{ "docId": "019dd923-622c-750b-8b98-4c28389e5dec", "docSlug": "e9969ec8b548", "documentTitle": "CoreWeave, Inc. (CRWV)", "authorId": "53_Kerrisdale_Capital", "authorName": "Kerrisdale Capital", "documentKindSlug": "research-note", "documentKindLabel": "Research note", "sourceTypeSlug": "short_seller", "sourceTypeLabel": "Short seller", "presentationDate": "2025-09-01 00:00:00", "orientation": "portrait", "aspectRatio": 0.77272725, "pageNumber": 22, "pageCount": 30, "prevPage": 21, "nextPage": 23, "slideType": "preempt_rebuttal", "function": "preempt_rebuttal", "density": "overcrowded", "nDataPoints": 8, "notes": "This slide serves as a detailed critique of a specific financial model, likely presented in a previous slide (referenced as 'this chart').", "elementsJson": [ "paragraph", "numbered_list" ], "metadataConfidence": 0.95, "imagePath": null, "slideHref": "/slides/019dd923-622c-750b-8b98-4c28389e5dec/22", "deckHref": "/decks/019dd923-622c-750b-8b98-4c28389e5dec", "deckJsonHref": "/decks/019dd923-622c-750b-8b98-4c28389e5dec.json", "deckAnchorHref": "/decks/019dd923-622c-750b-8b98-4c28389e5dec#slide-22", "components": [ { "bbox": null, "kind": "callout", "text": "It may look compelling on paper, but the figures collapse under even modest scrutiny.", "attrs": null, "subkind": null, "toolName": "Visual emphasis", "toolSlug": "visual-emphasis", "confidence": null, "componentId": "019dd953-5375-7572-b867-cd01d9b968b6", "frameworkName": null, "frameworkSlug": null }, { "bbox": { "h": 0.09, "w": 0.76, "x": 0.12, "y": 0.85 }, "kind": "list", "text": "3. Free Capital from Nowhere. CoreWeave assumes 15% of total project costs are covered by a ParentCo “equity contribution” with no associated cost. That is economically incoherent. CoreWeave has no retained earnings, no positive cash flow, and its parent entity is not a self-funding machine. We treat that “equity contribution” as additional equipment financing with a 10% interest rate, and repayment of that debt in year 4.", "attrs": null, "subkind": "numbered", "toolName": null, "toolSlug": null, "confidence": null, "componentId": "09331e79-efbe-4e57-bbb9-19dbf91608ae", "frameworkName": null, "frameworkSlug": null }, { "bbox": { "h": 0.13, "w": 0.76, "x": 0.12, "y": 0.48 }, "kind": "list", "text": "1. Overstated Margins from Day One. CoreWeave assumes EBITDA margins hit 80% immediately, a figure a former CoreWeave engineer called “utopian.” In practice, we assume margins build more gradually – starting at 70% in Year 1 and rising to 75% by Year 4 – which better reflects real-world operating ramps and industry feedback. Note, for context, in 2Q25 CoreWeave’s corporate EBITDA margin was 63% and consensus estimates for 2027E are 72%.", "attrs": null, "subkind": "numbered", "toolName": null, "toolSlug": null, "confidence": null, "componentId": "14b7745b-ee28-44d9-90fc-5ba8163eeed9", "frameworkName": null, "frameworkSlug": null }, { "bbox": { "h": 0.22, "w": 0.76, "x": 0.12, "y": 0.62 }, "kind": "list", "text": "2. Residual Value Assumptions Border on Fantasy. CoreWeave assumes it can recoup significant residual value from a glut of aging GPUs – despite the fact they will be five years old in a rapidly changing market. The company says its estimate is based on “current pricing data,” but that’s meaningless – there is no liquid market for obsolete AI chips five years out. Residual value will depend on a cocktail of unknowables: future architecture shifts, the rise of custom silicon, and even trade politics. Assuming a meaningful back-end payoff is pure speculation. We model a -60% drop in Year 5 revenue versus the contract period average, followed by another -15% in Year 6. This is entirely consistent with industry precedent: neocloud often feature steep discounts for longer lock-ins, and revenue degradation of -80% from on-demand Year 1 pricing is plausible given the rate of chip obsolescence.", "attrs": null, "subkind": "numbered", "toolName": null, "toolSlug": null, "confidence": null, "componentId": "66012a96-d8bf-4b51-bef8-097f831e5293", "frameworkName": null, "frameworkSlug": null }, { "bbox": null, "kind": "metric", "text": "EBITDA margin: 80%", "attrs": null, "subkind": "primary", "toolName": "Quantification", "toolSlug": "quantification", "confidence": null, "componentId": "019dd953-5375-7572-b867-d569fb4475ef", "frameworkName": null, "frameworkSlug": null }, { "bbox": { "h": 0.13, "w": 0.76, "x": 0.12, "y": 0.34 }, "kind": "paragraph", "text": "Unsurprisingly, the “illustrative” contract being used to reassure investors touts astonishing IRRs: 41% during the contract term, rising to 60% when residual value in Years 5 and 6 is included (see Appendix III). It may look compelling on paper, but the figures collapse under even modest scrutiny. After reviewing the assumptions with multiple cloud infrastructure professionals, we believe the contract is engineered to tell a flattering story rather than reflect typical outcomes. A more realistic model paints a very different picture. We highlight three key assumptions that warrant serious scrutiny:", "attrs": null, "subkind": "paragraph", "toolName": null, "toolSlug": null, "confidence": null, "componentId": "0714c2c4-2fdb-4d89-a2e4-64fde2684c5b", "frameworkName": null, "frameworkSlug": null }, { "bbox": { "h": 0.12, "w": 0.76, "x": 0.12, "y": 0.21 }, "kind": "paragraph", "text": "What equity investors may have missed in this chart is the tightness of the cash flows servicing the debt during the contract period. In Year 1, the debt service coverage ratio (EBITDA / (interest + principal amortization)) is only 1.21x – thin even with a take-or-pay agreement and a creditworthy counterparty. It leaves little room for execution mistakes like cost overruns or hardware performance issues. There's little margin for error for a business exposed to potential cost spikes, hardware underperformance, or delays.", "attrs": null, "subkind": "paragraph", "toolName": null, "toolSlug": null, "confidence": null, "componentId": "451f2513-2f64-4df3-8584-5742e7ba5472", "frameworkName": null, "frameworkSlug": null }, { "bbox": { "h": 0.12, "w": 0.76, "x": 0.12, "y": 0.09 }, "kind": "paragraph", "text": "CoreWeave's illustrative contract shows that in year 3, cumulative cash flows from the deal turn positive after debt service ($258 million green box). In year 4, the debt is paid off. By years 5 and 6, CoreWeave is generating considerable cash flow from a cohort of 5-year-old chips it assumes it can re-contract or sell on-demand at a -25% ASP discount to the contract price for the entire cluster.", "attrs": null, "subkind": "paragraph", "toolName": null, "toolSlug": null, "confidence": null, "componentId": "4930d75d-8211-4833-8d5f-604a3c27522e", "frameworkName": null, "frameworkSlug": null }, { "bbox": null, "kind": "quote", "text": "Overstated Margins from Day One. CoreWeave assumes EBITDA margins hit 80% immediately, a figure a former CoreWeave engineer called “utopian.”", "attrs": null, "subkind": null, "toolName": "Authority citation", "toolSlug": "authority-citation", "confidence": null, "componentId": "019dd953-5375-7572-b867-d197f9ebbc7a", "frameworkName": null, "frameworkSlug": null } ], "metrics": [], "tools": [], "frameworks": [], "arcBeats": [], "loops": [], "imagePathAlt": null, "thumbSrc": null, "thumbSrcAlt": null, "locked": true }