CRWV · Forward model · Bear case
The Bear case, 18 quarters out
Each segment is projected from its own operating driver, rolled up into consolidated cash flow, and discounted back to a fair value per share. The assumptions below are editable — change them and every number on this page moves with them.
CoreWeave reports a single segment and publishes no revenue split, so this model runs one vertical carrying consolidated revenue. Every actual is a reported figure; nothing is apportioned and no quarter is estimated. Managed inference (>$100M ARR) and the non-GPU services (>$400M ARR) are disclosed only as ARR points inside that same total, so neither is split out - inventing sub-lines for them would apportion reported revenue into segments the company has never published. What is ours: the utilisation reading, the price drift, the margin and capex glides, the tax rate and the exit multiple. Corporate overhead is set to zero deliberately, because adjusted EBITDA is already struck after sales, marketing and general and administrative, and charging an overhead percentage on top would double count it. Segment actuals reconcile to $5,130.9M for 2025 and to $2,078M and $2,575M for 2026 Q1 and Q2. Share count is held flat, so roughly 13% a year of dilution is charged nowhere.
The July pricing step does not hold on renewal and the build slows once the contracted 4.2 GW is spent, while the interest bill does not wait: Q3 interest is guided at $860-940M against $200-260M of adjusted operating income. 2026 still lands inside the guided range, at the bottom of it - the damage is all in what comes after. The fair value goes negative, and that is the case rather than a rounding error: enterprise value falls to roughly the $29.5B of net debt, so the equity is the residual behind the lenders.
Latest: $14.66B (2030Q4E)
| Period | Value |
|---|---|
| 2024Q1 | $189M |
| 2024Q2 | $395M |
| 2024Q3 | $584M |
| 2024Q4 | $747M |
| 2025Q1 | $982M |
| 2025Q2 | $1.21B |
| 2025Q3 | $1.36B |
| 2025Q4 | $1.57B |
| 2026Q1 | $2.08B |
| 2026Q2 | $2.58B |
| 2026Q3E | $3.53B |
| 2026Q4E | $4.31B |
| 2027Q1E | $4.95B |
| 2027Q2E | $5.53B |
| 2027Q3E | $6.07B |
| 2027Q4E | $6.60B |
| 2028Q1E | $7.14B |
| 2028Q2E | $7.69B |
| 2028Q3E | $8.26B |
| 2028Q4E | $8.86B |
| 2029Q1E | $9.47B |
| 2029Q2E | $10.12B |
| 2029Q3E | $10.79B |
| 2029Q4E | $11.50B |
| 2030Q1E | $12.24B |
| 2030Q2E | $13.01B |
| 2030Q3E | $13.82B |
| 2030Q4E | $14.66B |
What drives each segment
CoreWeave Cloud
Capacity × utilisation × priceOne business: contracted GPU clusters sold on multi-year reserved contracts, together with the storage, networking, CPU and managed-inference services layered on top of them. What caps revenue is not demand but energised power - management describes demand from multiple customers for every GPU brought online, against a $104 billion revenue backlog. So the projection here is a build schedule rather than a sales forecast: megawatts switched on, the share of them actually earning across a quarter, and what a megawatt-quarter of CoreWeave Cloud sells for.
Latest: $14.66B (2030Q4E)
| Period | Value |
|---|---|
| 2024Q1 | $189M |
| 2024Q2 | $395M |
| 2024Q3 | $584M |
| 2024Q4 | $747M |
| 2025Q1 | $982M |
| 2025Q2 | $1.21B |
| 2025Q3 | $1.36B |
| 2025Q4 | $1.57B |
| 2026Q1 | $2.08B |
| 2026Q2 | $2.58B |
| 2026Q3E | $3.53B |
| 2026Q4E | $4.31B |
| 2027Q1E | $4.95B |
| 2027Q2E | $5.53B |
| 2027Q3E | $6.07B |
| 2027Q4E | $6.60B |
| 2028Q1E | $7.14B |
| 2028Q2E | $7.69B |
| 2028Q3E | $8.26B |
| 2028Q4E | $8.86B |
| 2029Q1E | $9.47B |
| 2029Q2E | $10.12B |
| 2029Q3E | $10.79B |
| 2029Q4E | $11.50B |
| 2030Q1E | $12.24B |
| 2030Q2E | $13.01B |
| 2030Q3E | $13.82B |
| 2030Q4E | $14.66B |
Assumptions & reasoning
- Every quarter here is consolidated revenue as reported. CoreWeave publishes one reportable segment and no revenue split, so this vertical is the whole company: nothing is apportioned and no quarter is estimated.
- Managed inference ARR went from $1M to more than $100M and the non-GPU services already exceed $400M of ARR, but both are ARR disclosures inside the same total and never segment revenue, so neither is broken out into its own line.
- Utilisation here is the share of energised megawatts earning across the quarter, not a contracted share. It reads 77% for Q2 because 300 of the 500 MW added landed in June, which is also why management says that power comes through in Q3 and Q4.
- Adjusted EBITDA is 92% depreciation and amortisation added back - D&A was $1,393M of the $1,510M. This margin is therefore a pre-depreciation margin on a fleet whose depreciation is the real cost of producing the revenue.
- Capex follows the guided $35-39B for 2026, which includes finance-leased hardware. The stored house free-cash-flow series counts only the cash property and equipment line, so this model burns considerably more than that series shows.
- Share count is held flat at the 551 million diluted shares of Q2, so the 13% year-over-year dilution and the $997M of common stock issued in the quarter are charged nowhere in the fair value.
Where each case comes from
Bear case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the Bear column is what happens if they are taken at face value.
Q2 2026 earnings call, guidance section
Q2 2026 release, debt and interest
Bull case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the Bull column is what happens if they are taken at face value.
Q2 2026 earnings call, pricing and Vera Rubin
Power slips case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the Power slips column is what happens if they are taken at face value.
From cash flow to fair value
| Present value of free cash flow, 18 quarters | -$62.80B |
| Terminal-year revenue | $53.73B |
| Terminal-year EBITDA | $30.04B |
| Exit multiple, on revenue | 2.2x |
| Terminal value | $118.20B |
| Discounted at 15.0% a year, terminal value becomes | $63.02B |
| Enterprise value | $223M |
| Net cash | -$29.54B |
| Equity value | -$29.32B |
| Diluted shares | 0.55B |
| Fair value per share | $-53.21 |
| Against the current price of $89.76 | -159% |
15% on $35.1B of debt against $5.0B of equity, in an engine whose free cash flow charges no interest at all - the guided $860-940M a quarter of it has to be priced somewhere. The exit multiple prices the steady state beyond the build, not the terminal quarter: at 2030 Q4 this model still spends 68% of revenue on capex and free cash flow is negative. Once the fleet only replaces itself, 62% EBITDA less roughly 55% replacement capex is a 7% cash margin, so 3.0x revenue is already about 43x it. The shares trade near 6x forward revenue.
Read the other way round: at $89.76 the market is paying 5.0x terminal-year revenue, holding every other assumption on this page fixed. That is the number to argue about.
The projected path
| Quarter | CoreWeave Cloud | Revenue | YoY | EBITDA | Capex | FCF | R40 | PV of FCF |
|---|---|---|---|---|---|---|---|---|
| 2026 Q3E | $3.53B | $3.53B | +159% | $1.91B | $8.97B | -$7.05B | -41 | -$6.81B |
| 2026 Q4E | $4.31B | $4.31B | +174% | $2.34B | $9.65B | -$7.31B | +4 | -$6.82B |
| 2027 Q1E | $4.95B | $4.95B | +138% | $2.70B | $9.84B | -$7.14B | -6 | -$6.43B |
| 2027 Q2E | $5.53B | $5.53B | +115% | $3.02B | $9.80B | -$6.77B | -8 | -$5.89B |
| 2027 Q3E | $6.07B | $6.07B | +72% | $3.33B | $9.64B | -$6.32B | -32 | -$5.30B |
| 2027 Q4E | $6.60B | $6.60B | +53% | $3.63B | $9.46B | -$5.83B | -35 | -$4.73B |
| 2028 Q1E | $7.14B | $7.14B | +44% | $3.93B | $9.29B | -$5.35B | -31 | -$4.19B |
| 2028 Q2E | $7.69B | $7.69B | +39% | $4.24B | $9.14B | -$4.89B | -24 | -$3.70B |
| 2028 Q3E | $8.26B | $8.26B | +36% | $4.57B | $9.02B | -$4.46B | -18 | -$3.25B |
| 2028 Q4E | $8.86B | $8.86B | +34% | $4.90B | $8.95B | -$4.05B | -12 | -$2.85B |
| 2029 Q1E | $9.47B | $9.47B | +33% | $5.26B | $8.92B | -$3.66B | -6 | -$2.49B |
| 2029 Q2E | $10.12B | $10.12B | +32% | $5.62B | $8.93B | -$3.31B | -1 | -$2.18B |
| 2029 Q3E | $10.79B | $10.79B | +31% | $6.01B | $8.99B | -$2.98B | +3 | -$1.89B |
| 2029 Q4E | $11.50B | $11.50B | +30% | $6.41B | $9.09B | -$2.68B | +7 | -$1.64B |
| 2030 Q1E | $12.24B | $12.24B | +29% | $6.83B | $9.23B | -$2.40B | +10 | -$1.42B |
| 2030 Q2E | $13.01B | $13.01B | +29% | $7.27B | $9.41B | -$2.15B | +12 | -$1.23B |
| 2030 Q3E | $13.82B | $13.82B | +28% | $7.73B | $9.64B | -$1.91B | +14 | -$1.05B |
| 2030 Q4E | $14.66B | $14.66B | +28% | $8.21B | $9.91B | -$1.69B | +16 | -$902M |
Every row is projected. A year-over-year change is shown only where the quarter it compares against exists — an em dash means there is no comparable quarter, not a flat year.
Model revisions
Assumptions are marked to reality as each quarter prints. Every change is appended here, with the fair value the model produced at the time, so the model's own history stays visible.
| Date | Changed | Fair value then | Note |
|---|---|---|---|
| 2026-08-23 | all | $61.24 | First build, on the 2026 Q2 basis, calibrated to the raised full-year guide and the more-than-1.85 GW year-end power path given on the Q2 call. |