Neocloud Brief · neocloud stocks · what is a neocloud · neocloud ETF
CRWV — CoreWeave: the leverage-built AI landlord
The purest expression of the neocloud trade: borrow tens of billions to buy Nvidia’s best GPUs, rent them out on multi-year contracts, and hope the buildout outruns the balance sheet.
01The 30-second read
Fast read
Demand is not the question for CoreWeave — a $104.2B contracted backlog (+264% YoY) means years of revenue are already signed. The question is whether the balance sheet survives the buildout: ~$35B of debt, negative free cash flow, and a $35–39B 2026 capex plan against a $12.4–13.2B revenue guide.
Bulls see an index on AI compute with a ~$144 consensus target (~+74% upside). Bears see a financing machine priced for perfection. Both are right about different things.
02What CoreWeave actually is
CoreWeave rents high-end GPU capacity to AI companies through a cloud platform built for speed, not general-purpose workloads. Think “AI data-center landlord”: sign long-term contracts with AI labs and enterprises, then design, build, and operate liquid-cooled bare-metal GPU clusters — financed with debt, like a utility builder.
It went public in March 2025 at $40/share and has more than doubled since, even after a violent drawdown. It skips most of the software stack: no broad cloud services, just optimized GPU density, power, and deployment speed.
Nvidia is simultaneously its largest supplier, a $2B equity investor, and a strategic partner on a $6.3B collaboration to build 5+ GW of “AI factories” by 2030. That intimacy means every Nvidia product cycle (Grace Blackwell → Vera Rubin) moves the stock — in both directions.
03The bull case
1 · Backlog is the story
- $104.2B contracted backlog as of Jun 30, 2026 (+264% YoY); delivery begun on more than half
- +$25B net new commitments added early Q3, outside the June figure
- ~5 years of guided exit run-rate ($18.5–19.5B) already on contract
2 · Pricing is inflecting up
- ~25% price increase across SKUs in July 2026
- New contracts carry contribution margins 5–10pp above recent quarters
- First to deploy Vera Rubin NVL72 — the scarcest silicon, while demand broadens into enterprise, defense, and industrial AI
3 · Margins are scaling
- Q2 2026 adjusted EBITDA ~$1.5B at ~59% margin
- Each new deployment is a smaller slice of a growing base — operating leverage compounds
4 · The Street is still constructive
- Consensus Buy / “Moderate Buy” — 38 analysts: 5 Strong Buy, 21 Buy, 10 Hold, 1 Sell, 1 Strong Sell
- Average target ~$144, range $41–$250 — an enormous spread
04The bear case
1 · Debt is the business model
- ~$35B total debt at Q2 2026, up sharply YoY
- D/E 5.53× (brief convention) up to ~8.9× (some screens) — deeply levered either way
2 · The funding treadmill never stops
- $3B convertible notes due 2033 + ATM share sale up to ~$2.9B (35M shares) — in market now
- Second major debt raise of 2026, after 9.25% and 9.0% senior notes totaling $3.75B
- Convertibles and share sales dilute; coupons are heavy
3 · The cash math is tight
- Q2 capex $9.4B; FCF −$5.74B
- Interest expense nearly tripled to $640M (from $267M)
- Reported credit terms: cash lockboxes, a 1.35× debt-service-coverage covenant, full amortization
4 · Concentration, overhang, and the tape
- Meta and OpenAI dominate the backlog — one contract dispute lands hard
- Securities fraud class action alleging concealed construction delays (reported; unproven)
- 11 downward Q3 EPS revisions vs. 3 upward; stock −21% / month, −30% / year, ~45% off the high
05The numbers
| Metric | Value | Note |
|---|---|---|
| Q2 2026 revenue | $2.575B | +112.5% YoY |
| Q2 net loss | ~$654M | −25.41% net margin |
| TTM EPS | −$2.68 | Loss-making |
| Q2 adj. EBITDA | ~$1.5B | ~59% margin |
| Free cash flow (Q2) | −$5.74B | |
| Cash | ~$5.5B | |
| Total debt | ~$35B | LT $27.56B + ST $7.51B |
| Debt / equity | 5.53× | ~8.9× on some screens |
| Backlog | $104.2B | +264% YoY; +$25B early-Q3 adds |
| FY26 revenue guide | $12.4–13.2B | Raised twice; exit run-rate $18.5–19.5B |
| FY26 capex plan | $35–39B | ~3× guided revenue |
| Q3 guide | $3.45–3.60B | Pricing hike lands here |
| Consensus | Buy | 38 analysts |
| Avg target | ~$142–144 | Range $41–$250 |
| Implied upside | ~+74% | vs. $81.08 |
06What could change the story
Bull triggers
- Q3 print (Nov 16) hits the $3.45–3.60B guide with margin inflection (low-teens adj. op. margin target for Q4); 25% July price hike shows up as promised
- Refinancing terms better than feared; delivery stays on schedule
- Vera Rubin ramp extends the scarcity window; OpenAI’s spend funnels toward Nvidia-preferred partners
Note: the Nasdaq-100 inclusion already happened (effective Jun 22, 2026) — it’s now a passive-flow fact, not a pending catalyst.
Bear triggers
- A broken construction/delivery schedule; refinancing at punitive rates
- Customer-concentration event with Meta or OpenAI (they dominate backlog)
- Dilution running ahead of market digestion — convertibles + ATM are already in motion
- Covenant pressure (1.35× DSCR) if cash flow disappoints
07What’s next
2026
Q3 earnings, after close. The make-or-break test: pricing landing, margin inflection, funding updates.
2026
Deal flow. Watch the $3B convertible pricing, ATM execution, and any refinancing headlines.
Delivery vs. backlog. Construction schedule — the class action’s core allegation — becomes the class of risk the market prices.
08FAQ
Is CoreWeave profitable?
No. Q2 2026 net loss ~$654M; TTM EPS −$2.68. It’s EBITDA-positive (~59% adj. margin) but deeply negative at the bottom line and in free cash flow.
Why is the stock down if revenue grows 112%?
The market prices the funding risk: ~$35B debt, negative FCF, active dilution, a class action. Growth without cash-flow comfort trades at a discount.
Is CRWV a buy?
Consensus says Buy, ~$144 avg target, ~+74% upside — but 12 of 38 analysts are Hold-or-worse and the spread is $41–$250. High-conviction, high-disagreement. This page is research, not a recommendation.
What does CoreWeave actually do?
Rents Nvidia GPU capacity on multi-year contracts; revenue = contracted capacity times a price that just stepped up ~25% in July 2026.
Sources & method
- Company disclosures + watcher summaries as of Sep 17–18, 2026 (24/7 Wall St., Yahoo Finance, Ticker Report, StockAnalysis consensus aggregation, TipRanks earnings calendar: Nov 16, 2026, after close)
- Q2 2026 quarterly filing figures as reported: revenue $2.575B (+112.5%), LT debt $27.56B, ST debt $7.51B, backlog $104.2B, capex guide $35–39B
- Consensus: 38 analysts via S&P Global aggregation (5 Strong Buy / 21 Buy / 10 Hold / 1 Sell / 1 Strong Sell); average target ~$144 (MarketBeat $141.90, StockAnalysis $144.46, Public.com $142.97)
- Nasdaq-100 inclusion effective Jun 22, 2026; class-action and covenant/lockbox details as reported (Fitch, press) — marked as reported, unverified by us