CRWVCoreWeave, Inc.
Industry AI datacenter Updated 24 Sep 2026
The idea
A levered equity on a backlog: $103.7bn of RPO and 4.2GW contracted against $35.55bn of debt on $5.02bn of book equity, with $10.6bn of principal due inside eighteen months at effective rates to 15%. The September package is the evidence that matters — $4.2bn of 2.875% converts plus a 35m-share ATM, against an existing stack paying 8.5-9.75%. A market writing 2.875% paper is not pricing the bear case.
Events to watch
- Q3 2026 results, early November — whether the 5-10 point contribution-margin uplift management refused to decompose shows up in reported margin
- RPO and its 41% 24-month share
- the debt table's next-twelve-months column after the convert
- any disclosed delay on a contracted gigawatt
- whether the material weaknesses are still open in the FY2026 10-K
How we got the target
Dossier model table, EV/EBITDA rather than earnings because none exist: FY2028 revenue $20bn/$30bn/$34bn at a 40-47% EBITDA margin, 5x/8x/10x, less ~$50bn of net debt, ~600m shares. Bear is ~$20 of option value, not a going-concern value — at 5x the enterprise is worth less than the debt. Time frame: FY2028 (Dec 2028).
Target: from the research notes. Prices: market data get_latest_stock_price, close of 2026-09-22 (fetched 2026-09-23)
What would prove it wrong
7 warning signsRPO stops growing, or its front end stretches.
Details
$103.7bn at 2026-06-30, with 41% expected in the initial 24 months, 39% in months 25–48 and the rest through month 78. Both numbers are disclosed quarterly. A falling 24-month percentage means the backlog is being re-phased, and re-phasing is what a contract renegotiation looks like two quarters before it is announced.
Customer A's share stops falling for the wrong reason.
Details
Customer A was 71% of Q2 2025 revenue and 36% of Q2 2026; the top three are now 36/26/10. The fall is dilution by new contracts, not the customer shrinking. If the top three re-concentrate above ~80%, the OpenAI (~$6.5bn to 2031) and Meta (~$21.0bn) commitments are the business rather than anchors within it.
The margin uplift is not repeated, or is finally decomposed and turns out to be term mix.
Details
Management claims recent-deal contribution margins 5–10 points above recent quarters and refused to break it down. If Q3 2026 does not show it in reported numbers, the uplift was shorter-duration pricing, which does not extend.
Interest expense outruns revenue growth.
Details
Net interest expense rose 140% year over year in Q2 2026 on a revenue increase of a similar order. The debt is the thesis' financing and its clock: $10.6bn of principal is due inside eighteen months of 2026-06-30 and the stack's effective rates run to 15%.
The FY2025 receivable pattern repeats at 2026-12-31 (added 2026-09-24 from S3).
Details
Receivables ran to 183 days of the closing quarter's revenue at 2025-12-31, 68% of them owed by Customer A — roughly $2.2bn from one name — and halved to ~90 days by 2026-06-30. A second year-end at or above 150 days, with the same single-name concentration, stops being a billing-cycle artefact: it is the anchor customer financing itself on CoreWeave's balance sheet, at a company that is itself borrowing at 9-15% to fund the assets that customer uses. Both numbers are disclosed annually in the concentration note.
Capex lands outside its first guided range for a third straight year (added 2026-09-24 from S8).
Details
FY2025 was guided $20-23bn and came in at $14.9bn after a cut to $12-14bn; FY2026 was guided $30-35bn and is now $35-39bn. The financing stack is sized against the guide, so the guide being wrong in either direction is a financing error, not a forecasting one: too low and the company raises again on worse terms, too high and it has paid 9-15% for capital it did not deploy.
The material weaknesses are still unremediated in the FY2026 10-K.
Details
They cover IT general controls, segregation of duties and accounting headcount, and they were still open at 2026-06-30 — a second full year unremediated, at a company now carrying $35.5bn of debt and $103.7bn of RPO, is a different fact from an IPO-stage disclosure.
How far along
7 of 10 steps done · 3 done in a group review
Next step
read the FY2025 10-K (2026-03-02) and the DEF 14A (2026-04-22). This pass built S2 and S3 almost entirely off the Q2 2026 10-Q and the September 8-K because the promotion made those the live documents; the annual report's risk factors, the auditor's report and the Class B voting arithmetic are all still unread. Second: the DDTL 5.5 facility, announced on the Q2 call as the instrument that finances shorter-duration customer contracts — it is the mechanism behind the margin uplift management would not decompose, and it should appear in the next 10-Q's debt table.
Needs your decision
S6 product test / expert call, S7 conviction.
Failed checks
none. S0/S4/S7 are not closed by design (cluster stages, carried forward). S3 closed on the 10-Q and the 8-K rather than the 10-K — see the note there; the 2026-09-24 pass added the 10-K's accounting policies, PP&E, concentration and liquidity sections, and the rest is still unread.
Who runs it
The Front OfficeStep by step
- S0How it does as AI gets cheapDone in a group review
carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)
that file puts CRWV in the debt-funded half of the complex at 4.7x its cash in long-term debt, and records top-three customer concentration of 73% of H1 2026 revenue. Owner checks dropped 2026-09-21 (README §Stage 0 override).
- S1Where the idea came fromDone
watchlist (owner, Apple Stocks, 2026-09-20), note "recent IPO; S-1 is the primer", coverage sec_domestic, cik 1769628, cluster ai_datacenter. Tier 0 promotion: queue.json promoted CRWV on the 8-K filed 2026-09-22 (priority 2)
Full notes
That 8-K is the $4.2bn convertible note closing — see S3.
- S2Quick deal-breaker checkDone
revenue $229m (2023) → $1,915m (2024) → $5,131m (2025) → $4,653m in H1 2026 (+112% YoY; Q2 $2,575m). Operating income +$324m (2024) → −$46m (2025) → −$193m (H1 2026). Net loss −$1,167m (2025) → −$1,366m (H1 2026)
Full notes
OCF $3,058m (2025) and $3,663m (H1 2026). ❗ Capex $14,117m in H1 2026 against $3,663m of operating cash flow — free cash flow of −$10.45bn in six months. Total debt $35,551m of principal at 2026-06-30 (from $22.7bn at Q1 2026) against $5,524m of cash and $5,024m of total stockholders' equity: ~7x book equity.
⚠️ The maturity wall is near: $4,413m due in the rest of 2026 and $6,184m in 2027 — $10.6bn inside eighteen months of the balance-sheet date.
⚠️ The cost of that debt is the tell: DDTL 1.0 at a 15% effective rate, the Magnetar Loan at 12%, DDTL 2.0 and the OEM financings at 11%, five series of senior notes at 9-10%. Survives on access to capital, not on cash generation — and the September financings show the access is currently open. · capital allocation (2026-09-24, dossier-2026-09-24.md): neutral — the measure does not resolve yet, and saying so is the finding. No buyback, no dividend, one $108m acquisition (Weights & Biases; Core Scientific was terminated, 8-K Item 1.02 2025-10-31): every dollar went into gross PP&E, $12,880m (2024-12-31) → $33,941m → $52,622m (2026-06-30), 4.1x in eighteen months. Free cash flow -$5,953m (2024), -$7,251m (2025), -$10,454m (H1 2026). GAAP operating income -$46m (2025) and -$193m (H1 2026), so ROIC is negative — but $11.9bn of the asset base is construction in progress, never switched on, so that arithmetic measures the build schedule, not the returns. No goodwill impairment; goodwill is 15 months old. What would resolve it: the first deployment cohort's disclosed asset-level economics, or four straight quarters where operating cash flow covers cash interest plus maintenance capex. Management's claim — deployments "fully repaying asset-level debt used to fund the CapEx while generating significant additional free cash flow" (call 2026-08-11) — is on the record with no figure behind it.
- S3Reading the company's reportsDone
(gate-minimum)
Q2 2026 10-Q in full (Notes 2, 3, 10, Part II Item 1A, Item 4); the 2026-09-22 8-K (the promoting filing) and the 2026-09-17 424B5. Q2 2026 transcript turns 9-43 of 43 read Q&A-first per SKILL §3.1; 13F, Form 4 and short interest for S5.
Full notes
FY2025 10-K (2026-03-02) and DEF 14A (2026-04-22) located but not read (see
⚠️). Three disclosed risks named with locators in dossier-2026-09-23.md. ❗ Disclosure controls were NOT effective at 2026-06-30. Three material weaknesses — IT general controls, segregation of duties, and too few qualified accounting and finance personnel — "continued to exist as of June 30, 2026". CoreWeave says they produced no material misstatement. EQ: 7 flags, no clean lines (2026-09-24). ❗ Accounts receivable $417m → $3,169m over FY2025 (a $2,749m cash outflow) on revenue of $5,131m — 183 days of Q4 2025 revenue, up from 79 a year earlier, with unbilled amounts "not material"; the 10-K names the holder: "Customer A and D accounted for 68 % and 11 % of accounts receivable, net" at 2025-12-31, i.e. ~$2.2bn owed by one customer. Improves to ~90 days by 2026-06-30. ❗ Adjusted operating income excludes interest: Q2 2026 AOI $128m against interest expense, net of $(640)m — 5x — and an adjusted net loss of $(567)m, (22)% of revenue, against $(130)m / (11)% a year earlier. SBC $630m = 12.3% of FY2025 revenue with no positive FCF to set it against. Six-year computing-equipment life (extended from five effective 2023-01-01, worth $20m then): on FY2025 D&A of $2,454m a five-year life would add ~$490m — more than two-thirds of FY2025 adjusted operating income of $668m. Capitalised interest $182m (2025), $159m (2024), $41m (2023), with interest expense struck after it and CIP at $11.9bn and rising. Capitalised software: disclosed, not quantified.
⚠️ Three FY2025 interest figures in the same 10-K do not reconcile ($1.0bn of interest payments "inclusive of $159 million related to capitalized interest" vs "interest paid, net of capitalized amounts, of $869 million" vs the PP&E note's $182m) — trivial in size, not trivial at a company that has disclosed "too few qualified accounting and finance personnel" as a material weakness. FY2025 10-K now partly read: accounting policies, PP&E, concentration, liquidity risk factor, MD&A cash flows. Risk factors in full, the auditor's report and the DEF 14A remain unread.
- S4Rivals and suppliersDone in a group review
carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)
- S5Who owns the stockDone
13F register at 2026-06-30 via 3spread (676 filers, 406.4m shares), FINRA short interest, and the September 2026 financing stack.
⚠️ The largest shareholder is also a lender. Magnetar Financial LLC holds 52,062,927 shares ($5,179.7m, trimmed 16.5m in the quarter) and the 10-Q's debt table carries a "Magnetar Loan" maturing Jan 2029 at a 12% effective rate. Equity and 12% paper in the same hands.
⚠️ NVIDIA holds 47,213,353 shares ($4,699.6m), unchanged in the quarter — and is simultaneously the 4th largest 13F holder of INTC (see INTC/scorecard).
⚠️ Short interest peaked at 80,963,200 (2026-06-30) and fell 31% to 55,568,683 (2026-08-31): shorts covered into the quarter, not after it.
- S6What workers and customers sayDone
carry-forward (cluster: AI-capex complex — cluster pass 2: S6 for 41 names, S0/S4/S7 for three, 2026-09-21, §CRWV row: Kerrisdale Capital short thesis; that file's search-summary figure of ~$35bn of total debt at 2026-06-30 is confirmed from the primary document here at $35,551m)
Full notes
Product test / expert call open OPEN (user)
- S7The case for it, and pricesDone in a group review
carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)
conviction open OPEN (user) · valuation (bear/base/bull; moved from S8 2026-09-23): $86.76 close 2026-09-22; 551m shares (457m Class A + 94m Class B) → ~$47.8bn market value, ~$78bn enterprise value on $30.0bn of net debt.
Full notes
No earnings multiple exists. The EV/EBITDA model in dossier-2026-09-23.md puts BASE at ~$97 and BULL at ~$142 against a bear case near ~$20 — the widest spread of the three names in this slot, because the equity is a thin sliver on top of $35.5bn of debt. No haircut applied (2026-09-24): S8 scores reliable, not over-promises. But the base case takes capital spending from the top of the guided range, not the midpoint — the FY2025 capex guide ran $20-23bn → $12-14bn → $14.9bn actual, and the FY2026 guide has gone $30-35bn → $31-35bn → $35-39bn. This name's equity value is a residual after debt service, so the capex line moves the bear case more than the revenue line moves the bull.
- S8Do the bosses keep their word?Done
12 rows: 7 beat, 2 met, 3 missed, 0 withdrawn
reliable, on a six-quarter record. Michael Intrator (CEO) and Nitin Agrawal (CFO) made every row and hold both seats: no departed rows, and no pre-IPO history to score (first call 2025-05-14).
⚠️ The split is clean and it matters: eight of eight quarterly guides met or beaten (six beaten, revenue and adjusted operating income alike), and three of the four FY2025 annual commitments missed — adjusted operating income $800-830m → cut to $690-720m → $668m, short of the cut guide too; capex $20-23bn → cut to $12-14bn → $14.9bn; year-end active power "over 900 megawatts" → cut to "over 850" → "more than 850". All three cuts were announced, with one disclosed cause (a third-party powered-shell delay).
⚠️ The capex guide has never landed inside its first range: FY2025 came in 35% below its, and the FY2026 range has moved up until its low end equals the top of the first one — on a name whose whole risk is financing capex. dossier-2026-09-24.md
- S9What to watch nextDone
trigger written. Tier 0 EDGAR sweep covers the feed (cik 1769628)
Sources and notes
Sources 7 documents
- Full notes, 24 Sep 2026
- Full notes, 23 Sep 2026
- AI-capex complex — cluster pass, 2026-09-20
- AI-capex complex — cluster pass 2: S6 for 41 names, S0/S4/S7 for three, 2026-09-21
- EDGAR filings CIK 1769628
| Short name | Document | Where it came from | Read on |
|---|---|---|---|
| 10Q-Q2-26 | CoreWeave 10-Q for the quarter ended 2026-06-30, filed 2026-08-12 | https://www.sec.gov/Archives/edgar/data/1769628/000176962826000366/crwv-20260630.htm | 2026-09-23 |
| 8K-2026-09-22 | CoreWeave 8-K filed 2026-09-22 — Items 1.01/2.03/3.02/8.01, the $4.2bn 2.875% convertible notes due 2033 and the capped calls. THE TIER 0 PROMOTING FILING. | https://www.sec.gov/Archives/edgar/data/1769628/000176962826000432/crwv-20260917.htm | 2026-09-23 |
| 424B5-2026-09-17 | CoreWeave 424B5 filed 2026-09-17 — prospectus supplement for up to 35,000,000 Class A shares under an Equity Distribution Agreement with collared forward sales (Reg. No. 333-296553) | https://www.sec.gov/Archives/edgar/data/1769628/000162828026062362/coreweave-424b5.htm | 2026-09-23 |
| CALL-Q2-26 | CoreWeave Second Quarter 2026 Earnings Conference Call, 2026-08-11 — turns 9-43 of 43 | Equibles GetEarningsCallTranscript CRWV 2026 Q2 — https://equibles.com/stocks/crwv/earnings-calls | 2026-09-23 |
| 13F-Q2-26-3S | 3spread 13F holders, CUSIP 21873S108, report date 2026-06-30 (676 filers, 406,431,992 shares; prior quarter 792 filers) | python ledger/research/tools/threespread.py holders CRWV | 2026-09-23 |
| SI-FINRA-EQ | FINRA bi-monthly short interest, settlements 2026-05-15 to 2026-08-31 | Equibles GetShortInterest CRWV — https://equibles.com/stocks/crwv/short-interest | 2026-09-23 |
| QUEUE-2026-09-23 | ledger/research/queue.json — CRWV promoted 2026-09-23, priority 2, 'why': 8-K filed 2026-09-22 (current report) | ledger/research/queue.json | 2026-09-23 |
Links to other companies 28 links
| From | Link | To | Date | Where it says so |
|---|---|---|---|---|
| 2.875% Convertible Senior Notes due 2033, $4.2bn | issuer of | CRWV | 2026-09-22 | 8K-2026-09-22 — Item 1.01, Indenture and Notes |
| The nine capped-call Option Counterparties (Barclays, Crédit Agricole CIB, Citibank, Deutsche Bank London, Banco Santander as agent, Bank of Nova Scotia, Deutsche Bank Securities, Goldman Sachs & Co., HSBC Bank USA, Wells Fargo Bank) | counsel to | CRWV | 2026-09-18 | 8K-2026-09-22 — Item 1.01, Capped Call Transactions |
| Equity Distribution Agreement dated 2026-09-17, up to 35,000,000 Class A shares with collared forward sales | issuer of | CRWV | 2026-09-17 | 424B5-2026-09-17 — prospectus supplement cover page |
| U.S. Bank Trust Company, National Association (indenture trustee) | counsel to | CRWV | 2026-09-22 | 8K-2026-09-22 — Item 1.01, Indenture and Notes; Exhibit 4.1 |
| CRWV | context | CRWV | 2026-09-22 | 8K-2026-09-22 — Item 1.01, Indenture and Notes — list of guaranteed existing series |
| CoreWeave 'Customer A' (not named in the 10-Q's concentration table) | channel concentration | CRWV | 2026-06-30 | 10Q-Q2-26 — Part II Item 1A Risk Factors — customer concentration; and the concentration table in the notes |
| OpenAI OpCo, LLC | customer of | CRWV | 2026-06-30 | 10Q-Q2-26 — Part II Item 1A Risk Factors — customer concentration |
| Meta Platforms, Inc. | customer of | CRWV | 2026-06-30 | 10Q-Q2-26 — Part II Item 1A Risk Factors — customer concentration |
| CoreWeave 'Customer A' (not named in the 10-Q's concentration table) | context | CRWV | 2026-06-30 | 10Q-Q2-26 — Note 2, Remaining Performance Obligations |
| Magnetar Financial LLC | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 1 of the holders table, with its SEC filing URL |
| Magnetar Financial LLC | financing partner | CRWV | 2026-06-30 | 10Q-Q2-26 — Note 10, Debt — total debt obligations table, 'Magnetar Loan' row |
| NVIDIA Corporation | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 3 of the holders table |
| Jane Street Group, LLC | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 2 of the holders table |
| Susquehanna International Group, LLP | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 4 of the holders table |
| Goldman Sachs Group Inc | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 5 of the holders table |
| BlackRock, Inc. | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 6 of the holders table |
| Alyeska Investment Group, L.P. | holds | CRWV | 2026-06-30 | 13F-Q2-26-3S — row 9 of the holders table |
| CRWV | context | CRWV | 2026-08-31 | SI-FINRA-EQ — settlements 2026-05-15 through 2026-08-31 |
| DDTL 1.0 Facility, Mar 2028, 15% effective rate | issuer of | CRWV | 2026-06-30 | 10Q-Q2-26 — Note 10, Debt — total debt obligations table and future principal payments table |
| CRWV | context | CRWV | 2026-06-30 | 10Q-Q2-26 — Part II Item 1A Risk Factors — indebtedness |
| The third-party data centre provider whose November 2025 delivery delays CoreWeave announced (not named) | supplier dependency | CRWV | 2026-06-30 | 10Q-Q2-26 — Part II Item 1A Risk Factors — third-party data centres |
| Two unconsolidated data-centre development joint ventures (VIEs, not named) | financing partner | CRWV | 2026-06-30 | 10Q-Q2-26 — Note 3, Unconsolidated Variable Interest Entities |
| Supply agreement counterparty rendered as 'Solidine' in the Q2 2026 transcript (name not resolvable from the source) | supplier dependency | CRWV | 2026-08-11 | CALL-Q2-26 — Q&A, Samik Chatterjee (J.P. Morgan) exchange, turn 9 |
| NVIDIA Corporation | supplier dependency | CRWV | 2026-08-11 | CALL-Q2-26 — Q&A, Samik Chatterjee (J.P. Morgan) exchange, CEO answer |
| Michael Intrator (CEO) | officer of | CRWV | 2026-09-22 | 8K-2026-09-22 — signature page |
| Nitin Agrawal (CFO) | officer of | CRWV | 2026-08-11 | CALL-Q2-26 — Q&A, Irvin Lu (Evercore ISI) and Michael Turrin (Wells Fargo) exchanges |
| CRWV | governance structure | CRWV | 2026-06-30 | 10Q-Q2-26 — Part I Item 4, Controls and Procedures |
| CRWV | event | CRWV | 2026-09-23 | QUEUE-2026-09-23 — promoted[0] |
History 3 entries
- 2026-09-23 — first pass, slot
names-ai-capex-4, model-drafted unattended by the cloud research routine, on a Tier 0 promotion (8-K of 2026-09-22). S1, S2, S3 (gate-minimum, built on the Q2 2026 10-Q and the September filings rather than the 10-K), S5 and S9 closed; S0/S4/S7 carried forward from the 2026-09-20 cluster file and S6 from the 2026-09-21 one; S8 modelled and left open. Evidence and access limits in dossier-2026-09-23.md. Nothing here is human-verified. - 2026-09-23 — stage redefinition, owner decision (WATCHLIST_PLAN.md §25): S8 is now Management credibility and is not yet run for this name; the old S8 valuation figures moved to the
valuation:line under S7, and any proposed buy price or position size was removed (the pipeline no longer drafts either). No research was redone. S2 capital allocation and S3 earnings quality are also pending, in the same backfill slot. - 2026-09-24 — backfill, slot
backfill-2026-09-23-4, model-drafted unattended by the cloud research routine. S8 closed ✅ reliable on a 12-row, six-quarter record with no departed promisers; S2 gained its capital-allocation table — no verdict, deliberately, with the two tests that would produce one written down; S3 gained its earnings-quality check — 7 flags, ✅ kept at gate-minimum. Glyphs on S2 and S3 unchanged. Two kill criteria added (the year-end receivable concentration repeating; capex missing its first range a third year), the ⚠️ block extended, and the S7 valuation line told to take capital spending from the top of the guided range. Evidence in dossier-2026-09-24.md; 0 Equibles calls, 0 market data calls, EDGAR and the transcript archive only. One transcript defect recorded (FY2025-Q1 "$19 billion" for $1.9bn of Q1 capex). Nothing here is human-verified.
Full notes, 24 Sep 2026 18 min read
CRWV — dossier, 2026-09-24 (backfill: S8, S2 capital allocation, S3 earnings quality)
Slot backfill-2026-09-23-4. Model-drafted unattended by the cloud research routine; nothing here is human-verified. This file holds only the three checks added on 2026-09-23 (WATCHLIST_PLAN.md §25) plus the transcript work for the calls read for them. Every other stage stands as written in dossier-2026-09-23.md. Anything from background knowledge is marked [background].
The window is short, and that is the finding as much as anything in it. CoreWeave has reported six quarters as a public company (first call 2025-05-14). There is no 4–8 quarter record before the IPO to score. The two people who made every promise below — Michael Intrator (CEO) and Nitin Agrawal (CFO) — have been in their seats for all six, so unlike ORCL in the same slot there is no attribution problem; the limit is calendar, not turnover.
Sources this pass. EDGAR only, User-Agent ddboard-research gorkem@celebico.com (CIK 1769628): the FY2025 10-K (2026-03-02, accession 0001769628-26-000104), the Q2 2026 earnings release (8-K Item 2.02, 2026-08-11, Ex. 99.1), XBRL companyfacts. Transcripts from the archive ledger/research/_transcripts/CRWV/ (market data, all six calls, FY2025-Q1 to FY2026-Q2; call dates from _transcripts/index.json) — the FY2026-Q2 call was read in dossier-2026-09-23.md and the other five are read here for the guidance path. Equibles: 0 calls. market data: 0 calls.
Quote discipline: transcript quotes checked by exact substring against the archived JSON bytes; filing quotes against a whitespace-collapsed text extraction of the EDGAR HTML. Turn numbers [n] are line positions in the market data content string. One transcript defect found and worked around: the FY2025-Q1 file reads "CapEx in Q1 totaled $19 billion", which is impossible — the FY2025-Q4 call gives full-year capex of $14.9bn with $8.2bn in Q4, and Q2 and Q3 were $2.9bn and $1.9bn, leaving $1.9bn for Q1. It is a transcription error for $1.9 billion and the figure is not quoted from that file.
S8 — management credibility
How CoreWeave guides. On each call the CFO gives, for the next quarter: revenue as a range, adjusted operating income as a range, interest expense as a range and capex as a range; and for the full year: revenue, adjusted operating income, capex and active power in megawatts at year-end. The last of those is unusual and useful — it is a physical commitment that cannot be re-defined. Scored against the first guide with the revision path recorded.
Two definitional notes that decide how the rows read:
- "CapEx" here is management's measure, not the cash-flow statement's. The CFO: "CapEx consists primarily of investments in property and technology equipment, and is calculated as a change in gross PP and E minus the change in construction in progress" (call 2025-08-12 [3]), and "CapEx has reported net of construction in progress as that represents infrastructure not yet in service" (call 2025-05-14 [3]). XBRL
PaymentsToAcquirePropertyPlantAndEquipmentfor FY2025 is $10,309m against the $14.9bn management reports. Guide and result below are both on management's measure; the difference is itself in the earnings-quality section. - "Adjusted operating income" excludes interest, which for this company is the largest cost line in the business. See S3 below. The rows score what was promised on the basis it was promised.
Who promised. Michael Intrator, Chief Executive Officer, and Nitin Agrawal, Chief Financial Officer, on every one of the six calls; the market data speaker labels are legible and consistent throughout ("Michael Intrator", "Nitin Agrawal"), with the quantitative guidance always in Agrawal's turn [3]. No departed rows. No Item 5.02 in the window touches either role — the only executive-change item the routine found is the appointment of a first Chief Revenue Officer, named on the call of 2025-11-10 [2] ("we are excited to welcome John Jones as our first Chief Revenue Officer").
Say vs do (guides for Q2 2025 → Q2 2026, plus the FY2025 annual set)
Revenue actuals from XBRL RevenueFromContractWithCustomerExcludingAssessedTax; adjusted operating income, capex and active power from the following call, which is where CoreWeave reports them.
| # | Promise (period · metric · range) | Made where · by whom | Result · locator | Score |
|---|---|---|---|---|
| 1 | Q2 2025 · revenue $1.06–1.10bn | call 2025-05-14 [3] · Nitin Agrawal (CFO) | $1,212m (XBRL, 10-Q 0001769628-26-000366) — 10% above the top; call 2025-08-12 [3]: "Q2 revenue was $1.2 billion growing 207% year over year" | beat |
| 2 | Q2 2025 · adjusted operating income $140–170m | same | $200m — call 2025-08-12 [3] | beat |
| 3 | Q3 2025 · revenue $1.26–1.30bn | call 2025-08-12 [3] · Agrawal | $1,364.7m (XBRL) — above the top; call 2025-11-10 [2]: "In Q3, we beat expectations, delivering revenue of $1.4 billion, up 134% year over year" | beat |
| 4 | Q3 2025 · adjusted operating income $160–190m | same | $217m — call 2025-11-10 [3], which credits "higher revenue, lower costs due to timing of data center deliveries from our third-party partners, and improved fleet efficiencies" | beat |
| 5 | FY2025 · revenue $4.9–5.1bn | call 2025-05-14 [3] · Agrawal. Raised 2025-08-12 [3] to $5.15–5.35bn; cut 2025-11-10 [3] to "$5.05 billion to $5.15 billion" | $5,131m (FY2025 10-K) — 0.6% above the top of the first range, below the raised one, inside the cut one | beat (marginally, after a raise and a cut) |
| 6 | FY2025 · adjusted operating income $800–830m | call 2025-05-14 [3] · Agrawal; reaffirmed 2025-08-12 [3] "unchanged from our prior guidance"; cut 2025-11-10 [3] to $690–720m | $668m (163 + 200 + 217 + 88, each read off the call that reported it). Below the cut guide as well as the original — 19% short of the first range's midpoint | missed |
| 7 | FY2025 · capex $20–23bn (management's measure) | call 2025-05-14 [3] · Agrawal; reaffirmed 2025-08-12 [3]; cut 40% on 2025-11-10 [3] to "$12 billion to $14 billion", explained as a shift into construction in progress after a third-party powered-shell delay | $14.9bn — call 2026-02-26 [3]: "CapEx in Q4 totaled $8.2 billion and $14.9 billion for the full year, higher than anticipated due to our team's ability to put infrastructure in service ahead of our expectations." Above the cut range and 35% below the first one | missed |
| 8 | FY2025 · active power "over 900 megawatts … before the end of the year" | call 2025-08-12 [2] · Michael Intrator (CEO): "We are now on track to deliver over 900 megawatts of active power before the end of the year"; cut 2025-11-10 [3] to "over 850 megawatts" | "We ended the year with more than 850 megawatts of active power, adding approximately 260 megawatts in the fourth quarter alone across 43 active data centers" — call 2026-02-26 [2]. Meets the cut figure, misses the first | missed |
| 9 | Q1 2026 · revenue $1.9–2.0bn | call 2026-02-26 [3] · Agrawal | $2,078m (XBRL) — above the top | beat |
| 10 | Q1 2026 · adjusted operating income $0–40m | same | $21m — call 2026-05-07 [3]: "Adjusted operating income for Q1 was $21 million … just above the midpoint of our guidance" | met |
| 11 | Q2 2026 · revenue $2.45–2.6bn | call 2026-05-07 [3] · Agrawal | $2,575m (XBRL; Q2 2026 release) — inside, near the top | met |
| 12 | Q2 2026 · adjusted operating income $30–90m | same | $128m — Q2 2026 release reconciliation; call 2026-08-11 [3]: "well above the high end of our guidance" | beat |
Not scored (open): Q3 2026 revenue $3.45–3.6bn and adjusted operating income $200–260m, and FY2026 capex $11.5–13.5bn for Q3 (all call 2026-08-11 [3]); FY2026 revenue, first guided $12–13bn (2026-02-26 [3]), reaffirmed 2026-05-07 [3] and raised to $12.4–13.2bn (2026-08-11 [3]); FY2026 adjusted operating income $900m–1.1bn raised to $960m–1.15bn on the same path; FY2026 capex $30–35bn → $31–35bn (2026-05-07 [3], "The increase on the low end … is related to increases in component pricing") → $35–39bn (2026-08-11 [3]); FY2026 active power "more than 1.7 gigawatts by year-end" raised to "more than 1.85 gigawatts"; and the two long-dated ones — "we expect to exit 2026 with annualized run rate revenue of $17 billion to $19 billion, which we expect to grow to more than $30 billion of annualized run rate revenue as we exit 2027" (2026-02-26 [3]) and "at least 8 gigawatts by 2030" (2026-08-11 [2]).
Tally: 12 scored rows — 7 beat, 2 met, 3 missed, 0 withdrawn.
The shape. Eight of eight quarterly guides met or beaten, six of them beaten, across both revenue and adjusted operating income. Three of the four annual FY2025 commitments missed against their first guide — and all three were publicly cut first, all three for the same disclosed reason (a third-party powered-shell delivery delay), and one of them, adjusted operating income, missed even after the cut. That is the pattern to carry: the quarter is reliably sandbagged; the year is guided from a build schedule the company does not fully control.
The capex guide's dispersion is a finding in its own right. FY2025: $20–23bn → $12–14bn → $14.9bn actual. FY2026: $30–35bn → $31–35bn → $35–39bn, still open. For a business whose entire risk is financing capital expenditure — $35.6bn of debt at 2026-06-30, $10.6bn of principal due inside eighteen months, effective rates to 15% (scorecard.md S2) — a capital-spending forecast that has was cut ~40% at the midpoint in 2025, and has since been raised until its low end equals the top of its first 2026 range, is not a detail. It is the input to every financing decision the company makes, and it has never yet landed inside its first range.
Dodge-log roll-up (§3.1). The three dodges recorded on 2026-09-23 are all from the FY2026-Q2 call and each is marked first observation. This pass read the five earlier calls for guidance rather than for a full Q&A dodge sweep, so no second observation is claimed for any of them — the topic that would matter most, the refusal to decompose the 5–10 point margin uplift, becomes a finding only if it recurs on the Q3 2026 call. What this pass can add is the shape of the refusals in the prepared remarks: the FY2026-Q2 call states the economic claim on which the whole structure rests — "The deployment delivers attractive returns, fully repaying asset-level debt used to fund the CapEx while generating significant additional free cash flow" (call 2026-08-11 [3]) — without a return figure, a payback period or a single worked deployment. That is the sentence to hold management to, and it is now on the record with a locator.
Verdict — reliable, on a six-quarter record. Every quarterly number met or beaten; the annual misses are all FY2025, all pre-announced, all traceable to one disclosed third-party delay, and the company has since raised FY2026 revenue, adjusted operating income and active power. It is not conservative: a conservative guider does not cut three of four annual commitments and then miss one of the cut ones. It is not over-promises: 3 of 12, none concealed. The honest qualifier is that six quarters is barely a record at all — this is a company eighteen months public, with three unremediated material weaknesses in its accounting function (scorecard.md S3), whose longest promise ("at least 8 gigawatts by 2030", "more than $30 billion of annualized run rate revenue as we exit 2027") is far longer than anything it has yet been scored on.
S2 — capital allocation
The five-year table the stage asks for does not exist: CoreWeave's first full year as a public company is 2025 and XBRL carries 2024 onward. What follows is what the filings support.
| $m | FY2024 | FY2025 | H1 2026 |
|---|---|---|---|
| Revenue | 1,915 | 5,131 | 4,653 |
| Operating cash flow | 2,749 | 3,058 | 3,663 |
Capex (cash, XBRL PaymentsToAcquirePropertyPlantAndEquipment) | 8,702 | 10,309 | 14,117 |
| Free cash flow | (5,953) | (7,251) | (10,454) |
| Acquisitions (cash) | 0 | 108 | 0 |
| Buybacks | 0 | 0 | 0 |
| Dividends | 0 | 0 | 0 |
| Stock-based comp | 31 | 630 | 318 |
| SBC as % of revenue | 1.6% | 12.3% | 6.8% |
| Gross PP&E, period end | 12,880 | 33,941 | 52,622 |
| Depreciation & amortisation | 863 | 2,454 | 2,540 |
Where the cash went. All of it, and $23.7bn more, into property and equipment: gross PP&E went $12.9bn → $52.6bn in eighteen months, a 4.1x. There is no other use — no buyback, no dividend, one small acquisition (Weights & Biases, $108m of cash in FY2025; the much larger Core Scientific deal was terminated, 8-K Item 1.02 filed 2025-10-31). The funding is in scorecard.md S2: $35,551m of principal at 2026-06-30, up from $22.7bn one quarter earlier, at effective rates from 9% to 15%, plus the September 2026 equity-linked stack.
Buybacks at what price: none. Dividends: none. Neither is a criticism for a company at this stage; both are recorded because the stage asks and because the issuance side is the live question — the 2026-09-17 424B5 registers up to 35,000,000 Class A shares and the 2026-09-22 8-K adds $4.2bn of converts with a maximum of 52,578,540 shares on conversion (scorecard.md, §The September 2026 refinancing).
Share count: the five-year series does not exist — CoreWeave IPO'd in March 2025 [background] and the first public share count is Q1 2025. 551m shares at 2026-09-22 (457m Class A + 94m Class B, scorecard.md S7). The dilution question here is forward-looking and sits in the two September filings, not in a historical series.
What the acquisitions were worth: goodwill $20m (2024-12-31) → $1,101m (2025-12-31), all of it the Weights & Biases purchase. GoodwillImpairmentLoss is not tagged in any period — no impairment has been recorded, and the goodwill is fifteen months old.
Returns. The stage's measure does not resolve on this company yet, and saying so precisely is more useful than forcing it:
- GAAP operating income was −$46m in FY2025 and −$193m in H1 2026 (the Q2 2026 release reconciliation). NOPAT is therefore negative and ROIC is negative on an invested-capital base that went from roughly $13bn of gross PP&E to $52.6bn in six quarters. Mechanically, that is capital destruction.
- Mechanically is not the same as actually. At 2026-06-30, $11.9 billion of that base is construction in progress — infrastructure not in service, earning nothing and depreciating nothing (call 2026-08-11 [3]: "Construction in progress, CIP, increased to $11.9 billion from $9.6 billion quarter-over-quarter"). An incremental-return calculation over a window in which a quarter of the asset base has never been switched on measures the build schedule, not the returns.
- The test that does resolve it is management's own claim, quoted above: that each deployment "fully repay[s] asset-level debt used to fund the CapEx while generating significant additional free cash flow". That is checkable, deployment by deployment, and CoreWeave has not published the numbers to check it with.
Verdict: neutral — the measure does not resolve yet, and that is itself the state of the evidence. Not compounds: there are no returns to compound, only a negative operating result and $35.6bn of debt behind it. Not destroys: the arithmetic that would say so is being run over an asset base a quarter of which is not yet in service, on a company eighteen months public. The specific thing that converts this line into a verdict is the first cohort of deployments to complete a full contract term with disclosed asset-level economics — or, failing disclosure, four consecutive quarters in which operating cash flow covers cash interest plus maintenance capex. Until one of those exists, this stage has no honest verdict to give, and any scorecard that prints one is printing an opinion.
S3 addendum — earnings quality
Sources: FY2025 10-K (accession 0001769628-26-000104); Q2 2026 earnings release (8-K 2026-08-11, Ex. 99.1); XBRL companyfacts.
- The non-GAAP bridge. Flag 1 — the headline profit measure omits the largest cost in the business. Q2 2026 reconciliation (release 2026-08-11): operating loss $(49)m + stock-based compensation $165m + acquisition costs $1m + amortisation of acquired intangibles $11m = adjusted operating income $128m. In the same quarter interest expense, net was $(640)m (H1 2026: $(1,176)m) — five times the adjusted operating income the company leads with. The release's own adjusted net loss line shows what that does: $(567)m in Q2 2026, a (22)% margin, against $(130)m and (11)% a year earlier. Adjusted operating income is up as a headline ($128m vs the guide) while the adjusted net loss quadrupled. For a business funded with $35.6bn of 9–15% debt, "adjusted operating income" is a measure of the business without its financing, which is the one thing this business cannot be separated from. Second-order: the amortisation add-back is new — footnote (2) to the reconciliation: "In the second quarter of 2025, we began including an adjustment for the amortization of acquired intangibles in our calculation of adjusted operating income. Prior period non-GAAP calculations for acquired intangible amortization are not being adjusted as these amounts were insignificant." Disclosed, small ($11m), and correctly flagged as a definition change.
- Stock-based compensation. Flag 2. FY2025 SBC $630m = 12.3% of revenue, from 1.6% in FY2024 — the IPO grants, as the CFO said would happen ("we expect stock-based compensation to remain slightly elevated through the year for the grants issued in connection with the IPO", call 2025-05-14 [3]). H1 2026 is $318m, 6.8% of revenue, so it is normalising. Free cash flow is −$7,251m (FY2025) and −$10,454m (H1 2026), so there is no positive FCF to measure it against; SBC is the entire adjusted operating income of FY2025 ($630m against $668m) restated as stock.
- Cash conversion. Flag 3 — the good-looking ratio is prepayments, and the receivable behind it is one customer. Operating cash flow was positive in every period ($3,058m in FY2025) against a net loss of $1,167m, which looks like strong conversion and is mostly depreciation ($2,454m) plus customer money arriving early. The 10-K MD&A says so: net cash from operations rose "primarily driven by cash received for upfront payments from our committed contracts from new and expanded customer contracts. The increase was partially offset by an increase in accounts receivable." That offset is the flag — see 4.
- Working capital. Flag 4, and it is the loudest finding in this addendum. Accounts receivable went $417m (2024-12-31) → $3,169m (2025-12-31), a $2,749m cash outflow on the FY2025 cash-flow statement, on revenue of $5,131m. Measured against the quarter the balance sheet closes, receivables at 2025-12-31 were 183 days of Q4 2025 revenue ($3,169m against $1,572m), up from 79 days a year earlier, on revenue growing 168%. Unbilled amounts are not the explanation — the 10-K says unbilled receivables "have not been material for the periods presented". What the 10-K does say is who owes it: "Customer A and D accounted for 68 % and 11 % of accounts receivable, net, respectively, as of December 31, 2025." Sixty-eight per cent of $3,169m is roughly $2.2 billion owed by a single customer at year-end, at a company whose top-three concentration is already a kill criterion. It improved through H1 2026 — $2,541m at 2026-06-30, ≈90 days of the quarter's revenue — but the FY2025 year-end position, and the fact that it was one name, is the single most checkable soft spot in the accounts.
- Costs moved onto the balance sheet. Flags 5 and 6 — both disclosed, both material, both growing.
- Useful life of computing equipment, extended once and still at the long end. FY2025 10-K: "Effective January 1, 2023, the Company changed its estimate of the useful life for its computing equipment utilized in data centers from five to six years, reflecting continuous advancements in hardware performance, software optimization, and data center design improvements. The effects of this change in estimate for the year ended December 31, 2023 … was a reduction in total expenses of $ 20 million." The change is old and properly disclosed; its consequence is not. FY2025 D&A was $2,454m. If computing equipment is the bulk of it, running it over five years instead of six would add roughly $490m of depreciation — against FY2025 adjusted operating income of $668m and a GAAP operating loss of $46m. That arithmetic is this run's, on the stated assumption that most of D&A is six-year computing equipment; the point is not that the estimate is wrong, it is that a single year of estimated life is now worth more than two-thirds of the company's headline profit, where in 2023 it was worth $20m.
- Capitalised interest, and it is compounding with CIP. FY2025 10-K, PP&E note: "There was $ 182 million, $ 159 million, and $ 41 million of interest capitalized during the years ended December 31, 2025, 2024, and 2023, respectively." Reported interest expense is struck after it — "Interest expense, net is reflected net of capitalized interest." With construction in progress at $11.9bn and rising, the amount of interest that lands in the asset rather than the income statement grows with the build.
- Capitalised software is disclosed and amortised over three to six years, included in PP&E; no separate magnitude is given, so this part of the test is recorded as not quantified rather than clean.
- Flag 7 — three FY2025 interest figures in the same 10-K do not reconcile. The liquidity risk factor states: "For the year ended December 31, 2025, our cash flows dedicated for debt service requirements totaled $4.4 billion, which includes principal payments of $3.4 billion and interest payments of $1.0 billion, inclusive of $159 million related to capitalized interest", and, in the same paragraph, that operating cash flow "includes interest paid, net of capitalized amounts, of $869 million". $1,000m less $159m is $841m, not $869m; and the PP&E note (above) puts FY2025 capitalised interest at $182m, not $159m — $159m is the 2024 figure. One of the three is wrong, or they are three different bases with none of them labelled. This is recorded, not resolved: the routine cannot tell which, and it matters because CoreWeave disclosed that its disclosure controls were not effective at 2026-06-30 with three material weaknesses, one of them "too few qualified accounting and finance personnel" (
scorecard.mdS3). A $28m rounding inconsistency is trivial on its own; the same inconsistency at a company that has told you its accounting function is understaffed is the kind of thing this check exists to catch.
- Sector equivalents: n/a.
EQ summary: 7 flags, no clean lines. The headline profit measure excludes interest that is 5x its size; SBC at 12.3% of FY2025 revenue with no positive free cash flow to set it against; operating cash flow flattered by customer prepayments while $2.7bn of revenue went into receivables, 68% of them from one customer; a six-year equipment life whose marginal year is now worth more than the company's adjusted operating income; capitalised interest of $182m growing with an $11.9bn construction-in-progress balance; a new (small, disclosed) add-back; and three FY2025 interest figures in one 10-K that do not agree, at a company with unremediated accounting-personnel material weaknesses. Everything here is disclosed in the filings — none of it is hidden, and all of it is the ordinary consequence of financing a $52.6bn asset base in eighteen months. It is also, taken together, the reason S3's gate stays at ✅ (gate-minimum).
What this pass read of the FY2025 10-K, and what it did not. Read: the significant-accounting- policies note (useful lives, capitalised interest, capitalised software, ARO), the property and equipment note, the concentration-of-credit-risk note, the liquidity and debt-service risk factor, and the MD&A cash-flow discussion. Still unread and still the next action on scorecard.md: the risk factors in full, the auditor's report and critical audit matters, and the DEF 14A (2026-04-22) with the Class B voting arithmetic.
Full notes, 23 Sep 2026 16 min read
CRWV — dossier, 2026-09-23
Slot names-ai-capex-4, on a Tier 0 promotion. Drafted by Claude in an unattended cloud routine. Nothing here is human-verified. Every figure carries the document it was read from; anything from background knowledge is marked [background]. Quotes were located by substring search in the source and copied, never re-typed. Never edited after today.
Price reference: $86.76 close, 2026-09-22, volume 26,428,459 (market data NASDAQ:CRWV). The 2026-09-22 8-K gives an earlier fix: $79.88 on 2026-09-17.
The one-line version
CoreWeave burned $10.45 billion of free cash flow in six months, carries $35.55 billion of debt against $5.02 billion of book equity, has $10.6 billion of principal falling due inside eighteen months at effective rates running to 15% — and in the week before this pass it raised ~$7.2bn across a 2.875% convertible and a 35-million-share ATM. The coupon is the news: the existing stack pays 8.5% to 9.75%. Either the credit has been repriced, or the equity optionality is paying the coupon. Everything else — $103.7bn of RPO, 4.2GW contracted, inference from $1m to $100m in a quarter — is the reason someone was willing to write it.
S1 — Source
watchlist (owner, Apple Stocks, 2026-09-20), row: {ticker: "CRWV", label: "CoreWeave, Inc.", coverage: sec_domestic, cluster: ai_datacenter, cik: 1769628, note: "recent IPO; S-1 is the primer"}.
Tier 0 promotion. queue.json promoted CRWV at priority 2 on 2026-09-23 — "8-K filed 2026-09-22 (current report)". Under SKILL §0 a promoted name with no scorecard.md replaces the last member of the batch; CRWV was already the last member of names-ai-capex-4, so the batch is unchanged and the promotion was handled in place. It is removed from promoted by this slot.
S2 — Kill test
From XBRL company facts (data.sec.gov, User-Agent sent) and the Q2 2026 10-Q.
| $m | 2023 | 2024 | 2025 | H1 2025 | H1 2026 |
|---|---|---|---|---|---|
| Revenue | 229 | 1,915 | 5,131 | 2,194 | 4,653 |
| Operating income (loss) | (14) | 324 | (46) | — | (193) |
| Net loss | (594) | (863) | (1,167) | — | (1,366) |
| Operating cash flow | 1,833 | 2,749 | 3,058 | (190) | 3,663 |
| Capex (PP&E) | 2,943 | 8,702 | 10,309 | 3,860 | 14,117 |
| Free cash flow | (1,110) | (5,953) | (7,251) | (4,050) | (10,454) |
Q2 2026 alone: revenue $2,575m, operating loss $(49)m, net loss $(626)m.
Balance sheet at 2026-06-30: cash $5,524m; total assets $77,070m; total stockholders' equity $5,024m; deferred revenue (current + non-current) $9.7bn; shares outstanding 457m Class A + 94m Class B = 551m.
The debt, in management's own table (Note 10, effective interest rates as filed):
| Instrument | Maturity | Eff. rate | 2026-06-30 $m |
|---|---|---|---|
| DDTL 1.0 Facility | Mar 2028 | 15% | 1,300 |
| DDTL 2.0 Facility | Aug 2030 | 11% | 3,190 |
| DDTL 2.1 Facility | Mar 2031 | 9% | 3,000 |
| DDTL 3.0 Facility | Aug 2030 | 9% | 2,215 |
| DDTL 5.0 Facility | Nov 2031 | 9% | 1,101 |
| 2030 Senior Notes | Jun 2030 | 10% | 2,000 |
| 2031 9.00% Senior Notes | Feb 2031 | 10% | 1,750 |
| 2031 9.75% Senior Notes | Oct 2031 | 10% | 2,750 |
| 2032 9.625% Senior Notes | Jul 2032 | 10% | 1,250 |
| 2032 EUR Senior Notes | Jul 2032 | 9% | 2,279 |
| 2031 Convertible Senior Notes | Dec 2031 | 2% | 2,588 |
| 2032 Convertible Senior Notes | Oct 2032 | 2% | 4,000 |
| OEM and software licence financings | Dec 2026 – Jul 2030 | 11% | 4,220 |
| Magnetar Loan | Jan 2029 | 12% | 189 |
| DDTL 4.0 (non-recourse) | Mar 2032 | 7% | 2,837 |
| OEM financings (non-recourse) | Aug 2026 – Aug 2028 | 9% | 882 |
Total future principal $35,551m, of which $4,413m falls in the remainder of 2026 and $6,184m in 2027. Net interest expense rose 140% year over year in Q2 2026 and 121% in H1, "primarily attributable to increased borrowing levels and total debt obligations."
Survives-or-dies. CoreWeave survives on access to capital and nothing else. Operating cash flow of $3.66bn against $14.1bn of capex is not a business that funds itself, and $5.02bn of book equity under $35.55bn of debt is roughly 7x. The counterweight is real and also documented: $103.7bn of RPO, two named counterparties committing ~$6.5bn (OpenAI, to 2031) and ~$21.0bn (Meta, to 2032), 4.2GW contracted, and — decisively for this quarter — a capital market that has just written $4.2bn at 2.875% and stands ready to take 35m shares. The kill test is passed conditionally: it survives as long as that window stays open, and the RPO is what keeps it open.
S3 — Filings, gate-minimum
Read: the Q2 2026 10-Q in full (Notes 2, 3, 10; Part I Item 4; Part II Item 1A; MD&A); the 2026-09-22 8-K (the promoting filing); the 2026-09-17 424B5; the Q2 2026 call (turns 9–43 of 43). Not read: the FY2025 10-K (2026-03-02), the DEF 14A (2026-04-22), the S-1/424B4, any earlier transcript, and the efts.sec.gov auditor/restatement sweep. The promotion put the September filings and the latest 10-Q in front of the annual report; that is a deliberate ordering and it is also this pass's largest gap.
The three biggest risks management itself discloses
- "A substantial portion of our revenue is driven by a limited number of customers." — Q2 2026 10-Q, Part II Item 1A. The numbers in the same paragraph: top three at "approximately 36%, 26%, and 10%" of Q2 2026 revenue; top customer alone at ~71% of Q2 2025; "OpenAI committed to pay us up to approximately $6.5 billion through May 31, 2031"; "Meta Platforms, Inc. ("Meta") initially committed to pay us up to approximately $21.0 billion". And the forward statement: "We expect that our customer concentration with a limited number of top customers is likely to continue in future years because of the long-term nature of contracts with those customers."
- "Our substantial indebtedness could materially adversely affect our financial condition, our ability to raise additional capital to fund our operations, our ability to operate our business, our ability to react to changes in the economy or our industry, our ability to meet our obligations under our outstanding indebtedness and could divert our cash flow from operations for debt payments, and we may still incur substantially more indebtedness in the future." — Q2 2026 10-Q, Part II Item 1A, heading, immediately followed by "We have a substantial amount of debt, which requires significant interest and principal payments." The body names DDTL lenders "instituting foreclosure proceedings against their collateral" as a specific consequence of default.
- Reliance on third-party data centres CoreWeave does not control, with a precedent already on the record: "the delays that we announced in November 2025 with respect to the delivery of certain data centers to be provided by a third-party data center provider". CoreWeave leases or licenses all of its space.
Controls: not effective
Part I Item 4: "our disclosure controls and procedures were not effective at the reasonable assurance level as of June 30, 2026 due to the material weaknesses in our internal control over financial reporting described below." The three weaknesses are IT general controls over the applications supporting financial reporting, insufficient segregation of duties, and "lack of sufficient number of qualified personnel within our accounting, finance, and operations functions". "We have also concluded that these material weaknesses continued to exist as of June 30, 2026." CoreWeave adds that they "did not result in a material misstatement to our financial statements."
This is the same disclosure names-ai-capex-3 recorded at CBRS. Two of the cluster's recent IPOs are running unremediated material weaknesses while raising billions in the capital markets.
The RPO, and the shape it shares with Oracle
"As of June 30, 2026, the Company had $ 103.7 billion of unsatisfied RPO, of which 41 % was expected to be recognized over the initial 24 months ending June 30, 2028, 39 % between months 25 and 48, and the remaining balance recognized between months 49 and 78." (Note 2.)
Against annualised Q2 revenue of ~$10.3bn that is ~10 years of sales booked and undelivered — the same ratio ORCL shows at $664bn against $67.4bn. Two companies at opposite ends of the size range, funding very different balance sheets, both now valued on a backlog roughly ten times revenue whose counterparties they will not name. CoreWeave at least names two of them.
One nuance the definition carries and Oracle's does not: CoreWeave's RPO is explicitly net of estimated variable consideration, including "amounts that may not be recognized as revenue due to delivery delays" and "estimates of committed cloud computing capacity that the Company has the right to resell". The $103.7bn is already after management's own haircut for the delivery risk the November 2025 disclosure demonstrated.
Off-balance-sheet development capital
Note 3: CoreWeave committed up to $1.7bn in H1 2026 (about $500m of it in Q2) to acquire equity in two joint ventures each holding a data centre development project. It is not the primary beneficiary, so the VIEs are unconsolidated; carrying value $479m at 2026-06-30, maximum exposure $1.7bn. That is development capital which does not show in the $14.1bn capex line.
Transcript — Q2 2026 call, 2026-08-11 (read Q&A-first, SKILL §3.1)
Turns 9–43 of 43. Prepared remarks not read. Six analysts: J.P. Morgan, Deutsche Bank, Evercore ISI, Barclays, Wells Fargo, Cantor Fitzgerald.
Guidance and changes.
- Exit ARR guidance raised to $18.5–19.5bn. The exchange is worth quoting carefully: the analyst asked about "the $18 to $19 billion in ARR as kind of the exit target for 2027" and the CFO answered "we increased the exit ARR number that we provided in guidance to you folks right now at 18.5 to 19.5 for 2026." The two years do not match. This transcript is machine-produced and the year in either sentence could be a transcription artefact; the number is recorded here as the CFO said it, and the year is flagged as unresolved rather than silently corrected. The Q3 press release settles it.
- Power: 4.2GW contracted, plus ~1.5GW of powered-land options and executed LOIs — "close to about six gigawatts already" — against a stated goal of >8GW of active power by end-2030.
- 500MW of active power added in Q2, 300MW of it in June alone, "higher than any amount of power that we've added in any history prior quarter".
- Contribution margins on recent deals 5–10 percentage points higher than recent quarters.
- Managed inference went "from $1 million to $100 million inside of a single quarter."
- DDTL 5.5 closed (announced 2026-08-10) — a facility that finances shorter-duration customer contracts, which the CEO says opens the enterprise market: "These clients want to buy compute for two years or three years, and that's not a market that was easily accessible to us until we were able to bring DDTL 5.5 to market."
- Vera Rubin is seeing margin expansion "right from the start", and is where much of the 5–10 point step function sits.
- On data-centre moratoriums: "we feel like moratoriums are not going to impact the demand for this infrastructure. They are going to impact where this infrastructure gets built" and "none of those numbers will be impacted by the regulatory pushback as of today."
What the analysts asked about. Supply chain security; managed inference and capacity allocation; the drivers of the margin uplift; regulatory pushback against data centres; fleet evolution for inference; the gap between power added and revenue added; and recontracting GPUs coming off contract. Four of the seven were, underneath, the same question: is the margin improvement structural or is it term mix.
Dodges.
- The margin uplift. Zelnick: "can you help unpack the drivers? How much is a function of shorter duration deals versus strong competitive differentiation or other factors?" Intrator: "So it's a combination of a lot of things, And it's difficult to deconstruct it." He then gave three qualitative reasons — platform quality, customers monetising their own products, premium pricing of a premium product — and no attribution. This matters because the CFO had just credited DDTL 5.5 with opening short-duration contracts, and short-duration pricing is exactly the "other factor" the question was trying to isolate.
- Power-to-revenue linearity. Turrin observed that 500MW went in while sequential revenue adds were "fairly consistent with last quarter". The CFO answered that 300MW landed in June so the effect arrives in Q3/Q4; the CEO then moved to Vera Rubin. The gap itself was never quantified.
- Long-term supply agreements. Chatterjee asked whether CoreWeave needs broader long-term supply agreements. The CEO answered on relationship quality — "it's part of what we do every single day is nurture these relationships" — and named no agreement, term or volume. The analyst's question referenced a counterparty the transcript renders as "Solidine"; that name is not resolvable against anything read in this pass and is not guessed here or in
relationships.yaml.
S5 — Ownership
13F register at 2026-06-30 (3spread, CUSIP 21873S108; 676 filers against 792 the prior quarter — ingestion incomplete, no exit read from an absence; 406.4m of 551m shares, the balance being founder/insider Class B):
| Holder | Shares | $m | QoQ |
|---|---|---|---|
| Magnetar Financial LLC | 52,062,927 | 5,179.7 | −16,503,467 |
| Jane Street Group | 51,478,324 | 5,124.2 | +4,396,533 |
| NVIDIA Corp | 47,213,353 | 4,699.6 | +0 (held) |
| Susquehanna International Group | 26,074,395 | 2,595.4 | −8,792,052 |
| Goldman Sachs Group | 25,994,549 | 2,587.5 | −14,084,998 |
| BlackRock, Inc. | 24,105,966 | 2,399.5 | +14,727,734 |
| Citadel Advisors | 18,743,746 | 1,865.8 | +3,344,189 |
| Invesco Ltd. | 12,408,675 | 1,235.2 | +11,415,570 |
| Alyeska Investment Group | 10,891,267 | 1,084.1 | +3,897,741 |
| Bank of America | 8,272,471 | 823.4 | +2,620,342 |
The largest shareholder is also a lender. Magnetar Financial holds 52.1m shares, and the 10-Q's debt table carries a "Magnetar Loan" maturing January 2029 at a 12% effective rate ($189m at 2026-06-30, down from $273m). Equity and 12% paper in the same hands, with the equity being trimmed by 16.5m shares while the loan amortises. Whatever the intent, the position is paid on either resolution of the capital structure, and that is a different kind of holder from an index fund.
NVIDIA holds 47.2m shares and did not move them. Read this beside the INTC pass in the same slot, where NVIDIA is the fourth largest 13F holder with 214.8m shares, also +0 QoQ. The routine's standing cross-cutting finding has been about contractual support — guarantees, warrants, take-or-pay, vendor loans. This is the equity version, and it is passive: NVIDIA sits on the register of the complex's most leveraged buildout and of its most capital-hungry manufacturer, and traded neither in the quarter.
Goldman trimmed 14.1m shares in the quarter and is, since September, a sales agent and forward purchaser under the equity distribution agreement and a capped-call counterparty on the converts. No inference is drawn from the sequence; both facts are in the record.
- Short interest (FINRA, via Equibles): 51,444,662 (2026-05-15) → 80,963,200 (2026-06-30, the peak) → 64,362,421 (2026-07-15) → 55,568,683 (2026-08-31). A 31% decline off the peak, and the covering ran through the 2026-08-11 results rather than starting after them. Days to cover 2.6 → 2.7 on thinner volume. Equibles also returns a model estimate for 2026-09-15 (~56.5m, +1.7%, P10–P90 49.7–64.7m); it is an estimate, not a FINRA figure, and is not used here.
- Shelf / ATM / buyback: the S-3ASR of 2026-06-05 (333-296553) is live and was drawn on by the 2026-09-17 424B5 for up to 35,000,000 Class A shares with collared forward sales. No buyback. Dilution instruments in issue: two existing convertible series, the new 2033 series (max 52,578,540 shares), the ATM (35,000,000), plus the equity plan. CoreWeave is an issuer of shares from every direction, with $566.2m spent on capped calls to slow one of them down.
S8 — Valuation and sizing ⏳ OPEN (user)
Today. $86.76 (2026-09-22). 551m shares → ~$47.8bn of market value. Net debt at 2026-06-30 was $35,551m − $5,524m = $30.0bn; the September convert adds $4.2bn of principal against $4,137m of proceeds less $566m of capped-call cost, so net debt is roughly unchanged at ~$30.4bn. EV ≈ $78bn, on annualised Q2 revenue of ~$10.3bn — about 7.6x EV/revenue. There is no earnings multiple.
Out-year model. This one is cruder than the other two names in this slot, and the reason is structural: the equity is a $47.8bn sliver on top of $30bn of net debt, so small changes in the enterprise value swing the equity a long way. Assumptions: revenue $19bn in 2027 (the exit-ARR guide read as a run rate) and $30bn in 2028; EBITDA margin 45%, which is roughly what a contract-compute business with the disclosed contribution margins should carry before D&A; net debt at $50bn by 2028, because the build continues; and ~600m shares after the ATM but before any convert conversion.
| FY2028 case | Revenue | EBITDA | EV/EBITDA | EV | Less net debt | Equity | Per share |
|---|---|---|---|---|---|---|---|
| Bear — a top-three contract is renegotiated or delivery slips again; growth stalls | $20bn | $8.0bn | 5x | $40bn | $50bn | ~0 | ~$20, option value only |
| Base — the RPO converts on schedule, margins hold | $30bn | $13.5bn | 8x | $108bn | $50bn | $58bn | ~$97 |
| Bull — 5-10pt margin uplift proves structural, inference scales, >8GW on track | $34bn | $16.0bn | 10x | $160bn | $50bn | $110bn | ~$142 |
What that says. Base lands ~12% above the traded price — the only name in this slot where it does — and the bear case takes the equity to roughly nothing, because at 5x EBITDA the enterprise is worth less than the debt. That asymmetry is the investment case and the objection at once: this is a levered equity on a backlog, and the distribution of outcomes is wide in both directions rather than centred near the price. The September financing package is evidence about the left tail specifically: a market that writes 2.875% paper is not pricing the bear case, and a company that pays $566.2m for capped calls is not either.
Buy price and size are owner-only (skill §5). Liquidity: 26.4m shares on 2026-09-22, ~$2.3bn — ample, though the float is 457m Class A rather than the full 551m.
S9 — Monitoring
- Next earnings: Q3 2026, on the 10-Q cadence (Q1 filed 2026-05-08, Q2 filed 2026-08-12) — early November 2026. A precise date was not obtained; market data's calendar needs a paid plan.
- Feed: Tier 0 EDGAR sweep covers cik 1769628, and it is what promoted this name.
- Leading indicator between quarters: the RPO line in Note 2 together with its 24-month percentage — $103.7bn and 41% at 2026-06-30. Unlike revenue, it moves before delivery does, and unlike ARR guidance it is audited-adjacent and net of management's own delivery-risk haircut. Second: the debt table's next-twelve-months principal column, which the September convert should visibly reduce.
- Trigger that converts this to a buy: the 5–10 point contribution-margin uplift showing up in reported gross margin for two consecutive quarters while RPO grows and the 24-month share holds at or above 41% — i.e. the margin being structural rather than term mix, confirmed in the financial statements rather than in a qualitative answer management declined to decompose.
- Trigger in the other direction: any disclosed delay on a contracted gigawatt, a top-three customer renegotiation, or the material weaknesses still open in the FY2026 10-K.
Access and evidence limits, this run
- The FY2025 10-K and the DEF 14A were not read. The promotion put the September filings and the Q2 10-Q first; the annual risk factors, the auditor's report, auditor tenure and the Class A/Class B voting arithmetic are all unread. Largest gap; it is the scorecard's next action.
- One transcript, not two, and prepared remarks not read — so the 5–10 point margin figure is reported as management restated it in Q&A, not as it was first framed.
- The exit-ARR year is unresolved (analyst said 2027, CFO answered 2026). Recorded as said, flagged, not silently reconciled.
- "Solidine" not resolved. The analyst named a supply counterparty the transcript renders that way; no filing read here corroborates it and no name is guessed.
- No
efts.sec.govfull-text sweep for auditor change or restatement language. - No
GetCustomerConcentration/GetGoingConcernStatus/GetAtmProgramscalls — Equibles budget. Customer concentration came from the 10-Q's own risk factor, which is the better source; the ATM came from the 424B5. - Form 4 / Form 144 not pulled for CRWV — 3spread time, and the lock-up analysis in
CBRS/dossier-2026-09-21.mdis the template for when it is done. - The out-year model is the routine's own, on stated assumptions, and is an EV/EBITDA frame because no earnings exist. On a balance sheet this levered, treat the three cases as a spread of outcomes rather than a valuation.