NBIS
Nebius Group N.V.
AI FOUNDRY / NEOCLOUD · INFRASTRUCTURE
~$176–193
NASDAQ · ~$44–49B mkt cap 1
52W: $27.20 – $197.89
Next earnings: 13 May 2026 (pre-market)
⬡ WATCH / SPEC ADD
Entry zone: $120–145 · Do not add above $160
The market is pricing NBIS as a momentum neocloud at ~13–14× forward 2026 revenue, when the correct frame is a capital-compounding AI utility: $46B+ in contracted backlog from Meta and Microsoft, Nvidia preferred-partner status, and a hidden NAV floor of ~$7–8B in portfolio assets (ClickHouse ~28%, Avride ~83%) that most screeners ignore. The variant perception is not growth — it is balance-sheet architecture and the structural premium of being the only neocloud with owned European data sovereignty infrastructure as the EU AI Act tightens.
TTM P/E
N/M
Net loss $447M FY25
Fwd P/S (2026E)
~13–15×
$3.0–3.4B guide
Rev Growth (Q4 YoY)
+547%
$227.7M vs $35.2M
Gross Margin
70%
Q4 2025 (stable QoQ)
Adj EBITDA Mgn
24%
Q4 core AI; ~40% guided 2026
FCF Margin
Neg
CapEx $16–20B 2026E
Cash Position
$3.7B+
YE25; ~$4.9B latest est.
2026 ARR Target
$7–9B
YE2026; $1.2B at YE25
Next Earnings
13 May
Pre-market Q1 2026
Valuation vs Growth
C+
TTM P/S: ~87× — irrelevant; use forward. Fwd P/S 2026E: ~13–15× vs CoreWeave at ~1.9×. CoreWeave's premium discount reflects debt load, not necessarily business quality. Nebius's cleaner balance sheet justifies some premium but 13–15× fwd P/S vs a peer at 1.9× is a 7× valuation gap for ~4× more projected growth — not obviously justified.
EPS guidance 2026E: consensus −$2.17 to −$3.95. No P/E anchor possible. Valuation is purely revenue + EBITDA multiple.
2026 consensus rev: $3.1–3.4B (company guide). 2027 analyst consensus: ~$10.1B (+210% YoY). If 2027 ARR target met at $7–9B, fwd P/S on current price compresses to ~4.5–6× — reasonable for a durable infrastructure compounder.
PEG: Not calculable (losses). Revenue-based PEG analog: fwd P/S / 2-yr CAGR ≈ 14×/250% ≈ 0.056 — optically compelling, but only valid if execution on the $16–20B CapEx plan lands.
EPS guidance 2026E: consensus −$2.17 to −$3.95. No P/E anchor possible. Valuation is purely revenue + EBITDA multiple.
2026 consensus rev: $3.1–3.4B (company guide). 2027 analyst consensus: ~$10.1B (+210% YoY). If 2027 ARR target met at $7–9B, fwd P/S on current price compresses to ~4.5–6× — reasonable for a durable infrastructure compounder.
PEG: Not calculable (losses). Revenue-based PEG analog: fwd P/S / 2-yr CAGR ≈ 14×/250% ≈ 0.056 — optically compelling, but only valid if execution on the $16–20B CapEx plan lands.
Priced to perfection on a trailing basis; conditionally justified on forward 2027 multiples — but the multiple gap vs CRWV needs a better explanation than "balance sheet is cleaner."
Fundamentals vs Hype
B
Revenue quality: High. 94% from AI cloud. $46B+ in contracted backlog (Meta $27B, Microsoft $19.4B) — legally binding, not LOIs. Sold out through Q1 2026. Pipeline >$4B.
Gross margin trajectory: Q1 2025: 30% → Q4 2025: 70%. Structural expansion from fixed-cost leverage; guided to remain ~70%+ as utilisation rises.
FCF: Deeply negative. Q4 operating CF was +$834M (inflated by customer prepayments) but CapEx was $2.06B — net burn >$1.2B that quarter. FY26 CapEx guide $16–20B vs rev guide $3–3.4B = capital intensity ratio ~5–6×. This is not a FCF story until 2027–2028.
AI revenue measurability: Excellent — 100% of core revenue is GPU-hours billed. No narrative-only component.
Red flags: EPS miss Q4 ($−0.68 vs −$0.54 est.). Convertible issuance overhang. Avride/ClickHouse portfolio values are ⚠ unverifiable from open-access sources — company's own estimates. Yandex lineage creates ESG/governance perception risk for certain LP mandates.
Gross margin trajectory: Q1 2025: 30% → Q4 2025: 70%. Structural expansion from fixed-cost leverage; guided to remain ~70%+ as utilisation rises.
FCF: Deeply negative. Q4 operating CF was +$834M (inflated by customer prepayments) but CapEx was $2.06B — net burn >$1.2B that quarter. FY26 CapEx guide $16–20B vs rev guide $3–3.4B = capital intensity ratio ~5–6×. This is not a FCF story until 2027–2028.
AI revenue measurability: Excellent — 100% of core revenue is GPU-hours billed. No narrative-only component.
Red flags: EPS miss Q4 ($−0.68 vs −$0.54 est.). Convertible issuance overhang. Avride/ClickHouse portfolio values are ⚠ unverifiable from open-access sources — company's own estimates. Yandex lineage creates ESG/governance perception risk for certain LP mandates.
Grounded where measurable — revenue quality and margin trajectory are real. The FCF hole is large but a known feature of the infrastructure build model, not a concealed problem.
Institutional vs Retail Flow
B+
Institutional ownership: 660 filers, ~105.9M shares. Key names: Orbis/Allan Gray, Fred Alger, Susquehanna, Citadel, Jericho Capital, Jane Street, Accel Leaders 4. BlackRock added 9.43M shares (38,000%+ position increase) as of Dec 31 2025 13F-HR.
Ownership trajectory: ~38.36% → 44.6% institutional ownership increase (+6.24%) despite a 33.6% price crash — institutions accumulated on weakness. Fintel estimates ~46.3%. Low relative to large-cap norms (60–80%) due to unusual Yandex IPO history — mechanical upside as index inclusion occurs.
Analyst coverage: Rapid expansion. 16–20 analysts now covering. Initiations: Citi Buy (Mar 2026), BofA Buy (Mar 2026), Cantor Overweight (Apr 2026), Morgan Stanley Equal Weight (Jan 2026), Wolfe Peer Perform (Apr 2026). Average PT: $154–177 range; high $291.
Retail: WallStreetBets coverage active. Beta 3.13 — high retail participation amplifies volatility in both directions. Short borrow rates elevated ⚠.
Insider activity: No material open-market buying or selling flagged in recent 90-day open-access data ⚠.
Ownership trajectory: ~38.36% → 44.6% institutional ownership increase (+6.24%) despite a 33.6% price crash — institutions accumulated on weakness. Fintel estimates ~46.3%. Low relative to large-cap norms (60–80%) due to unusual Yandex IPO history — mechanical upside as index inclusion occurs.
Analyst coverage: Rapid expansion. 16–20 analysts now covering. Initiations: Citi Buy (Mar 2026), BofA Buy (Mar 2026), Cantor Overweight (Apr 2026), Morgan Stanley Equal Weight (Jan 2026), Wolfe Peer Perform (Apr 2026). Average PT: $154–177 range; high $291.
Retail: WallStreetBets coverage active. Beta 3.13 — high retail participation amplifies volatility in both directions. Short borrow rates elevated ⚠.
Insider activity: No material open-market buying or selling flagged in recent 90-day open-access data ⚠.
Institutionally accelerating but not yet institutionally dominated — BlackRock-led accumulation on weakness is a strong signal. Retail momentum is a double-edged amplifier at current ATH levels.
Position Verdict
Watch / Speculative Add (sub-$145). This is not a current holding — treat as a fresh entry decision. The thesis is structurally sound: contracted backlog exceeds market cap, NVIDIA preferred-partner access is a genuine moat in a GPU-constrained world, and 70% gross margins at the infrastructure layer are exceptional for a company 18 months into its post-Yandex life. The Eigen AI acquisition ($643M) moves Nebius up-stack toward inference optimisation — if executed, it transforms the unit economics of the Token Factory from compute-margin to software-margin.
However, at $176–193 the price is above any defensible entry zone. The stock has run +110% YTD and is printing ATHs the day before earnings. Tomorrow's Q1 print (13 May, pre-market) is the right frame: consensus expects ~$316–389M revenue. If print exceeds $400M and ARR trajectory confirms >$2B run-rate, re-evaluate at post-print price.
Do not add above $160. Entry zone: $120–145 represents a ~15–20% pullback from pre-earnings levels, where the 2026 forward P/S compresses to ~10–11× — closer to a defensible infrastructure compounder valuation. Set a watchlist alert and wait for a catalyst-driven dip rather than chasing ATHs.
However, at $176–193 the price is above any defensible entry zone. The stock has run +110% YTD and is printing ATHs the day before earnings. Tomorrow's Q1 print (13 May, pre-market) is the right frame: consensus expects ~$316–389M revenue. If print exceeds $400M and ARR trajectory confirms >$2B run-rate, re-evaluate at post-print price.
Do not add above $160. Entry zone: $120–145 represents a ~15–20% pullback from pre-earnings levels, where the 2026 forward P/S compresses to ~10–11× — closer to a defensible infrastructure compounder valuation. Set a watchlist alert and wait for a catalyst-driven dip rather than chasing ATHs.
New position: 0% now. If entry into $120–145 occurs post-earnings dip: medium conviction 10–15% max given execution risk and price level. Do not add above $160. Hard cap 20% single stock.
Key Risks
▾
- R1 Execution on $16–20B CapEx plan. Nebius must connect 800MW–1GW of capacity in 2026 across 16 data centres on 3 continents simultaneously. A single major site delay misses the $7–9B ARR target. At current valuation, a 30% ARR miss = >40% stock drawdown based on comparable infrastructure growth-to-multiple re-rates.
- R2 Customer concentration — Meta + Microsoft = ~92% of backlog. The $27B Meta contract is for Nvidia Vera Rubin infrastructure starting early 2027 — any delay in Vera Rubin availability cascades directly to Nebius revenue recognition. If Meta capex growth decelerates post-2026, renewal risk is non-trivial.
- R3 Dilution overhang. $16–20B CapEx on $3–3.4B revenue requires continuous external capital. ATM program (25M Class A shares) is live. Convertible debt already issued. At $46B mkt cap, each additional 5% dilution = ~$2.3B book value transfer.
- R4 Yandex geopolitical overhang — ESG/compliance exclusion risk. Despite CEO Volozh's Russian citizenship renunciation and OFAC clean-break status, some LP mandates and ESG-screened institutions cannot hold NBIS. This structurally caps institutional ownership below 60–70% norms and creates periodic forced-selling episodes. A real multiple limiter vs CRWV.
- R5 Hyperscaler in-sourcing — the neocloud window may be temporary. AWS, Azure, and Google are expanding GPU capacity aggressively. By 2028–2029, their own Blackwell and post-Blackwell deployments may satisfy 70–80% of demand currently outsourced to neoclouds. Current stock price implies a 5+ year premium franchise — not the 3-year neocloud boom scenario.
DCF Sanity Check
▾
Discount rate: 15% — speculative / pre-profitability profile; high execution risk, capital intensity, and Yandex governance overhang. Terminal growth: 5%. FCF modelled from adj. EBITDA minus CapEx maintenance from 2028 onward, assuming CapEx normalises to ~30% of revenue as infrastructure matures.
| Assumption | Value | Note |
|---|---|---|
| 2026E Revenue | $3.4B | Top of guide |
| 2027E Revenue | $10B | Analyst consensus |
| 2028–2030 CAGR | 35% | Deceleration from hyper-growth |
| Normalised FCF Margin (2028+) | 18% | After CapEx normalisation; adj EBITDA 40% - maintenance CapEx |
| Discount Rate | 15% | Pre-profitability speculative |
| Terminal Growth | 5% | Infrastructure compounder |
| Implied Fair Value | ~$130–160/share | Bull case: full execution |
⚠ Current price ($176–193) is above the bull-case DCF range. The market is pricing in a scenario more optimistic than the bull case — implying either a lower discount rate or higher terminal growth assumptions than are defensible.
| Assumption | Value | Note |
|---|---|---|
| 2026E Revenue | $2.5B | 30% miss vs guide midpoint |
| 2027E Revenue | $5B | Neocloud window compression |
| 2028–2030 CAGR | 20% | Hyperscaler in-sourcing headwind |
| Normalised FCF Margin (2028+) | 10% | Margin shortfall; ongoing dilution |
| Discount Rate | 15% | Pre-profitability speculative |
| Terminal Growth | 4% | Commoditising compute |
| Implied Fair Value | ~$50–70/share | Bear case: execution shortfalls |
Bear case implies 65–75% downside from current levels. Only requires one of: major CapEx delay, Meta contract restructuring, or AI capex deceleration from hyperscalers by 2027.
SUMMARY: Current price is above both scenarios. Bull case: ~$130–160. Bear case: ~$50–70. Current: ~$185. The stock is pricing in either a sub-10% discount rate or terminal growth assumptions of 7%+, neither of which is defensible for a pre-profitability, capital-intensive neocloud.
Analyst Quotes (Tracked Sources)
▾
"Nebius is the purest Neocloud and AI-infra asymmetry left. This is the highest revenue Neocloud… with no plaguing uncertainty that $IREN, $ORCL face from full-stack execution. No high interest debt that $CRWV, $APLD, and others face… $8B midpoint ARR next year, $4.7B+ in cash, diversification in enterprise clients from Meta, Microsoft, Accenture, Shopify, Governments, hyper-growth portfolio companies, and a proven full-stack high-margin business."
Serenity (@aleabitoreddit) · May 2026 · Price at time: ~$86–92 · PT: $400 / 1Y from that post
"$NBIS vs. $IREN. The difference is night and day. Nebius: $2B dilution from $NVDA, zero immediate selling pressure to the public float… Very clear which company leads to higher share value appreciation from capex financing. One is strategic with Nvidia, the other is toxic financing."
Serenity (@aleabitoreddit) · 11 March 2026 · Commenting on NVDA $2B strategic investment
"Incredibly undervalued and detached from fundamentals. $7–9B forward ARR, 20–30% EBIT, enterprise contracts from Shopify, Accenture, Cursor, foreign governments and hyperscaler contracts from Meta and Microsoft give Nebius revenue visibility. With $4.8B+ in cash, it's isolated from credit tightening affecting data centres."
Serenity (@aleabitoreddit) · November 2025 · Rating: Strong Buy
Beth Kindig / IO Fund (@bethkindig) — No public commentary on NBIS found on io-fund.com free tier or X as of May 2026. NBIS does not appear in IO Fund's tracked portfolio.
Dylan Patel / SemiAnalysis — No coverage of NBIS found on newsletter.semianalysis.com free tier as of May 2026. SemiAnalysis focus is on semiconductor supply chain; neocloud operators are outside their typical scope.
NOTE: Serenity's $400 PT was set at ~$86–92 stock price. At current ~$185, the implied remaining upside is ~115%. His analysis is supply-chain-grounded and has been directionally correct on NBIS. However, his PTs are not risk-adjusted; they reflect conviction scenarios, not probability-weighted estimates.
Monitoring Triggers
▾
| Trigger | Bull Signal | Bear Signal | Timing |
|---|---|---|---|
| Q1 2026 Revenue | ≥$400M | <$280M | 13 May 2026 |
| Q1 ARR Exit Rate | ≥$2.0B run-rate | <$1.6B | 13 May 2026 |
| Missouri DC energisation | On-track for Q3 2026 | Delay flagged | Q1 call / Q2 2026 |
| Meta contract delivery cadence | Rubin GPU deployment confirmed | Pushout to 2028 | H2 2026 onwards |
| Adj EBITDA margin vs 40% guide | Q2/Q3 tracking ≥30% | <20% — cost overruns | Q2 Aug 2026 |
| Dilution events (ATM draws) | ATM <$1B total drawn | ATM >$3B or new converts | Continuous |
Portfolio Context & What NOT to Do
▾
Not a current holding. No sizing decision is active — this is a fresh coverage initiation.
Where NBIS fits in the framework: AI Foundry & Connectivity tier (neocloud / GPU infrastructure). No overlap with existing positions. Would sit alongside TSM and AVGO as a compute-layer name.
Relative to Stage framework: NBIS is a Stage 3 / speculative-tactical position — higher risk profile, requires earnings validation and a price pullback to enter. Not Stage 1 (TSM, NOW, GOOGL) or Stage 2 (AVGO, NVDA).
The timing problem: Tomorrow's Q1 print is the single biggest near-term risk. Entering a pre-profitability ATH stock the day before earnings is the textbook "thesis correct, entry timing wrong" error. Wait for the print. If NBIS beats and rips higher, the entry discipline says pass and look for the post-euphoria fade. If NBIS misses, the $120–145 entry zone becomes live.
What NOT to do:
- Do not FOMO into ATHs pre-earnings. A 547% YoY growth company just hit all-time highs — the easy money has been made.
- Do not use Serenity's $400 PT as a price target — it was set at $86–92 and has not been updated for the current $185 level.
- Do not treat the $46B contracted backlog as equivalent to revenue visibility — $27B of that is 5-year Meta infrastructure for Vera Rubin starting 2027, contingent on GPU delivery schedules.
- Do not size this as a high-conviction position. If entering: medium conviction (10–15%) is the maximum given execution risk and price level.
Sources
SEC EDGAR 6-K (Feb 12, 2026) · Stockanalysis.com · Macrotrends.net · Yahoo Finance · Fintel.io (institutional 13F filings) · Serenity @aleabitoreddit (X, Mar–May 2026) · InsiderFinance · TradingView · Robinhood / Capital.com (price data) · Q1 revenue consensus range $316.9M–$389M across providers (Zacks vs others) — will be resolved at 13 May print.
⚠ = figure not independently verifiable from open-access sources. Portfolio asset valuations (ClickHouse ~$4.2B, Avride ~$2.2–2.3B) sourced from third-party analysis; treat as indicative, not confirmed. FCF data ambiguous: Q4 operating CF of +$834M was partially inflated by customer prepayments.
SEC EDGAR 6-K (Feb 12, 2026) · Stockanalysis.com · Macrotrends.net · Yahoo Finance · Fintel.io (institutional 13F filings) · Serenity @aleabitoreddit (X, Mar–May 2026) · InsiderFinance · TradingView · Robinhood / Capital.com (price data) · Q1 revenue consensus range $316.9M–$389M across providers (Zacks vs others) — will be resolved at 13 May print.
⚠ = figure not independently verifiable from open-access sources. Portfolio asset valuations (ClickHouse ~$4.2B, Avride ~$2.2–2.3B) sourced from third-party analysis; treat as indicative, not confirmed. FCF data ambiguous: Q4 operating CF of +$834M was partially inflated by customer prepayments.