$58 IREN Is Pricier Than $208 NBIS — Here's Why

TL;DR

  • All three Neoclouds are up huge over the last year (NBIS +481%, CRWV +70%, IREN +642%). Per-share price doesn't tell you cheap vs expensive — market cap divided by sales (P/S) does
  • P/S comparison: CRWV 6.5x (cheapest) < NBIS 14.1x < IREN 31.8x. On ARR basis, however, all three converge to 5–6x
  • Recommended weights: NBIS 45 / CRWV 35 / IREN 20. Cycle average 45°C (early mid-cycle); only NBIS is at 30°C, the most buy-friendly zone

📅 Published 2026-05-13 · 🏢 Nebius / CoreWeave / IREN · 🎯 "Which AI GPU rental company should I buy?" 12-month horizon

🎯 The Gist

"AI is the new neighborhood, and these three companies are GPU warehouse landlords. NBIS is the well-rounded honor student. CRWV is the giant kid who got too big too fast. IREN is the dark horse — used to mine Bitcoin, now pivoting to AI. They all took the same exam this week, and only NBIS aced it." 🚀

#1 for 12-month stability = NBIS. #1 for 3–5 year multibagger = IREN (up to +1,200% if the NVDA 5GW plan works). CRWV did well, but the market said "we already knew" and gave it the cold shoulder.

💰 Market Cap and P/S — This Is the Real Signal

CompanyShare priceMarket capFY26 revenue estimateP/S multiple
NBIS$208$45B$3.2B14.1x 🟡
CRWV$112$59B$9.0B6.5x 🟢 cheapest
IREN$58$19B$0.6B31.8x 🔴 most expensive*

*IREN's 31.8x is "BTC + AI combined." Its AI Cloud alone trades at roughly 5x P/ARR — actually cheap. IREN is the weird stock where BTC mining is essentially thrown into the market cap for free.

Read (Neocloud peer range 8–12x, general cloud SaaS 5–10x):

  • CRWV 6.5x → below average 🟢 most reasonable price
  • NBIS 14.1x → above average 🟡 expensive, but +684% growth justifies it
  • IREN 31.8x → 🔴 expensive by normal standards, but it's really an option bet on a successful AI pivot

Share price ($58 vs $208) is just a function of how many shares are outstanding. What matters is how many times revenue the entire company costs (market cap).

🏢 What Do These Three Companies Actually Do?

They all do the same thing: "Buy tens of thousands of NVIDIA GPUs, rent them out to AI companies." This business is called Neocloud.

Analogy: "Real estate landlords renting GPU warehouses in the new AI city." Younger cousins to AWS and Azure, but exclusively focused on AI and GPUs.

NBIS = 🥇 "The European Honor Student About to Turn Profitable"

In one line: "Full-stack landlord — owns the warehouses (data centers) + the operating system + the AI marketplace, three tools in one."

  • 🏭 Data centers: Finland, Israel, and a new 1.2 GW site in Pennsylvania (announced this quarter)
  • 💸 Revenue +684% YoY ($51M → $399M)
  • 📈 Operating cash flow turned positive for the first time = starting to actually make money
  • 🇪🇺 Headquartered in Europe → wins business from European governments and enterprises that don't want their data leaving the EU

CRWV = 🥈 "The North American Backlog Champion"

In one line: "Super landlord with 4 years of lease contracts already in the books — but also a lot of debt."

  • 🏭 Data centers: 1 GW active (#1 of the three) · 3.5 GW contracted
  • 💸 Revenue +112% YoY ($2,078M — biggest of the three by far)
  • 📦 Backlog $99.4B — 4 years of future revenue already committed
  • ⚠️ Microsoft is 60%+ of the business — single-customer risk
  • 💰 Debt above $20B — rate hikes hurt

IREN = 🥉 "The Dark Horse Pivoting from Mining to AI"

In one line: "A Bitcoin miner that just announced a 5 GW AI buildout co-developed with NVIDIA itself."

  • 🏭 Data centers: 250 MW (small), but 5 GW plan with NVDA announced (5/7)
  • 💸 Core BTC revenue declining; AI Cloud only $33M per quarter (still small)
  • NVDA received a 30M-share option ($2.1B) = NVDA is betting on IREN's future
  • 🟢 Self-owned renewable power (Texas, Canada) → cheap electricity

🌍 How Big Are They?

ItemNBISCRWVIRENTranslated
Active GPUs (est)80K250K30KCRWV dominates
Active power380 MW1,000 MW250 MWCRWV 3x NBIS
Quarterly revenue$399M$2,078M$145MCRWV 5x NBIS
Market cap$45B$59B$19BCap gap only 1.3x and 3x
Post-earnings move+16% 🟢-19% 🔴 (1 week)+7% 🟢 (D+1)Opposite reactions

Interesting finding: CRWV has 5x the revenue and 3x the GPUs, but only 1.3x the market cap of NBIS. The market is paying NBIS a premium for growth rate and profitability visibility.

🚀 Earnings Diagnostics — All Three Reported Within One Week

🟢 NBIS (5/13 intraday) — Surprise Beat

ItemActualEstimateResult
Revenue$399M$379MBeat +5.3% 🟢
YoY growth+684%Explosive
EPS-$0.23-$0.78Beat by $0.55 🟢
FY26 guide$3.0–3.4B + ARR $7–9BmaintainedConservative

→ Stock gapped up +16% ($179 → $207). "Growth + profitability" achieved together.

🔴 CRWV (5/7) — "Good, but the market cooled"

ItemActualEstimateResult
Revenue$2,078M~$2,050Min-line
YoY growth+112%Strong
Backlog (RPO)$99.4B+$40B new!
New debt$8.5B DDTLConcern

→ Strong numbers, but stock -19% in 1 week. Reason: "already priced in" + "more debt" + "capex too large."

🟡 IREN (5/7) — "Core miss + NVDA mega-news, both"

ItemActualEstimateResult
Revenue$144.8M$219.9MMiss -34% 🔴
AI Cloud revenue$33.6MQoQ +97% acceleration 🟢
EPS-$0.30-$0.22Miss
NVDA partnership$3.4B/5y + 5GW + 30M-share optionMega catalyst 🚀

→ Core revenue blew up, but NVDA news papered over it. Stock +7% (D+1). Market is bidding the pivot option.

🟢🟡🔴 Fundamental Traffic Light Summary

ItemNBISCRWVIREN
Revenue speed🟢 explosive🟢 strong🟡 core ↓ AI ↑
Profitability visibility🟢 close🟡 takes time🟡 BTC noise
Backlog🟢🟢🟢 strongest🟡
Customer diversification🟢 good🔴 MSFT 60%🔴 MSFT+NVDA 84%
Funding🟢 self-funded🟡 debt-reliant🟡 NVDA option dilution
Margin🟡 29% → 40%🟢 56% leader🟡 41% (BTC included)
Composite🟢 excellent🟢/🟡 good🟡 option bet

🚦 Cycle Diagnostics — Where Are We Now?

Neocloud is a capex-cycle industry (AI data centers). In cycle industries, buying expensive at the peak is fatal — so always check "where in the cycle are we."

🌡️ Neocloud Cycle Thermometer(2026-05-13)
NBIS·$207.80CRWV·$111.74IREN·$57.54
0°C
30°C
50°C
70°C
85°C
100°C
45°C
Early mid-cycle (3-name average)
85~100°C · Peak (danger)
+50%~
NBIS$340+
CRWV$180+
IREN$110+
🔴Exit candidate· Cycle end
70~85°C · Overheating
+25~+40%
NBIS$280~340
CRWV$150~180
IREN$85~110
🟠Cut 30~50%· Caution stage
50~70°C · Mid (caution)
+10~+20%
NBIS$230~280
CRWV$120~150
IREN$65~85
🟡Hold / trim· CRWV·IREN zone
★ 지금 여기
30~50°C · Growth (normal)
-15~-25%
NBIS$180~230
CRWV$95~120
IREN$50~65
🟢Scale in· NBIS current zone
0~30°C · Early (cheap)
-40~-60%
NBIS$120~150
CRWV$70~90
IREN$30~45
🟢Aggressive buy· Cycle trough

Per-name cycle diagnostics

  • NBIS at 30°C (trough zone) — Revenue +684% accelerating, first positive OCF, new PA 1.2GW capex → 🟢 buy-friendly
  • CRWV at 55°C (mid) — YoY decelerating (209→175→140→112%), -19% in a week = valuation capitulation → 🟡 hold, scale in on dip
  • IREN at 60°C (volatility peak) — Revenue miss (-34%) + 9.5% dilution option = peakish patterns → 🟡 enter carefully, must average in

💡 One-liner: NBIS = OK to buy now. CRWV = wait for the dip. IREN = scale in slowly.

🧠 The Big Picture — ★ Most Important Section ★

Why did three stocks in the same industry react so differently?

Same earnings week (5/7–5/13), but NBIS +16%, CRWV -19%, IREN +7%. One reason:

🎯 The gap between "market expectation" and "actual results" was different in each case.

  • NBIS: Low expectations → blowout numbers → +5% beat → stock +16% 🟢
  • CRWV: Numbers exactly matched consensus → "we already knew" → no fuel → stock -19% 🔴
  • IREN: Revenue missed by -34%, but NVDA news obscured it → stock +7% 🟡

"Safety vs Optionality" — Which Side Do You Bet On?

StylePickReason
🛡️ Safe growthNBISProfit visible + diversified + accelerating
📦 Want contracted revenueCRWV$99B backlog visibility
🎢 Enjoy volatility / optionsIRENMultibagger if 5GW plan delivers
🌱 Only one nameNBISEven single-name, stable enough

Who's Actually Making Money? (Margin Comparison)

CompanyEBITDA per dollar of revenue
CRWV56 cents 🥇 (most efficient warehouse)
IREN41 cents (BTC included)
NBIS29 cents → 40 cents guided

→ CRWV is the efficiency leader. NBIS is lower because of capex cycle, but ramping fast.

🚀 Multibagger Lens — "Where Is the Maximum Future Value?"

Forget 12-month +13–18%. Switch to "what could rise the most in 3–5 years" and the ranking flips.

If everything goes right (Bull scenario)

NameCurrent capBull-case capPotential multiple
IREN 🚀$19B$240–300B11–14x
NBIS$45B$250–350B5–7x
CRWV$59B$200–300B3–5x

Why IREN has the highest ceiling

  1. Smallest starting point — $19B cap means the same ARR delivers a bigger multiple
  2. NVDA option = "official stamp" — NVDA is committing to buy 30M shares at $70, public endorsement of IREN's future
  3. 5GW sheer scale — if realized, more capacity than CRWV (3.5GW) + NBIS (2GW) combined
  4. Backlog isn't priced in yet — CRWV and NBIS already trade for their future; IREN is the doubter case

But — failure probability is also highest

NameBull probBear lossBull payoffProb × payoff
NBIS35%-36%+500%+175%
CRWV30%-31%+300%+90%
IREN15%-31%+1,200%+180%

→ Probability-adjusted, IREN and NBIS are essentially tied (180% vs 175%).

💎 Single-Name Buy-and-Hold Outcome (Theoretical Upside)

If you buy-and-hold a single $10K position in one name for 5 years, here's the Bull-case theoretical terminal value:

NameEntrySharesBull Terminal$10K → TerminalMultipleBull prob
IREN$57.54~174$700–880 (avg $743)~$129K ($110K–$150K)12.9x 🥇15%
NBIS$207.80~48~$1,250 (5–7x)~$60K ($50K–$70K)6x35%
CRWV$111.74~89~$447 (3–5x)~$40K ($30K–$50K)4x30%

Assumes flat FX. IREN figures already adjusted for 9.5% NVDA-option dilution.

→ At the Bull outcome, IREN delivers 2.1x NBIS and 3.2x CRWV. → But Bull probability is IREN 15% < CRWV 30% < NBIS 35% — payoff potential and realization probability run in opposite directions.

IREN path by year

PeriodShare price$10K position
YE 2026 (8 months)$80–120~$14K–21K
2027–2028$200–350~$35K–61K
2029–2030 (Bull complete)$700–880~$120K–150K

⚠ The table is a theoretical ceiling, not a recommendation to concentrate into one name. The realistic playbook is the next section.

🎯 Two Right Answers

"Stability winner" = NBIS (12-month +18% visible) 🚀 "Multibagger winner" = IREN (+1,200% if 5GW lands)

Recommendation: Core + Option Combo

💰 $10K multibagger allocation
NBIS:  $5–7K (stable core)
IREN:  $2–3K (option bet)
CRWV:  $1–2K (middle)

The single-name buy-and-hold $129K terminal above is the theoretical ceiling at only 15% probability — diversification is the realistic answer. If IREN fails, NBIS cushions you. If IREN delivers, $10K could become $20–40K in the diversified portfolio.

🎯 Action Guide — $10K Portfolio

🚦 Composite Signals

Name12M PTUpsideSignal
NBIS$245+18%🟢 Buy (gapped up — wait 3–5 days, then scale in)
CRWV$115+3%🟡 Hold (scale in below $90)
IREN$62+8%🟡 Hold (scale in near $50)

💰 $10K Simulation — Three Scenarios

12-month forward portfolio value.

StyleNBIS / CRWV / IRENBull (30%)Base (50%)Bear (20%)Prob-weighted
Stable60/25/15$15.2K$10.9K$7.0K$11.3K (+13%)
Balanced (recommended)45/35/20$15.6K$10.8K$6.7K$11.4K (+14%)
Aggressive30/35/35$15.8K$10.5K$6.6K$11.3K (+13%)

→ All three expected returns are similar (+13–14%), but Stable minimizes Bear-case drawdown. Pick Stable for risk aversion, Aggressive if you enjoy volatility.

Buying Only One?

  • 🥇 NBIS — Balanced and profit-visible (if it has to be one name, this)
  • 🥈 CRWV — Stable but limited short-term upside
  • 🥉 IREN — Highest volatility; could 10x if 5GW delivers

👀 Five Things to Watch

#ItemWhenThreshold
1NBIS ARR progressionevery quarterQ2 quarterly ARR >$2B?
2CRWV RPO net addsevery quarterbelow +$15B = decel signal 🔴
3IREN AI Cloud quarterly revenueQ4 FY26 (Jul)>$50M?
4Microsoft share at CRWV·IRENevery quarterbelow 50% relieves discount
5Fed funds ratequarterly>5% strains CRWV debt

🤔 FAQ

Q1. Why are Neoclouds so expensive? A. AWS and Azure can't absorb all the AI GPU demand, so this new "AI-only cloud" category emerged. NVIDIA hands them priority GPU allocation, which is the moat.

Q2. Isn't NBIS expensive at $208 per share? A. ❌ Don't judge by share price. Judge by market cap divided by revenue. NBIS P/S 14x > CRWV 6.5x — NBIS is the more expensive name. But growth-adjusted (PEG), they're closer.

Q3. CRWV has a $99B backlog. Why is the stock falling? A. The market already priced in "$100B backlog." Further upside requires faster conversion of backlog to revenue — if that's slow, the multiple compresses.

Q4. What's the NVDA option on IREN and why is it dangerous? A. NVDA holds the right to buy 30M IREN shares at $70 within 5 years. Full exercise dilutes existing shareholders by ~9.5%. But it also injects $2.1B in cash. Two-sided.

Q5. Can I buy all three? A. Yes — diversification works because business mixes differ. But all three are tied to AI macro, so a bubble pop hits them together.

Q6. Buy now or wait? A. NBIS gapped up — wait 3–5 days, then scale in. CRWV: buy below $90. IREN: scale in near $50 (pre-NVDA-announcement level).

Q7. What happens to all three if the AI bubble pops? A. -31% to -36% across the board in the Bear case. Diversification within Neocloud doesn't help much. Cap AI exposure at 50% and put the rest in defensive sectors.

🚨 Risk Warnings

  1. AI bubble concerns — Hyperscaler in-house GPUs (TPU/Trainium) ramping could slow Neocloud demand (probability ~30%)
  2. Rate hikes — CRWV most exposed, $20B+ debt
  3. Customer concentration — CRWV (MSFT 62%), IREN (MSFT+NVDA 84%). Customer loss = big shock
  4. NBIS short-term overbought — MA200 stretch +99.5%; profit-taking pressure
  5. NVDA next-gen GPU delays — Blackwell Ultra / Rubin slip hits all three
  6. Geopolitics — NBIS has legacy Yandex/Russia overhang

📚 Jargon Glossary (one-liner each)

  • Neocloud: NVIDIA-GPU-focused new cloud = "AI-exclusive GPU rental"
  • RPO/Backlog: Contracted revenue not yet recognized = "list of reserved customers"
  • ARR (Annualized Run Rate): Last-month revenue × 12 = "extrapolated annual revenue at current pace"
  • Adj EBITDA: Operating cash earnings before depreciation = "real cash earned after stripping out factory build costs"
  • P/S (Price/Sales): Market cap ÷ revenue = "how many times revenue you pay for the whole company"
  • SOTP (Sum of the Parts): Sum values of each business segment = "count each pocket separately"
  • GW (gigawatt): Data center power unit = "electricity for a million households at once"
  • GB200 NVL72: NVIDIA's 72-GPU rack system = "next-gen inference GPU cluster"
  • Inference vs Training: Using a model (post-graduation) vs building it (study) = "graduation vs studying"

📌 Bottom Line

NBIS = stability winner over 12 months. IREN = multibagger winner over 3–5 years. CRWV is the middle. Balanced $10K = NBIS 45% + CRWV 35% + IREN 20%. Single name = NBIS.

⚠️ Final Disclaimers

  • This report is educational only, not buy/sell advice
  • All decisions are your responsibility — never invest borrowed money
  • Earnings and guidance change quickly; refresh information after the report date
  • Price data is as of the report date — check live prices via your broker
  • NBIS data is intraday from the earnings day and may revise after close

Analysis date 2026-05-13 · Next earnings: NBIS Q2 2026 (Aug est.) / CRWV Q2 2026 (Aug) / IREN Q4 FY26 (Jul–Aug)

FAQ

Why isn't CRWV the Top Pick when its revenue is 5x NBIS?

Quarterly revenue at CRWV ($2,078M) is 5x NBIS ($399M), but its market cap is only 1.3x NBIS. The market awards NBIS a premium for +684% YoY acceleration, the operating cash flow flip to positive, and Microsoft concentration below 20%. CRWV leads the backlog scoreboard with $99B RPO, but the -19% one-week post-earnings drop tells you the visibility is already priced in. NBIS still has incremental fuel; CRWV has to convert backlog into revenue faster than expected to add upside from here.

Isn't NBIS the most expensive at $208 per share?

No — that's the price trap. Per-share price depends on how many shares were issued. What matters is market cap (price × shares). NBIS market cap is $45B, smaller than CRWV ($59B). For true expensiveness, use market cap divided by revenue (P/S). On that metric, IREN at 31.8x is the most expensive of the three. But IREN's number includes BTC mining — its AI-only P/ARR is closer to 5x, which is actually cheap.

What's this NVDA option on IREN, and is it dangerous?

NVIDIA received the right to buy 30M IREN shares at $70 per share within 5 years. If fully exercised, it dilutes existing shareholders by ~9.5%. But the exercise would inject $2.1B in cash that funds IREN's 5GW data center build. The market reads the option itself as NVIDIA's public bet on IREN's future — that's why the stock rose despite the revenue miss.

If the cycle peaks, do all three really drop 30-50%?

In the Bear case (20% probability), we model -31% to -36% across all three simultaneously. Bubble triggers include hyperscaler in-house silicon (TPU, Trainium) acceleration, simultaneous capex guidance cuts, backlog cancellations, or Fed funds above 5%. Diversification within Neocloud is limited — keep AI exposure below 30-40% of your total portfolio.

Microsoft is over 60% of CRWV and IREN revenue — is that really dangerous?

Yes. Single-customer concentration is one of the largest risks in the Neocloud space. CRWV is 62% Microsoft, IREN is 84% MSFT+NVDA combined. A single quarter of reduced Microsoft orders could shock the stock. CRWV is mitigating it — non-MSFT new contracts added $15B last quarter, OpenAI and IBM are entering — and a glide path from 60% toward 50% over 12-18 months would unwind the discount. IREN gained NVDA as a second anchor alongside MSFT (which exercised its 2027 expansion option in Q1), so the customer base is now two-pillared. Even so, keep any single-name position below 15% of your portfolio.

Should I buy now or wait?

NBIS gapped up +16% on earnings, so wait 3-5 days for short-term consolidation and scale in. CRWV becomes attractive below $90 (currently $112). IREN around $50 (where it traded before the NVDA announcement) is a good entry. Average cycle reads 45°C — early mid-cycle, not imminent peak — so 4-6 week scaling-in works.

What if the AI bubble pops?

All three drop together. Bear case is -31% to -36% across the board. Diversification within Neocloud doesn't help much. If you're worried about the bubble, cap AI exposure at 50% or less and put the rest in dividend-paying or defensive sectors. A $10K balanced portfolio would draw down to roughly $6.7K (-33%) in that scenario.