The backlog does not own Thursday’s CoreWeave print. The debt tape does. CoreWeave Class A shares closed on Thursday 15 October 2026 at $69.69, up $3.91, or 5.94 percent, from Wednesday’s $65.78. The bounce did not repair the week: the same shares closed at $82.08 on Friday 9 October. Using the last 90 daily log-returns in the Nasdaq file, and 53 New York sessions from Friday 16 October through Thursday 31 December 2026, the bull case is $107.76, the base case is $55.57, and the bear case is $35.73. Those three numbers are this stock’s own volatility, not a Street target. The fact the usual write-up still skips is the maturity gap Chamath Palihapitiya wrote down on 13 October: CoreWeave borrowed for five years against customer contracts averaging about three. A revenue backlog of about $104 billion can be real and still be the wrong length of money for bonds that traded below 80 cents on the dollar on Wednesday.
Key facts
- Nasdaq official close, Thursday 15 October 2026: $69.69, up $3.91, or 5.94 percent, on 130,537,200 shares. Day range $59.51 to $69.76. Source: Nasdaq historical file saved for this article.
- Nasdaq premarket, Friday 16 October 2026, 7:04 a.m. Eastern: last sale $66.6004, down $3.0896, or 4.43 percent. Status was still premarket. Not a close, and not a chart bar. Source: Nasdaq quote file saved that morning.
- Quarter ended 30 June 2026: revenue $2,575 million, operating loss $49 million, net interest $640 million, net loss $626 million ($1.14 a share). Adjusted EBITDA $1,510 million. Adjusted operating income $128 million. Source: CoreWeave earnings exhibit, 11 August 2026.
- Revenue backlog approximately $104 billion at 30 June 2026, excluding more than $25 billion of net new commitments added early in the third quarter. Source: the same exhibit. Reuters, via U.S. News on 14 October 2026, put the figure at $104.2 billion, up from $99.4 billion.
- At 30 June 2026: cash $5,524 million, debt $35,068 million, operating lease liabilities $16,319 million, stockholders’ equity $5,024 million. Leases not yet commenced: about $35.5 billion of undiscounted future payments. Source: Form 10-Q.
- Wednesday 14 October close $65.78, down 15.46 percent, on 111,214,200 shares. Reuters reported the 2032 unsecured notes below 80 cents on the dollar, citing Bloomberg and Trace.
- Short contracts of about three to six months were pricing near $40 million per megawatt, the company said on 17 September 2026. Annualizing second-quarter revenue over 1.5 gigawatts gives about $6.87 million per megawatt per year.
- Deutsche Bank’s Brad Zelnick kept Buy and a $150 target on 15 October and called the selloff a buying opportunity. That $150 is his number, not the $107.76 bull here. Source: Investing.com, via Yahoo Finance.
What just happened, and why the obvious reading is wrong
The obvious reading is that the business broke. It did not, on the last filing or this week. CoreWeave’s investor events page, checked on 16 October, listed nothing. No earnings release landed between 9 October and 15 October. The credit price moved, and the equity followed.
From $82.08 on Friday 9 October to $69.69 on Thursday, the drop is 15.10 percent. Monday closed at $80.26, Tuesday at $77.81, Wednesday down 15.46 percent at $65.78. Thursday opened at $61.43, ranged from $59.51 to $69.76, and closed at $69.69. In the saved Nasdaq file, $59.51 is the lowest daily low since 1 October 2025.
Tuesday volume was 34,284,880 shares, Wednesday 111,214,200, Thursday 130,537,200. That is repositioning. The second quarter was not a lost book: revenue rose from $1,212 million to $2,575 million, and the operating loss was $49 million. The $626 million net loss is that operating loss, minus $640 million of interest, plus $125 million of other income, minus $62 million of tax. A funding argument, not a revenue miss.
The 5.94 percent bounce is the second wrong reading. It leaves the stock 15.10 percent under the prior Friday. At 7:04 a.m. Eastern on 16 October, still premarket, the last sale was $66.6004, and it is not a chart bar. The backlog does not answer the bondholders either. The 11 August exhibit defines it as remaining performance obligations plus other estimated amounts, subject to delivery. Adjusted operating income was about 5 percent of revenue. It is not cash.
Who is exposed
CoreWeave, Nasdaq CRWV, has been public since the March 2025 offering at $40. At 30 June the 10-Q shows 457 million Class A shares and 94 million Class B. Times $69.69, that 551 million count is about $38.40 billion, only about $3.3 billion above $35,068 million of debt. It is not a live market cap. A later at-the-market program allows up to 35 million shares, and a convertible closed after the quarter.
Book equity is $5,024 million against $72,046 million of liabilities, about 14.3 times. Of the debt, $31,405 million is recourse and $3,663 million is non-recourse. Cash of $5,524 million covers about 8.6 quarters of the second-quarter interest bill, not the next build. Customers A and B each held 32 percent of receivables. The exhibit also names Bentley Systems, Caterpillar, Grammarly, Isomorphic Labs, Databricks, and Hudson River Trading, plus a $1 billion Jane Street investment.
Nvidia is on more than one side. TipRanks reported a January investment of $2 billion at $87.20 and a $6.3 billion unused-capacity purchase through 2032. Thursday is about 20 percent below that $87.20. A purchase agreement is not a bond guarantee.
Credit is the other company. TipRanks put the 2032 yield near 16 percent, against a 9.625 percent coupon, and wrote that prices implied a better than 60 percent chance of default inside five years. On 1 October zerohedge wrote that Goldman Sachs prime brokerage showed the 9 1/4 percent 2030s as the most-shorted high-yield bond. Zelnick said the company had closed a $4.2 billion 2.875 percent convertible after a $2.6 billion loan at SOFR plus 5.50 percent. An earlier FinanceFeeds report put a conversion price near $97.85 next to a Redburn sell of $54, and Truist moved to Buy while cutting its target to $126. Neither figure is today’s bull or bear. Leases add $16,319 million of liabilities, plus about $35.5 billion of payments not yet commenced. Do not add that $35.5 billion to the debt.
What the tape and the filings actually show
The equity tape and the filings are not the same firm. The right column is the 10-Q, the 11 August exhibit, or a named report of the bonds.
| Item | Equity tape | Credit and filings |
|---|---|---|
| Thursday 15 October | Close $69.69, up 5.94 percent. Range $59.51 to $69.76. | Wednesday’s 2032 notes were reported below 80 cents. The equity bounce did not reprint that bond. |
| The week | From $82.08 on 9 October to $69.69, down 15.10 percent. Wednesday alone was down 15.46 percent. | No earnings release. The events page was empty on 16 October. |
| Since October 2025 | 1 October 2025 close $137.05, now 49.15 percent lower. File high $143.08 is 51.29 percent higher than Thursday. | 30 June debt $35,068 million against book equity of $5,024 million. |
| Last quarter | The share is what remains after interest and dilution. | Interest of $640 million was five times adjusted operating income of $128 million. |
| Backlog versus the Street | TipRanks average $142.78 is 104.9 percent above $69.69. Zelnick’s $150 is 115.2 percent above it. | Backlog about $104 billion at 30 June, before more than $25 billion added early in the third quarter. |
| One megawatt | Short contracts near $40 million, three to six months. | Second-quarter revenue annualized over 1,500 megawatts is about $6.87 million a year, roughly 5.8 times lower. |
| This article’s year-end | Bull $107.76, up 54.63 percent. | Base $55.57, down 20.26 percent. Bear $35.73, down 48.73 percent. |
The 6 October close of $91.72 is 24.02 percent above Thursday, and the file high of $143.08 has already crossed the bull line. Active power was 1.5 gigawatts at 30 June, against contracted power of about 3.7 gigawatts, later described as about 4.2 gigawatts. FinanceFeeds has traced the contract price. A 7 October plan for 240 megawatts in Navi Mumbai did not hold the $88.45 close that day.
What they have said, and what is not scheduled
Michael Intrator, co-founder, chairman, and chief executive, told the demand story in the 11 August release. “CoreWeave reached an important inflection point this quarter as our scale began to translate into expanding operating leverage. Customer demand is accelerating, as enterprise adoption broadens and we continue to deepen our technology platform,” he said. That sentence is about scale. It is not about the 2032 bond.
On Wednesday, with the stock down about 15 percent, he answered staff in a memo. Reuters reported that he said the market can sometimes “misunderstand and act irrationally,” and that rivals’ delayed offerings and debt financings “are bleeding into the narrative around CoreWeave.” He told employees the backlog is large enough to more than support the debt. The bond was still reported under 80 cents.
Jensen Huang, Nvidia’s founder and chief executive, was Thursday’s equity answer. TipRanks, citing Bloomberg, reported that he said CoreWeave was becoming financially stronger and did not need Nvidia to rescue it. The words TipRanks put in quotes were that the “cost of capital is actually decreasing.” The same piece pointed at a 2032 yield near 16 percent. A new 2.875 percent convert can lower one sleeve while older unsecured notes fall. Thursday’s rise is a comment, not a refinancing.
Brad Zelnick kept Buy and $150. Investing.com quoted him: “The vast majority of CoreWeave’s backlog is associated with long-term take or pay contracts where contracted revenue fully covers all capital and operating costs associated with the contract and generates a return for the company.” He put 60 to 65 percent of the backlog with investment-grade customers at the end of the second quarter, and he forecast about $10 billion of operating cash flow over 12 months against about $45 billion of capital spending. That gap, near $35 billion, is the sentence that matches the bond market.
The Redburn $54 sits $1.57 under this base case by drift, not by design. The full-year range stayed on a call this piece did not read.
Where the credit tension actually sits
Palihapitiya wrote on 13 October that CoreWeave borrowed for five years against customer contracts averaging about three, and he set that next to Meta bonds running to 2049 against leases renewed every four years. A five-year loan against a three-year contract is a refinancing risk even when the customer pays. The backlog supports the debt only if the cash arrives before the loan matures. The same post cites a January Bank for International Settlements comparison: private loans to AI-related companies carried an average spread of 6.2 percentage points, against 6.1 points for other private credit. Lenders were not charging much extra for the risk the equity was celebrating.
The sleeves are not one yield. At 30 June, $13.6 billion was outstanding under delayed-draw term loans, secured by contributed contracts. That is not the unsecured 2032 note, and it is not the SOFR plus 5.50 percent loan or the 2.875 percent convert. Reuters posted at 10:40 GMT on 16 October: “Morning Bid: Calm after CoreWeave shock, but for how long?” The 7:04 a.m. Eastern premarket sale was $66.6004.
Bull, base, and bear through 31 December
The title numbers are a distribution around Thursday’s $69.69 official close. They are not the Friday premarket print, not Zelnick’s $150, not Truist’s $126, not the $142.78 average, and not the Redburn $54.
The last 90 daily log-returns use official closes from 8 June 2026, at $102.37, through 15 October. A log-return is the natural log of one close divided by the prior close. Mean mu is -0.004273. Sample sigma, dividing by n minus 1, is 0.060652, about 6.07 percent a day and about 96 percent scaled by the square root of 252.
N is 53, Friday 16 October through Thursday 31 December 2026. Weekends are out. Thanksgiving, 26 November, and Christmas, 25 December, are out. The 27 November and 24 December early closes still count. This close is Thursday’s.
Base is the spot times e to the power of N times mu. Bull adds 1.5 times sigma times the square root of N inside the exponent. Bear subtracts 1.0 times sigma times that square root. N times mu is about -0.2265. The square root of 53 is about 7.2801, so one standard deviation over the horizon is about 0.4416 in log terms. On the unrounded inputs the base is $55.57, the bull is $107.76, and the bear is $35.73: plus 54.63 percent, minus 20.26 percent, and minus 48.73 percent versus $69.69.
The base is under the spot because mu is negative. The bull still clears the spot, so no drift override was needed. The bear uses a smaller one-sigma shock on purpose. The window opens at the $102.37 close on 8 June and includes the 29 July close of $60.82.
| Case | Price | Versus $69.69 | What has to be true |
|---|---|---|---|
| Bull | $107.76 | Plus 54.63 percent | Negative drift continues, plus a 1.5 standard-deviation gain over 53 sessions. New money looks more like the convert coupon than like a 16 percent unsecured yield. Zelnick’s $150 is not required. |
| Base | $55.57 | Minus 20.26 percent | The last 90 sessions’ mean log-return persists, with no extra shock. The backlog can stay large. This case compounds the tape, not the contract list. |
| Bear | $35.73 | Minus 48.73 percent | The same drift, plus one standard deviation down, with the refinancing gap Zelnick sized near $35 billion still open. |
The $107.76 bull is $42.24 under Zelnick’s $150 and underwrites only this stock’s own volatility. The $35.73 bear is the formula, not an older $36 bear published next to a $303 bull.
What happens next
Friday 16 October is the first test, and it is the regular session, not the 7:04 a.m. Eastern premarket sale of $66.6004. A cash close under $69.69 would mean Thursday was a one-session response to Huang and Deutsche Bank. A close above Thursday’s $69.76 high would mean the equity is challenging the credit mark.
Monday 9 November 2026 is the ordinary 40-day mark after 30 September for a large accelerated filer’s 10-Q. CoreWeave has not confirmed that date, so this is a window, not a company announcement. If a report arrives, the test is whether backlog is still near $104 billion after a quarter of recognized revenue, and whether interest is still climbing from $640 million. A larger backlog with the bond under 80 cents means the two markets still disagree.
Thursday 31 December 2026 is the horizon. If mu and sigma persist, the base is $55.57, the bull at $107.76 needs the upside shock, and the bear at $35.73 needs one standard deviation down. None of the three is a bank target.
Frequently asked questions
Does Thursday’s CoreWeave print sit with the backlog or the debt?
It sits with the debt. The $69.69 close is 15.10 percent under the 9 October close of $82.08, and the last 90 sessions pull a 31 December base case to $55.57. A backlog of about $104 billion is contracted work still to deliver, not cash that retired Wednesday’s bonds. Reuters reported those 2032 notes below 80 cents on the dollar. The bull at $107.76 is a volatility outcome, not a verdict that the backlog won.
What are the CoreWeave bull, base, and bear by 31 December 2026?
The bull is $107.76, the base is $55.57, and the bear is $35.73. They use Thursday’s $69.69 official close, 53 New York sessions, a mean log-return of -0.004273, and a sample standard deviation of 0.060652. The base is below the spot because that mean is negative. They are not Deutsche Bank’s $150 target and not the $142.78 average TipRanks showed. This is not a recommendation.
Why did the shares fall if second-quarter revenue was $2,575 million?
Revenue grew from $1,212 million a year earlier, and the company still lost $626 million. Net interest was $640 million, five times adjusted operating income of $128 million. The operating loss was only $49 million, so the quarter was not a demand collapse. Wednesday’s 15.46 percent drop to $65.78 arrived with the 2032 bonds at distressed prices, and there was no new earnings release that day.
When is the next CoreWeave earnings report?
CoreWeave’s investor events page listed no date when checked on 16 October 2026. Third-party calendars do not agree, so this article does not treat 9, 11, or 16 November as a company announcement. The second-quarter release was 11 August. Whenever the date is set, the test is the new backlog against the bond price and the interest line, not the adjective in the headline.
Does Nvidia have to bail CoreWeave out?
TipRanks, citing Bloomberg, reported on 15 October that Jensen Huang said CoreWeave was becoming financially stronger and did not need Nvidia to rescue it, and that its “cost of capital is actually decreasing.” Nvidia has invested and has agreed to buy unused capacity. A purchase agreement is not a bond guarantee. The same TipRanks piece said credit prices still implied a better than 60 percent chance of default inside five years.
Is $40 million per megawatt the price of the whole fleet?
No. On 17 September the company said short contracts of about three to six months were pricing at about $40 million per megawatt. Annualizing second-quarter revenue of $2,575 million and dividing by 1.5 gigawatts of active power gives about $6.87 million per megawatt per year. The scarcity price and the fleet rate describe different slices of the same firm. One is a short contract. The other is power already on.
This is not financial advice. The closes used here are Thursday 15 October 2026 official Nasdaq prints. The Friday figure is a premarket last sale with a clock, not a forecast.