Rankings Software, cloud & internet CoreWeave
CoreWeave
40
15070100
A view of the same findings, not a new score. How it works
The model found several serious problems. That does not mean the stock will fall.
Biggest deductions
- Debt and financing needs−10
- Peers and EBITDA quality−7
- Margin and ROIC−6
Strengths
- Contracted order book+4
- Pricing power+2
How the score is built
From 100 to 40, step by step: points off for the business, the price and the cracks, back for documented strengths, then three data adjustments.
Start
Score
Hover or select a step to see what drove it. Steps with findings open them below.
Key numbers
As of 2 Nov 2026. Analyst targets tend to be optimistic; read upside as a hint, not a forecast.
- Price
- 50.15 USD
- Median analyst target
- 135.00 USD
- Upside to median
- +169.2%
- Analysts
- 42
- Target range
- 39.00 – 278.00 USD
- 52-week change
- −62%
- P/E (trailing)
- n/a
- Forward P/E
- n/a
- PEG
- n/a
Findings
Every deduction and every plus, with its source. Open a step to read them.
Business
Is the company getting stronger, not just bigger? · 4 findings−19 of −30
- −6Margin and ROICOperating margin −2.7% and net margin −25% TTM; ROIC −0.3%. Gross margin has fallen from 74–76% in 2024–25 to 66% in Q2 2026, and adjusted operating margin was only 5% in Q2 (USD 128m vs 200m a year earlier). Depreciation (4.0bn TTM) and interest eat all of EBITDA. The loss is counted only here.
- −6Capital allocation and dilutionCapex guidance for 2026 is USD 35–39bn against revenue of 12.4–13.2bn (about 3 times revenue). The share count rose about 13% in a year, share-based pay USD 619m (8% of revenue), USD 3.0bn in new shares TTM (incl. Nvidia and Jane Street), and in September came an ATM programme of up to 35m shares on top of convertible loans with a conversion price of USD 97.85.
- −3MoatThe core service is rental of Nvidia GPUs, which the hyperscalers, Nebius, IREN and Oracle also offer. The software layer is growing (managed inference ARR >100m, non-GPU products >400m ARR), but is under 5% of run-rate revenue; no real lock-in beyond the contract terms.
- −4Quality of growthRevenue is growing 112%, but the growth is debt-financed and does not turn into earnings: Q3 guidance implies interest expenses of USD 860–940m vs adjusted operating profit of only 200–260m, and consensus expects EPS of −3.94 in 2026 vs −2.81 in 2025 (the loss grows with revenue).
Price
Today's multiples against the company's own history and its peers. · 2 findings−13 of −35
- −6Multiples vs own historyEV/Sales 12.6 vs 10.2 at the end of 2025 and EV/EBITDA 25.3 vs 21.6; P/B 10.1 on equity of only USD 5.0bn. Only one year of history, but the multiples have risen even though the stock is -34% in a year, because debt has grown faster than revenue.
- −7Peers and EBITDA qualityEV/EBITDA 25 vs Oracle 15.5 and EV/Sales 12.6 vs Oracle 7.4 (Nebius 48 and IREN 26 are more expensive). Forward P/S 2.5 looks cheap, but half of the EV of 94.8bn is net debt (46bn), and EBITDA is misleading when GPUs are depreciated over six years and interest is 3.6bn annually; negative EPS and FCF -28.9bn are expected in 2026.
Cracks
Cash flow, customers, debt, insiders, guidance and what sits behind the headline numbers. · 6 findings−28 of −35
- −10Debt and financing needsTotal debt USD 51.6bn incl. leasing, net debt 46bn, debt/EBITDA 9.2, interest coverage −0.1, Altman Z 0.29, current ratio 0.46 and negative working capital of 11.4bn. The most expensive layer is 9.75% bonds due 2031 (2.75bn). Interest paid rose to 1.3bn TTM, Q3 interest guided at approx. 0.9bn for the quarter, and management itself says more debt and/or equity will be necessary.
- −6Customer concentrationMicrosoft accounted for approx. 67% of revenue in 2025; the largest customer was still 45% in Q1 and 36% in Q2 2026. The order book of USD 104bn sits with few buyers: OpenAI with up to 22.4bn in commitments (itself dependent on external financing), Meta and Anthropic. The value of the contracts depends on these customers' ability to pay.
- −4GPU depreciation and residual valueGPUs and network equipment are depreciated over six years (extended from five in 2023), which lowers depreciation by several hundred USD million a year. Critics put the useful life at 3–5 years under full load; with new Nvidia generations every year there is a risk of write-downs and lower prices on re-leasing.
- −4Insider selling and voting controlThe founders Intrator, McBee and Venturo have sold over USD 2.3bn since the IPO, and Magnetar over 5.5bn (stake halved); Intrator sold again in April, July and August 2026. A/B structure in which the founders' B shares give voting control, so minority shareholders have little influence. No insider purchases found.
- −2Circular Nvidia relationshipNvidia is at once supplier of nearly all hardware, shareholder (bought for USD 2bn at USD 87.20/share in January 2026) and customer/backstop (USD 6.3bn agreement to buy unused capacity). That artificially supports demand and makes CoreWeave dependent on one supplier's allocation and pricing.
- −2Market signalShort interest 16.7% of free float, the stock −34% over 52 weeks and below both the 50- and 200-day averages.
Plus
Only documented strengths. · 2 findings+6 of +15
- +4Contracted order bookOrder book of USD 104bn as of 30/6 (+246%) plus over 25bn in new commitments at the start of Q3; 4.2 GW of contracted power. Customer prepayments of USD 4.8bn TTM co-finance capex, and 2026 guidance was raised to 12.4–13.2bn.
- +2Pricing powerRaised prices by about 25% across GPU types in July, and the Q2 contracts have 5–10 percentage points higher expected contribution margin than the previous quarters.
Caveats
What the score does not capture, or where the data is weak.
Industry set to the closest: Software, cloud & internet (AI cloud/GPU rental, in reality capital-intensive infrastructure). Listed only since March 2025, so own multiple history is one fiscal year. Stockanalysis' share growth of +76% is distorted by the IPO conversion; the real increase is c. 13% (487 → 552m shares). Total debt of USD 51.6bn includes lease liabilities; the company itself reported c. USD 35bn in loans at 30/6 before new convertible loans of 4.2bn in September. The Core Scientific acquisition was rejected by Core Scientific's shareholders in October 2025 and has not been resumed. Not flagged cyclical: revenue rests on multi-year take-or-pay contracts, but GPU prices and residual values are a real risk. Price USD 91.72, market cap 48.7bn, EV USD 94.8bn (7 Oct 2026).
Also in Software, cloud & internet
The highest scores in the same industry.
