Who is steering CoreWeave through the selloff, and their record so far

Boardroom Alpha · Executive Brief

On October 14, the day shares of CoreWeave (CRWV) fell 15.5%, chief executive Michael Intrator told employees in a memo: “When you are building a new type of technology company there will be periods of time when the market misunderstands and acts irrationally.” He added: “We are confident in our backlog which more than supports all of our debt obligations.” By October 16 the stock was down 30% from the end of September, at $61.39. Its bonds fell too: yields on its unsecured notes rose past 15%. Last week, insuring its debt against default cost more than 10% of the amount a year, a level that Gavan Nolan of S&P Global Market Intelligence told CNBC is generally a sign of credit distress.

Three co-founders, all former commodity traders, control CoreWeave through shares with 10 votes each: more than 70% of the votes on about a fifth of the stock. They have delivered the growth they forecast, after one cut in 2025. Their margins and borrowing costs have fared worse. Since the March 2025 IPO they have sold $3.3 billion of stock, mostly through trading plans set up in advance, and bought none. Three outside directors who joined just before the IPO oversee them.

Nebius (NBIS), a rival, fell just as far, in a month when a higher-than-expected inflation report raised bets on Federal Reserve rate increases and Anthropic reported slower revenue growth. But Nebius has far less debt, and CoreWeave has more than $35 billion. CoreWeave’s stock is still up 53% from its $40 IPO price, about level with the Nasdaq-100, but down 67% from its June 2025 peak.

Intrator delivered the growth he forecast

Intrator spent about 16 years at Natsource, an emissions-credit trader, and from 2013 to 2018 ran Hudson Ridge, a natural-gas hedge fund, with Venturo. In 2017 the founders began mining Ethereum, then turned the chips to AI computing. The mining business closed in 2022.

Under Intrator, quarterly revenue more than doubled in a year, to $2.58 billion in the second quarter of 2026. CoreWeave has met or beaten its revenue guidance every quarter this year and raised its 2026 forecast to $12.4 billion to $13.2 billion. The backlog of signed contracts grew from $25.9 billion in March 2025 to $104.2 billion.

In November 2025 CoreWeave lowered its 2025 forecasts, citing a third-party data-center developer that was behind schedule. Revenue landed in the lowered range. Equipment spending, at $10.3 billion, fell below even the lowered $12 billion to $14 billion; the August forecast had been $20 billion to $23 billion. A shareholder filed a proposed class action in January, generally alleging false and misleading statements, and their lawyers tie it to the delays. CoreWeave says the claims are “without merit” and that it “intends to defend itself vigorously.”

The IPO priced at $40, below its $47 to $55 range. In October 2025 Core Scientific (CORZ) shareholders rejected CoreWeave’s all-stock takeover.

Margins are thinner than a year ago, and the backlog has yet to be earned

The adjusted operating margin was 5% in the second quarter, up from 1% in the first but down from 16% a year earlier. Intrator said on August 11 that “our scale began to translate into expanding operating leverage.” Net interest was five times adjusted operating income. On CoreWeave’s adjusted measure, profit before interest, taxes, depreciation and amortization was $1.51 billion, 2.4 times interest. That measure leaves out depreciation, the cost of its chips spread over six years. Under standard accounting CoreWeave lost $626 million in the quarter.

The backlog in Intrator’s memo is revenue CoreWeave has yet to earn. About $43 billion of it is expected within two years, much of it from capacity still to be built. At June 30 CoreWeave had $15.6 billion of cash and undrawn credit, against $10.6 billion of debt due by the end of 2027 and interest running at about $2.6 billion a year. In the first half it brought in $3.7 billion of cash from operations and spent $14.1 billion on equipment, so it will have to keep raising money.

Its three largest customers accounted for 72% of second-quarter revenue. CoreWeave’s 10-Q warns of “the risks of non-payment and non-performance” by its few large customers, and some of its loans let lenders hold back cash if a customer contract goes wrong. S&P rates CoreWeave B+, below investment grade.

CoreWeave’s debt has quadrupled, and unsecured borrowing costs more

Co-founder Brannin McBee has been CoreWeave’s public voice on its borrowing. Debt principal rose from $8.8 billion in March 2025 to $35.6 billion at June 30, 2026, on top of $16.3 billion of lease obligations. Since then CoreWeave has added a $2.6 billion delayed-draw loan facility and $4.2 billion of convertible notes, bonds that can be swapped for shares.

The IPO prospectus said CoreWeave had “managed to drive lower cost of capital across our financings.” That holds for loans backed by customer contracts and for convertibles. An $8.5 billion loan rated investment grade pays 2.25 percentage points over the benchmark lending rate on its floating-rate portion, and convertible notes sold since December 2025 pay 1.75% to 2.875%. Unsecured debt costs far more: the four dollar bond issues pay 9% to 9.75%. Even the newest contract-backed loan, in August, cost 5.5 percentage points over the benchmark.

During the selloff McBee told the Financial Times that insurers’ money for investment-grade debt is “unlimited” and called its chip-backed loans “accelerating debt and great debt.” Nvidia (NVDA) chief executive Jensen Huang said on October 15 that CoreWeave’s “cost of capital is actually decreasing.” Nvidia owns about 9% of CoreWeave, makes all of its GPUs and has a $6.3 billion agreement to buy capacity CoreWeave cannot sell through April 2032.

Since June 2025 two Magnetar veterans have run strategic financing: Ernie Rogers, Magnetar’s former chief operating officer and a CoreWeave director until March 2025, and Jessica Damrat, a former finance chief of its management company. Rogers’s 2025 award of 600,000 restricted stock units had a grant-date value of $77.9 million. CoreWeave says the pay of both was set by reference to market practice or to the pay of employees with no Magnetar ties. Magnetar and Blackstone (BX) had led a $2.3 billion chip-backed loan to CoreWeave in 2023, at about 14% interest.

If borrowing stays costly, more of the money may come from shareholders. A program set up in September lets CoreWeave sell up to 35 million new shares, about 6% of the total, into the market.

Agrawal’s spending forecast held, but control weaknesses from before the IPO are still open

Nitin Agrawal held finance roles at Google Cloud and Amazon Web Services and was finance chief of Mapbox, a private company, before joining CoreWeave. In November 2025 he cited “continued insatiable demand” and said 2026 spending would be “well in excess of double that of 2025.” That held: first-half spending alone exceeded all of 2025.

CoreWeave’s disclosure controls were “not effective” at June 30. Its material weaknesses, the most serious class of control gaps, are in IT controls, separation of duties and the size of its accounting staff, the same three it disclosed before the IPO. It says it is fixing them under audit committee oversight.

The founders have sold $3.3 billion of stock; no officer or director has bought since the IPO

Eleven executives and directors have sold $4.1 billion of stock since the post-IPO lock-up ended in August 2025, more than $110 million of it to cover taxes as stock awards vested. Measured against what each held at the IPO, Venturo has sold more than a third, McBee about 40% and Intrator about a sixth, Bloomberg calculated. The founders’ sales averaged about $97 a share.

Most of the selling ran through trading plans set up in advance, the usual route for executives. CoreWeave calls the sales “pre-scheduled, fully disclosed, and entirely standard,” and says the founders still own about 20% of the company.

Two of the biggest bursts came right after results. In the month after the lock-up expired on August 14, 2025, insiders sold $831 million, more than half of it by Jack Cogen, then a director. In the month after May 7, 2026 results, which Intrator called the “strongest bookings quarter in CoreWeave’s history,” they sold $483 million, including $214 million by Cogen with no trading plan disclosed.

The founders sold nothing in October through October 16. Agrawal sold $1.8 million on October 12 under a trading plan set up in August 2025 and changed in November. Magnetar has sold about $6 billion of stock since August 2025.

Hutchins, Boone and Whitman, all appointed just before the IPO, oversee the founders

Glenn Hutchins, the lead independent director, co-founded the private equity firm Silver Lake and was its co-chief executive until 2011. He is vice chair of Banco Santander (SAN) and sat on the board of the Federal Reserve Bank of New York. Nasdaq (NDAQ) returned 655% while he was a director from 2005 to 2017, against 170% for the S&P 500. AT&T (T) returned 93% over his 2014 to 2025 term, against 264%.

He and Boone are the only outside directors who have bought stock with their own money: about $20 million and $500,000, at about $47, before the IPO.

As pay committee chair, Hutchins oversees awards that vest over four years; apart from a condition met at the IPO, they carry no performance test. The committee called the executives’ 2025 performance “outstanding” and paid each a bonus of 100% of target. It paid them alike because it “believed that the executive team contributed equally to the Company’s achievements.”

Karen Boone heads the audit committee, which oversees the control fixes. Until 2012 she was an audit partner at Deloitte, now CoreWeave’s auditor. She then became finance chief and later president of RH (RH), whose stock rose 392% from its first day of trading in 2012 to her departure in 2018, against 125% for the S&P 500.

Boone has been a director of Peloton (PTON) since 2019. She was its chair from 2022 to 2024 and then its interim chief executive for eight months, at first jointly. The stock rose 170% during that stint, against 18% for the S&P 500. Over her whole time on the board, Peloton’s shares have fallen 80% from their first day of trading, against a 184% rise in the S&P 500. She also sits on the boards of Sonos (SONO) and Rivian (RIVN), both down since they listed while the S&P 500 rose.

Meg Whitman leads the nominating committee. She ran eBay (EBAY) as its stock rose more than 1,400% from its first day of trading in 1998 until she stepped down as chief executive in March 2008, against 47% for the S&P 500. Hewlett-Packard returned 23% under her from 2011 to its 2015 split, against 94% for the S&P 500. Hewlett Packard Enterprise returned 104% by early 2018, counting the two businesses it spun off, against 40% for the S&P 500. She then ran Quibi, a streaming start-up that raised $1.75 billion and announced its shutdown six months after launch. Since September 8 she has also been executive chair of Albertsons (ACI), an employee role in which she advises the chief executive on strategy and operations.

The board has had five members since Cogen, a director since 2017 and Intrator’s and Venturo’s former boss, left in June. The founders’ 10-vote shares convert to ordinary stock by May 2032 at the latest, or earlier if Intrator stops working at CoreWeave full time, though not if the board removes him without cause. Intrator was re-elected in June with 93% of the votes cast. CoreWeave does not report the vote by share class, but if the founders voted all their shares for him, about half of the other holders’ votes were withheld.

Our view

Intrator, Venturo and McBee have built the business they promised. Revenue has met or beaten guidance every quarter this year, the backlog has quadrupled since the IPO, and the board added three directors with serious résumés before the listing. But CoreWeave has financed the growth with costly debt, its margins are thinner than a year ago, and the founders have sold $3.3 billion of their own stock. Intrator is asking investors to look to the backlog.

With the November results, Agrawal should set out how CoreWeave will fund or refinance the debt due in 2027 and how far contracted cash from investment-grade customers covers the next three years’ debt payments. The board should hold off on selling shares under the September program until then. The founders should not adopt new selling plans until the credit market settles. Directors who believe the market has CoreWeave wrong should buy stock with their own money when the trading window opens.

On the board, the nominating committee under Whitman should add an independent director who has worked through a credit downturn from the lenders’ side. Hutchins’s pay committee should tie the founders’ future awards to cash flow, margins or borrowing costs rather than time served. And Boone’s audit committee should publish a date for fixing the control weaknesses, which have been open since before the IPO.

What to watch

• October
  Sales of new shares can begin under the September program.
• Oct. 27–28
  The Federal Reserve meets.
• November
  Third-quarter results and the plan for 2027 maturities. Monthly repayments begin on a $3.1 billion loan.
• Form 4s
  Whether the founders’ selling resumes, and whether any director buys.
• 2026–27
  $4.4 billion of debt due in the rest of 2026 and $6.2 billion in 2027, as of June 30.

Sources: CoreWeave’s 2026 proxy statement, IPO prospectus, 10-K, 10-Q and 8-K filings; insider trades from Form 4 filings to Oct. 16, 2026; the October 14 memo, McBee’s interview, Huang’s remarks, guidance and bond yields as reported by CNBC, Bloomberg, Reuters, Fortune and the Financial Times; S&P Global Market Intelligence via CNBC; credit rating from S&P Global Ratings. Adjusted operating income, as reported by CoreWeave, is after depreciation and excludes stock-based pay. Returns to Oct. 16, 2026. Boardroom Alpha is an independent research and analytics company and does not invest. This brief is for information only and is not investment advice.