CoreWeave's $104 Billion Backlog Cannot Pay a $900 Million Interest Bill

$CoreWeave, Inc.(CRWV)$ reports Q3 2026 after Tuesday's close, and the Street will price the print off revenue: consensus sits near $3.54 billion, up roughly 159% year on year. That headline is almost irrelevant. The number that decides this company's fate is the interest bill — guided to $860-940 million for the quarter, up from $640 million last quarter and $267 million four quarters ago. CoreWeave is the largest pure-play renter of $NVIDIA(NVDA)$ GPUs, and it finances every one of them. Tonight's question is not whether the backlog grows. It is whether operating profit grows at least as fast as the interest expense. At the guidance midpoint, the answer is no — by a factor of more than two.

Figure 1 — The four numbers framing tonight's print: revenue guidance, the interest bill, the backlog, and the market's price on the equity. Sources: CoreWeave Q3 2026 guidance via Money & Markets, 9 Nov 2026; Gate Research note, 8 Nov 2026.

The quarter is an interest story wearing a growth costume

Start with the scissors, because it is the whole thesis. CoreWeave guided Q3 adjusted operating income to $200-260 million. The midpoint, $230 million, is roughly $100 million more than last quarter. Interest expense is guided to $860-940 million; the midpoint, $900 million, is roughly $260 million more than last quarter's $640 million. So the cost of carrying the machine is growing about 2.6 times faster than the profit it throws off — and operating income still covers barely a quarter of the interest bill. Twelve months ago the interest bill was $267 million a quarter; it has since grown 237%, computed from those reported and guided figures, while revenue is guided up about 37% sequentially from $2.58 billion.

Figure 2 — Quarterly interest expense, $267M a year ago to $640M last quarter to a $900M guidance

midpoint: +237% in four quarters. Sources: CoreWeave quarterly results and Q3 guidance via Money & Markets, 9 Nov 2026; midpoint computed.

None of this makes CoreWeave a failed business. It makes it a leveraged infrastructure developer whose equity is, arithmetically, a call option on refinancing terms. Tonight's report cannot change that arithmetic — the guidance already told us where both lines land.

Backlog, debt and market cap are three different numbers

The bulls' favorite number is the backlog: roughly $104 billion, with more than $25 billion of new customer commitments added early in the third quarter. The bears' favorite number is the debt: about $50 billion. The market's verdict sits below both — the equity is worth approximately $25 billion, after the stock hit a 12-month low, down more than 50% from its highs, including a roughly 25% loss in the week before 16 October.

Hold those three numbers together and the disconnect is obvious: the contracted future is four times the market cap, and the debt is twice the market cap. Why would the credit market price a company with $104 billion of contracted revenue at a 16% yield on its 2030 bonds? Because lenders do not get paid by backlog headlines — they get paid by cash flow, and the guided cash flows do not reach the interest bill for years. $NVIDIA(NVDA)$ has partially patched this by agreeing to buy CoreWeave's unused capacity through April 2032, but a backstop with a 2032 expiry is a bridge, not a balance sheet.

Figure 3 — Backlog $104B, total debt roughly $50B, market cap roughly $25B: the contracted future and the borrowed present both dwarf the equity. Sources: backlog and debt via Money & Markets, 9 Nov 2026; market cap and drawdown via Gate Research note, 8 Nov 2026.

The equity-bond disagreement is the most interesting signal here. Sell-side targets sit above $130 against a $48 share price — implying roughly 170% upside. The 2030 bonds yield near 16% — implying real default risk. Somebody is very wrong. My read: the bond market is doing the arithmetic in Figure 2, the equity market the backlog arithmetic in Figure 3, and one of them is about to be corrected by a cash flow statement.

What the guidance actually implies

Management's own targets frame the machine. Full-year 2026 revenue guidance has been raised to $12.4-13.2 billion, with a year-end annualized run-rate target of $18.5-19.5 billion and active power above 1.85 gigawatts. To get there, CoreWeave guided this quarter's capital expenditure to $11.5-13.5 billion — one quarter of guided capex at roughly 3.5 times guided revenue. Every incremental dollar of revenue arriving over the next year has already been pre-financed at whatever the debt market charges, and the debt market currently charges around 16% on the 2030 maturity.

Figure 5 — Q3 guidance in cash terms: revenue $3.45-3.6B, interest $860-940M, capex $11.5-13.5B per quarter; capex runs about 3.5x revenue at the midpoints. Sources: CoreWeave Q3 2026 guidance via Money & Markets, 9 Nov 2026; ratio computed.

This is not a criticism of the model — hyperscalers pre-finance too. The difference is that Microsoft and Oracle pre-finance off investment-grade balance sheets. CoreWeave pre-finances at high-yield rates, so its cost of winning a contract is structurally higher than its customers' cost of building the same capacity. That spread is the hidden tax on every dollar of the $104 billion backlog.

Bull Case

  1. Revenue beats the top of the $3.45-3.6 billion guide and management reiterates the $18.5-19.5 billion year-end run-rate target — that falsifies the demand-digestion thesis in one print.

  2. Backlog conversion accelerates: the $25 billion of new Q3 commitments start converting, and live power moves decisively toward 1.85GW.

  3. The Nvidia capacity backstop through April 2032 caps utilization risk on newly deployed clusters for five-plus years.

  4. The refinancing window reopens: Morgan Stanley, UBS and HSBC all called the sector selloff overdone this week per Reuters, and a risk-on tape would let CoreWeave term out debt below 16%.

  5. Positioning is washed out: the stock sits at a 12-month low after a more than 50% drawdown, so a clean print has less overhead supply than the megacaps.

Bear Case

The bear case is one line of arithmetic, and it is already guided:

  1. Adjusted operating income of $200-260 million covers roughly 23-28% of the guided interest bill of $860-940 million — the quarter is cash-negative before a single growth initiative.

  2. Money & Markets' bridge math puts the implied adjusted loss at roughly $1.43 per share at the guidance midpoint, versus the $1.28 consensus — an "in-line" quarter still loses more than the Street models.

  3. The 2030 bonds yield near 16% — the lender's honest forecast of refinancing terms, which equity holders inherit first.

  4. Post-earnings volatility is violent: CoreWeave shares have averaged about a 15% move in the week after results, per the Gate Research note.

  5. The $11.5-13.5 billion quarterly capex guide means the debt balance is still growing while interest coverage shrinks — the scissors widen before they close.

Figure 4 — From last quarter's $1.14 adjusted loss, the guided interest step adds roughly $0.29 per share across about 550 million shares, implying a $1.43 loss against $1.28 consensus. Sources: Money & Markets preview math, 9 Nov 2026.

Strategic Slam

On $CoreWeave, Inc.(CRWV)$ : if you believe the backlog, the bonds are the better instrument — you get paid 16% to wait for exactly the confirmation the equity needs, and you sit ahead of shareholders if the math breaks. The equity is a refinancing trade, not a growth trade. On $NVIDIA(NVDA)$ : its 17 November print, with data-center revenue expected up more than 12% sequentially per Reuters, sets the demand backdrop the backlog depends on — but Nvidia owns the same story with none of the leverage, which is why the backstop economics favor Nvidia, not CoreWeave.

Catalyst Calendar

Figure 7 — The dated path: tonight's print, Nvidia's read-through, the year-end run-rate target, and the April 2032 backstop expiry. Sources: Money & Markets, 9 Nov 2026; Reuters via Economic Times, 10 Nov 2026; Gate Research note, 8 Nov 2026.

  • 10 November 2026: CoreWeave Q3 2026 results, after the US close — consensus near $3.54 billion revenue and a $1.28 adjusted loss per share.

  • 17 November 2026: Nvidia Q3 results after the close — data-center revenue expected up more than 12% sequentially.

  • 31 December 2026 (company target): year-end annualized run-rate of $18.5-19.5 billion; active power above 1.85GW.

  • April 2032: expiry of the Nvidia commitment to purchase unused CoreWeave capacity.

Where the thesis confirms or breaks

Figure 6 — The two dials that matter tonight: where interest lands inside the $860-940M guide, and whether backlog conversion stays on schedule. Sources: CoreWeave Q3 guidance via Money & Markets, 9 Nov 2026.

Verdict

Structurally bearish on the equity into the print, and unimpressed by any backlog headline tonight. The company will almost certainly report spectacular growth — the guidance already promises it. But at the guidance midpoint, the interest bill grows $260 million this quarter while operating income grows $100 million, the implied loss deepens to roughly $1.43 per share against $1.28 consensus, and one quarter of capex runs 3.5 times revenue. The real price discovery is happening in the bond market at 16%. Watch two numbers tonight — interest expense and adjusted operating income — and ignore everything above the fold.

Key Takeaways

  1. CoreWeave reports Q3 after the close on 10 November; consensus is roughly $3.54 billion revenue, up about 159% year on year, with a $1.28 adjusted loss.

  2. Guided interest expense of $860-940 million has grown 237% in four quarters; adjusted operating income of $200-260 million covers barely a quarter of it.

  3. The scissors: interest grows about $260 million quarter on quarter at the midpoint, operating income about $100 million.

  4. Backlog of roughly $104 billion and debt of about $50 billion both dwarf a $25 billion market cap; the 2030 bonds yield near 16%.

  5. The verdict is decided by interest expense and refinancing terms, not by the backlog print.

If the backlog is real money, why is the lender charging 16% while the equity costs $25 billion — and which one do you trust tonight?

Tickers: $CRWV(CoreWeave)$ $NVDA(Nvidia)$

Sources: CoreWeave Q2 2026 results and Q3 2026 guidance — revenue $2.58B then $3.45-3.6B guided; adjusted operating income $200-260M guided; interest $267M, $640M, $860-940M; capex $11.5-13.5B; FY26 $12.4-13.2B; backlog ~$104B; debt ~$50B; ~550M shares — via Money & Markets, 9 Nov 2026. Consensus ~$3.54B (+159% y/y), -$1.28 EPS, stock ~$48 at a 12-month low, ~$25B market cap, targets above $130, ~16% 2030 bond yield, Nvidia backstop through April 2032, ~15% average post-earnings week move — via Gate Research note, 8 Nov 2026, and Money & Markets. Nvidia 17 Nov earnings date, expected +12% sequential data-center growth, and Morgan Stanley/UBS/HSBC/Jefferies commentary — via Reuters (Economic Times), 10 Nov 2026.

Disclaimer: This article is for information and discussion only and does not constitute investment advice. Figures are drawn from the named sources above; derived numbers are computed from those sources and labeled as such. Markets involve risk; do your own research.

# 💰Stocks to watch today?(11 November)

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Report

Comment

  • Top
  • Latest
empty
No comments yet