CoreWeave Crashes 16% as “AI Bond Blues” Grip Wall Street; Traders Pile Into “Bubble Puts”

The artificial intelligence trade had its ugliest day in months on Wednesday, October 14, as a violent selloff in both the shares and the bonds of AI infrastructure companies exposed how fragile the financing behind the AI buildout has become. CoreWeave, the Nvidia-backed “neocloud” that rents out AI computing power to developers, saw its shares crash roughly 16 percent on the day, while yields on its unsecured bonds climbed to about 15 percent — deep into territory that bond traders consider distressed — prompting a flood of bearish options bets and a one-word verdict from Elon Musk: “yikes,” according to CNBC.

The damage spread well beyond one company. Nebius fell around 10 to 11 percent, IREN dropped more than 12 percent, Oracle sank about 9 percent, and Dell tumbled roughly 11 percent, together erasing some $80 billion in market value in a single session, according to Reuters. The PHLX Semiconductor Index slid 4.2 percent for a sixth straight day — its longest losing streak in nearly two years — while the Nasdaq dropped 1.9 percent, its biggest one-day fall in three months, and the S&P 500 lost 1.1 percent, its first decline of at least 1 percent in more than 50 trading days, according to Reuters.

The numbers behind Wednesday’s rout

CoreWeave’s shares closed around $65.78, down about 15.5 percent from the previous close of $77.81, with trading volume of roughly 107 million shares — nearly four times the average, according to market-data reporting. The stock is now set to wipe out about a quarter of its value this month, Reuters reported.

But equity was only half the story. Yields on CoreWeave’s $1.25 billion of unsecured bonds maturing in 2032 touched an intraday high of 15 percent before closing at approximately 14.5 percent, according to CNBC, while a separate 9.75 percent bond due in 2031 yielded 16 percent on Wednesday and slumped to a new low in price, according to Reuters’ Trading Day commentary. The 2032 bonds traded below 80 cents on the dollar, according to TRACE pricing data cited by TheStreet — a level widely considered the threshold at which a borrower is regarded as financially strained.

The stress infected the credit of far bigger names too. Meta’s “Beignet Investor” bond maturing in 2049 hovered near its lows, and Oracle’s 4.9 percent bond maturing in 2033 registered a record low, Reuters reported, with both bonds down around 20 percent from their issue prices. The cost of insuring the debt of Oracle and CoreWeave against default surged to record highs, according to S&P Global Market Intelligence data cited by Reuters.

How a tweet turned into a bond-market rout

The spark for Wednesday’s selloff arrived the night before. Venture capitalist Chamath Palihapitiya, founder of Social Capital, posted a chart on X on Tuesday evening showing CoreWeave’s bond yields had reached 15 percent, and used it to question whether a private-credit bubble was forming around the AI industry, according to CNBC. Elon Musk, whose companies include X, replied with a single word: “yikes.”

Palihapitiya’s analysis pointed at something specific: how little extra compensation lenders appeared to be demanding for AI risk, according to reporting on his post. The observation landed just as the bond market was already pricing in a rougher outlook for indebted AI firms.

A Financial Times report added fuel, according to CNBC: CoreWeave co-founder Brannin McBee said the company was targeting insurance companies for further debt financing, calling insurers’ capital “unlimited.” McBee told the FT that securing investment-grade ratings on more of CoreWeave’s chip-backed debt would let it borrow from insurers at lower cost, and said the company was “actively” looking at the $500 billion financing program Nvidia unveiled in August alongside partners including Goldman Sachs, under which insurers could underwrite part of the risk on chip-backed loans.

Then came the options market. Trading in CoreWeave options surged to more than seven times the 30-day average, with traders buying just under 200,000 puts against 106,000 calls, according to CNBC. One new position — 26,000 December 18 puts struck at $37.50, bought for $1.9 million — would need CoreWeave to fall another 42 percent over roughly two months to pay off. Scott Bauer, chief executive of Prosper Trading Academy, told CNBC the trade looked less like a bet that the stock would go to zero and more like a hedge on a highly correlated AI complex.

The company at the center of the storm

CoreWeave has become the emblem of a particular AI-era business model: borrow enormous sums, buy Nvidia chips, build data centers, and rent out the computing capacity — in industry jargon, a “neocloud.” The company has grown from three data center facilities to 51 in four years, according to SiliconANGLE, and its debt and lease obligations totaled roughly $52 billion at the end of June, while it has raised more than $35 billion in financing this year alone.

The scale of the borrowing is what frightens bond investors. CoreWeave has issued $22 billion of bonds and taken on $26 billion of loans, according to LSEG data cited by Reuters, and its quarterly interest costs more than doubled from a year earlier to $640 million in the second quarter. Visible Alpha consensus estimates cited by AIStockWire expect the company to burn about $25 billion in negative free cash flow this year, $28 billion next year, and $18 billion in 2028.

That is the bear case. The bull case is demand. CoreWeave reported a revenue backlog of $104 billion, second-quarter revenue of $2.6 billion — up 112 percent from a year earlier — and full-year 2026 sales guidance of at least $12.4 billion, according to reporting on the company. Its second quarter did include a $626 million net loss, however, which analysts have cited as evidence of how punishing the debt load is.

The company’s own leadership struck back. Chief executive Mike Intrator wrote to employees in an internal memo seen by Bloomberg, arguing that the market was acting “irrationally” and that the company’s contract backlog “more than supports” its debt obligations, according to reporting from TheFly. “When you are building a new type of technology company, there will be periods of time when the market misunderstands and acts irrationally,” Intrator wrote. Paul Meeks, head of technology research at Freedom Capital Markets, told CNBC he expected CoreWeave’s margins to stay flat between June and September before improving between September and December as new data centers came online, adding that the situation was “definitely not a deteriorating situation” — just one not improving yet.

The skeptics remain vocal. Gil Luria, an analyst at D.A. Davidson, warned that CoreWeave’s equity could theoretically be wiped out if AI spending weakens despite the backlog. Luis Alvarado, co-head of global fixed income strategy at the Wells Fargo Investment Institute, told CNBC that CoreWeave’s balance sheet looked “relatively bad” compared with other neoclouds. And the timing was unkind: Jim Cramer had called CoreWeave the only company of its kind he trusted less than a day before the crash, according to BeInCrypto’s market coverage.

The macro fuse: hot inflation and the rate-hike repricing

None of this happened in a vacuum. Wednesday’s session opened with a hotter-than-expected September inflation report: headline consumer prices rose 3.7 percent from a year earlier, up from 3.4 percent in August, while core inflation came in at 2.6 percent, above economists’ forecasts. The data revived bets that the Federal Reserve could raise interest rates again as soon as its October 27–28 meeting, with traders pricing roughly a one-in-three to one-in-two chance of a hike, according to Reuters and Bloomberg data.

Higher rates are poison for companies that borrow heavily to build data centers and buy AI chips. The two-year Treasury yield jumped as the inflation figures landed, the 10-year yield traded above 5.3 percent intraday before easing to about 5.19 percent, and the dollar rose to a fresh 18-month high, according to Reuters and the Wall Street Journal. For the most indebted AI names, the market concluded, the financing arithmetic was getting worse by the hour.

Analysts made the selloff worse with downgrades. Barclays cut Dell, Hewlett Packard Enterprise, and a third enterprise-hardware stock to “equal-weight” from “overweight,” arguing that valuations had run too far: “Valuations have run, commodity costs have helped revenues, and we don’t see a meaningful AI inflection yet,” the bank said, according to Reuters. Dell shares have quadrupled this year, and 16 stocks in the S&P 500 technology sector have at least doubled, according to Reuters’ calculations.

“Investors are questioning whether the AI buildout will generate the returns that were expected,” Kim Forrest, chief investment officer at Bokeh Capital Partners, told Reuters. “We’re getting a whole lot of information suggesting that the largest AI companies may not be generating the earnings or revenue people anticipated, and I think that makes everyone nervous about companies like CoreWeave.”

Wall Street’s biggest banks, which both finance the buildout and advise on it, sounded cautious on their own earnings calls this week. Goldman Sachs chief executive David Solomon told analysts that “no one knows how the demands for capital and compute will evolve as the broader AI ecosystem matures,” and stressed vigilance about “the types of activity we finance and how we manage risk,” according to Reuters. Citigroup chief executive Jane Fraser pushed back, saying she was “not seeing a slowdown in investment in the broader AI ecosystem” because “there’s simply not enough compute,” while Bank of America chief executive Brian Moynihan called AI capital investment important for his bank’s debt and leveraged-finance businesses.

Analysis: Why It Matters

Wednesday’s rout was not really a verdict on whether artificial intelligence works. It was a verdict — at least a preliminary one — on how it is being financed.

The most important distinction in this story is the one the market itself drew: between revenue and solvency. CoreWeave does not obviously have a demand problem. A $104 billion backlog and triple-digit revenue growth are, by any ordinary standard, evidence of a business customers desperately want. The problem is the capital structure underneath it — tens of billions in debt taken on before the customer revenue has arrived, at floating anxiety about rates, with interest bills doubling year over year. The market’s judgment on Wednesday was that the margin of safety in that structure is gone.

That points to the deeper question now circulating on trading floors: the circularity of AI financing. Nvidia backs CoreWeave, lends to customers, invests in the companies that buy its chips, and is now assembling a $500 billion financing program with banks and insurers to fund more chip purchases. Each loop strengthens the ecosystem’s cash flows — and each loop concentrates the risk. If the revenue ever fails to materialize, the counterparties discover they were exposed to each other. Palihapitiya’s question about the “private credit bubble” was really a question about whether this loop has been priced as if it cannot break.

Wednesday also showed how quickly macro can convert a worry into a fire. A single inflation print, by repricing October rate-hike odds, turned a slow-burn concern about AI debt into a stampede out of it. Debt-heavy companies are rate-sensitive by definition, and the AI buildout is the most debt-heavy investment cycle in living memory. The reflexivity is the danger: higher borrowing costs deepen investor concern, concern drives selling, selling pushes yields higher still. Reuters’ Trading Day commentary asked, bluntly, whether the spiral ends with “some sort of intervention.”

That possibility is no longer abstract. Reuters’ Mike Dolan has argued that governments nursing stressed bond markets could be forgiven for quietly welcoming an AI stumble, since the AI boom has been competing directly with governments for fixed-income dollars and stoking inflation that complicates central bankers’ jobs. The Cato Institute counts 35 US government equity-stake deals in private companies since June 2025, eleven of them in semiconductor firms and nine in quantum computing — unheard of outside crisis periods until recently. Washington’s posture toward the tech sector has shifted from regulator to co-investor; if the AI buildout ever truly wobbles, the pressure for the state to step in — on national-security grounds, or simply because key players look “too big to fail” — will be immediate.

For readers, the practical lesson is narrower: the AI trade is no longer one trade. For two years it moved as a bloc — chips, cloud, data centers, power — but Wednesday split it into at least two stories. One is about earnings power, which upcoming results will test. The other is about financing durability, which markets are testing right now, in real time, and where CoreWeave has become the canary.

What to Watch Next

1. CoreWeave’s third-quarter results, expected in November. Revenue growth is no longer the question; margins, cash burn, and backlog conversion are. Watch whether management can show the new data centers turning into profitable capacity, as Paul Meeks expects, and what guidance says about 2027 financing needs.
2. The Federal Reserve’s October 27–28 meeting — and Kevin Warsh’s remarks Thursday at the IMF meetings in Bangkok. The entire AI-debt selloff is a bet on the rate path; any signal that October or December hikes are off the table would ease pressure on indebted AI firms fast.
3. Thursday’s data double-header: TSMC’s third-quarter earnings and US September retail sales. TSMC is the foundry at the heart of the AI chip boom — its order book is the cleanest read on whether the hardware demand story is intact beneath the financing panic.
4. Who finances the next round. McBee’s hunt for insurer capital and Nvidia’s $500 billion financing program will reveal whether private money still trusts AI debt at lower prices — or whether the industry quietly shifts toward state or supplier financing.
5. Credit markets, not just stocks. The CDS record highs on Oracle and CoreWeave debt and the distressed pricing on CoreWeave’s bonds are the early-warning system now. If spreads stabilize, Wednesday looks like a scare; if they keep widening, the “bubble puts” buyers will look prescient.

Sources

• Reuters, “AI infrastructure firms slump as US rate hike bets rise”
• Reuters, “Red hot CPI report fuels AI rout”
• Reuters Trading Day commentary, “Tech melt”
• The Wall Street Journal, “Stock Market Today: Treasury Yields Turn Lower as Stocks Slump”
• Motley Fool, “Stock Market Today, Oct. 14: CoreWeave Plummets on Gartner Neocloud Warning and Bond Yield Surge”
• SiliconANGLE, “Shares of Oracle, neoclouds drop on CoreWeave debt concerns”