value of $178.53, or -18.5% from the current $219.13.
- Cash burn and leverage remain severe: FCF is $-3.4B, total debt is $10.1B, debt/EBITDA is 20.32x, and debt/equity is 0.97 before the biggest capex phase.
- Guidance still implies a funding treadmill: FY26 capex of $20B-$25B dwarfs current revenue scale, and even with structured financing the business depends on sustained access to debt and customer prepayments.
- Trading setup is fragile: short interest is 29.80% of shares, the stock is 46.8% above its 200-day average, and options gamma is negative with zero-gamma at $232.45, which can amplify downside if momentum breaks.
Valuation & Price Target
Engine price target $178.53 (-18.5% vs current), Medium confidence.
- P/E Multiple: $12.06 (1% weight)
- P/S Multiple: $5.86 (2% weight)
- Quality-adjusted: $206.68 (20% weight)
The rating stays aligned with the quant baseline SELL. The engine's $178.53 target is reasonable because even after crediting strong forward indicators, including $3.0B-$3.4B FY26 revenue guidance, ~40% adjusted EBITDA margin, and $3.0B ARR, the equity still trades at 842.81x P/E and 45.28x sales, far beyond the peer set's 22.9x P/E and 2.6x P/S. The quality-adjusted method at $206.68 is closer to the market, which fairly reflects real AI demand and stronger-than-expected Q2 execution, but that still sits below the current $219.13. Sell-side consensus is Strong Buy with 6 buy / 2 hold / 0 sell, but that is low-signal here because the debate is not whether demand exists, it is whether a business with $10.1B debt, 20.32x debt/EBITDA, and $20B-$25B capex deserves this valuation today. It does not.
Risk Assessment
Risk score 59/100 (Moderate). The biggest risk to the SELL call is that FY26 guidance of $3.0B-$3.4B revenue and ~40% adjusted EBITDA margin, plus evidence of 2027 capacity already saleable, could justify a much higher earnings base faster than the current valuation framework assumes.
The Bottom Line
Rated SELL, Medium conviction. Nebius has a real business in a hot part of AI, and the recent quarter gave that story more substance than many skeptics expected. The problem is the stock still asks investors to pay today for a large chunk of 2027 and beyond, while the company is carrying extreme valuation, heavy leverage, and a $20B-$25B annual capex plan. A move materially closer to the $178.53 base value, or hard evidence that 2027 earnings power can absorb this capital intensity without balance-sheet stress, would change the view. Until then, the better call is to avoid chasing the story at this price.
Go deeper in the app
Live charts, options analytics, the full earnings-call breakdown, price alerts, and every other stock — updated in real time.
Get Tickr