Date: 2026-02-22 Quarter: Q4 FY25 (Dec-25) Market cap: ~$24.6B | EV/FY26 Rev: ~7.8x | Revenue growth: 547% YoY Prior coverage: None (first analysis)
The setup here is genuinely unusual. Nebius is a company I've had on my radar but never written about. No prior beliefs to test, which means I'm coming in fresh — and that cuts both ways.
Revenue came in at $227.7m — up 547% year-over-year and 56% sequentially. Core AI was $214.2m, now 94% of total revenue. The ARR run-rate hit $1.25B, up from just $90m a year ago. That's a 14x increase in 12 months. Let that sink in.
And they flipped to positive EBITDA for the first time — $15m at the group level. Core AI Adj EBITDA came in at $51.8m on a 24% margin, expanding from 19.1% last quarter. Gross margin is 70% GAAP. These are genuinely good numbers.
The deferred revenue balance is $1.578B. That's seven to eight quarters of current run-rate revenue already pre-paid by customers. Combined with $1.25B ARR, Nebius has more contracted demand on the books than they've generated in their entire public history.
EPS came in at -0.68vsconsensusof−0.54. A miss. But the EPS number is being dragged by the CapEx ramp, not by a deteriorating business.
This is where it gets interesting — and where I have to be honest about both sides.
What "is" delivering:
What's "could be":
Here's my concern on the EBITDA guidance: CapEx went from $544m in Q1 to $511m in Q2 to 956minQ3to * *2.056B in Q4**. That's a hockey stick nobody expected — management guided $1.5B for all of FY25 and spent closer to $4B. They guided $5B CapEx at the Q3 call, and then Q4 alone was $2B. If CapEx runs at even half that pace in FY26, Adj EBITDA margins of 40% are technically achievable (D&A isn't in Adj EBITDA), but FCF will be deeply negative.
Promising a 40% Adj EBITDA margin while withholding CapEx guidance is... selective. I don't think it's dishonest, but it's incomplete. I need management to clarify this.
The transcript wasn't published at the time of this analysis (as of Feb 21). That's a real gap. Management quotes matter here. The CapEx story needs explaining.
This is where I pump the brakes a bit. I don't have a long track record with Nebius management. What I do know:
To be fair, a company growing this fast will miss its own forecasts — usually to the upside on demand and to the upside on cost. That's not necessarily sinister. But the lack of CapEx guidance in an EBITDA-margin-guided year is something I'll be watching closely.
This is the most important analytical point and I want to spend time on it.
ARR run-rate is 1.25B.AnnualisedQ4revenueis 910m. That gap of ~$340m is contracted-but-not-yet-recognized revenue. Deferred revenue of $1.578B means customers have already paid Nebius for services not yet delivered — that's 7 quarters of current run-rate.
The reason? They're signing contracts faster than they can deploy capacity. Active power only moved from 100 MW to 170 MW in Q4 — the infrastructure to serve these contracts is still being built. Contracted power sits at 2.0 GW. They're operating 170 MW. There's a massive gap that will be filled as data centers come online.
If the FY26 trajectory is $3.0-3.4B, and deferred revenue of $1.578B is already pre-paid, a big chunk of that revenue is already won. This is not the same as "management says we'll grow 6x next year" with no contract backing. The demand side looks real.
| | Q324 | Q424 | Q125 | Q225 | Q325 | Q425 | | | Sep-24 | Dec-24 | Mar-25 | Jun-25 | Sep-25 | Dec-25 | |---|---|---|---|---|---|---| | Revenue ($m) | 32.1 | 35.2 | 50.9 | 105.1 | 146.1 | 227.7 | | YoY % | 542% | 425% | 347% | 625% | 355% | 547% | | QoQ % | 121% | 10% | 45% | 107% | 39% | 56% | | Core AI Rev ($m) | 27.1 | — | 38.2 | 93.6 | 131.1 | 214.2 | | Core AI % | — | — | 75% | — | 90% | 94% | | Gross Margin [GAAP] | — | — | — | — | — | 70.0% | | Op Margin % [GAAP] | -251% | -430% | -254% | -106% | -89% | -103% | | EBITDA (m)|−45.9|−75.5|−62.6|−21.0|−5.2|+15.0||CoreAIAdjEBITDAMargin|—|—|—|—|19.1|ARRRun − Rate(m) | — | 90 | 249 | 430 | 551 | 1,250 | | CapEx ($m) | — | 417.6 | 544.0 | 510.6 | 955.5 | 2,056.0 | | Data Centers | — | 1 | 2 | 4 | 4 | 7 | | Active Power (MW) | — | 100 | 100 | 100 | 100 | 170 |
Atlas gave this a Conviction 4/5 and landed on "fair-to-cheap at 7.8x FY26 revenue if execution holds." I don't disagree with the framing, but I want to add a few nuances from my own lens:
The EPS miss matters less than the CapEx story. Atlas flagged the -$0.14 miss. I think that's secondary. The real question is: can management reconcile $2B/quarter CapEx with 40% EBITDA margins in the same year? I want to hear that directly from the call.
Management credibility is an open question. Atlas noted the ARR guidance volatility. For me, miss-and-lower or guide-volatility gets weighted heavily. FY26 ARR guide went from $1.0B → $9.0B → actual $1.25B exit. I can't tell if the $7-9B FY26 exit target is aggressive, conservative, or noise. Without the transcript I can't score management credibility.
The deferred revenue is bullish but comes with concentration risk. $1.578B pre-paid sounds great. But what if it's two or three customers? Customer concentration at this scale could be a significant fragility. We don't have that data.
The "AWS of AI" thesis is real — but so is CoreWeave. Nebius is #2 in neoclouds. CoreWeave is 10x larger in GPU fleet. In infrastructure, scale matters. I wouldn't understate that.
At ~97/shareand 24.6B market cap:
The TTM number is misleading — Nebius was building infrastructure all year and revenue was ramping from a low base. The forward multiple is what matters, and 7.8x for a company that might genuinely do 5-6x revenue growth is not crazy. Wall Street consensus is 6 buy, 1 hold, average PT ~$148 (~50% upside from $97).
I'm not a DCF guy. But the $1.578B in deferred revenue alone is meaningful collateral at this valuation. The risk/reward skews positive if FY26 execution is even in the ballpark.
Here's how I'm thinking about it: Nebius is delivering genuinely exceptional numbers. The growth is real, the contracts are real, and the margin direction is real. But this is a company in the middle of a massive infrastructure build with no CapEx guidance for the year they're promising 40% EBITDA margins. That's a gap I can't resolve without the earnings call transcript.
The "is": 547% growth, EBITDA positive, $1.578B pre-paid, 14x ARR growth in 12 months.
The "could be": FY26 $3.0-3.4B, 40% margins. Credible given deferred revenue, uncertain given CapEx trajectory.
For me, this is a tryout. The growth is extraordinary and the deferred revenue visibility is compelling. But I'd want to see: (1) transcript with CapEx clarification, (2) at least one quarter of FY26 execution, and (3) any customer concentration data. At that point, this could earn a starter position.
I could be wrong. The momentum here is strong enough that waiting costs real upside. But the CapEx ambiguity is genuine, and in my experience, when management avoids a disclosure, there's a reason. Watch and see.
As usual, thanks for reading.
First analysis. No prior corpus writings on NBIS to reference. Atlas baseline: Conviction 4/5, "fair-to-cheap if FY26 executes." Joe alignment: largely agree — tryout rather than starter pending CapEx clarity and transcript. Data source: Scout brief NBIS_earnings-review_2026-02-21. Transcript not yet published as of analysis date.