CoreWeave面临高额资本支出压力
CoreWeave面临高额资本支出压力
I'll pull the current quote, recent news, and the latest financial data in parallel.
Conclusion: Demand is not what limits CoreWeave. Conversion is limited by cash. The "10× demand" figure measures qualified demand against deliverable capacity through 2027, not contracted revenue. Turning it into earnings requires about $2.90 of capex for every $1 of guided 2026 revenue, funded mainly by debt and customer prepayments, with GAAP losses persisting into 2028. The shares are under pressure because the market is pricing that funding and capital risk, plus Nvidia-related sentiment, rather than doubting demand.
Data as of: Quotes are as of the Nov 12, 2026 close (Thursday, NY time). Financials are for the quarter ended Sep 30, 2026, from the Nov 10 release. The 10-Q was filed Nov 12. Call commentary is from the Nov 10 call.
1. Demand and pricing are strong
- Management said it has 10 times as much qualified demand as available capacity through end-2027, and 2028 demand is already several times its capacity, not yet marketed.
- Pricing is rising. CoreWeave raised prices about 25% in July and about 10% later, and Ampere and Hopper pricing rose 30–50% quarter over quarter. Short-term capacity sells above $40M per megawatt a year, versus $20–25M for 4–5 year contracts.
- Renewals are a source of profit: 2027 renewals at current pricing would add about $4B of annualized revenue with no new capex, rising to nearly $10B including 2028.
- Revenue is real: Q3 revenue was , up year over year. Backlog was , about the FY2026 revenue guidance midpoint. Guidance was raised to for FY2026, with an exit run-rate of .
2. The conversion cost is high
- The Q3 quarter-end cash balance was , excluding restricted cash.
- Implied Q4 capex at the guidance midpoint is , a step-up from Q3.
- Depreciation and amortization was in Q3, versus $630M a year earlier. That is the main reason GAAP profit lags the adjusted figures.
3. Financing is the swing factor
- Management says it will not use the unsecured parent-level debt market at current yields through 2027. Its weighted-average cost of debt is , down 250 basis points year over year.
- Funding is shifting to customer prepayments and asset-level debt. CoreWeave expects of prepayments and deposits in the next three months, and collected prepayments on 70% of 2026 deals.
- Q4 interest guidance is , about above Q3.
- CoreWeave also raised of convertible notes and set up an at-the-market program for up to 35 million shares. That makes dilution a live risk.
4. Why the shares fell
- The stock rose about from the Nov 10 close of to on Nov 11. It then closed Nov 12 at , down . That is still below the 52-week high of .
- Nvidia, which owns more than 10% of CoreWeave and is a key chip supplier, saw its CEO resign with immediate effect. The news hit AI infrastructure names on Nov 12.
- Barclays cut its target from $105 to $85 and kept an equal-weight rating, a change, per .
5. Execution risks, ranked
- Funding cost and access. Interest already absorbs about 37.5% of adjusted EBITDA. Heavier debt or any parent-level equity issue would hit per-share value. Management expects EBITDA less interest to more than double in 2027, and adjusted net income to turn positive in the first half of 2028. Both depend on funding staying available.
- Pricing durability. Margins depend on scarcity pricing. Over 130,000 GPUs come off contract in 2027 and about 170,000 in 2028. Renewal pricing has to hold for the economics to work.
- Delivery pace and power. Active power is above 1.7 GW, with a year-end target of more than 1.85 GW and 4.6 GW contracted. Management pulled capex forward to deliver sooner, which raises near-term funding needs and vendor dependence.
- Concentration. The backlog is concentrated in a few large customers. An AI-generated of JPMorgan's list names Meta, OpenAI, Microsoft, Nvidia and Jane Street. That source is secondary and unverified. Nvidia is also a supplier and shareholder, which ties sentiment to its governance.
- Return on capital. A short-seller cited in the same secondary article puts capital intensity near $4 per $1 of revenue. This is not verified, but it is consistent with the capex figures above.
What would change the view
- Positive: Q4 revenue within or above the $13.0–13.3B range, active power at or above 1.85 GW, renewal pricing holding, and funding through asset-level loans and prepayments rather than equity.
- Negative: a rise in funding costs, a new equity or convertible issue at the parent, pricing on renewals falling, or further Nvidia-related disruption.
Limits: The transcript text is a partial capture, and the balance-sheet and cash-flow line items were read from a text extract of the release. The structured total-debt figure reconciles to the release once leases are included. Nothing here is a price target or a recommendation.
Sources
- CoreWeave Q3 2026 results press release, Nov 10, 2026 (investors.coreweave.com; the same release on BusinessWire)
- Q3 2026 earnings call transcript, Investing.com, Nov 10, 2026 (partial capture)
- Driven Market Data: CRWV quote (Nov 12, 2026 close) and quarterly balance sheet (filed Nov 12, 2026)
- Yahoo Finance, Nov 12, 2026: Nvidia CEO resignation and CoreWeave share move
- MarketBeat, Nov 12, 2026: Barclays target change
- AInvest, Nov 11, 2026 (secondary, AI-generated summary; used only for context)