Oracle's Credit Insurance Hit a Record High. CoreWeave's Now Prices In a 63% Default Risk.

Oracle's five-year credit default swaps hit 288 basis points on Thursday. Seven days earlier they were at 260. Twenty-eight basis points is a small move in absolute terms, but Oracle holds an investment-grade rating, and as recently as 2024 almost nobody was pricing meaningful default risk into its debt at all.

[[IMG1]]

CoreWeave's version of the same story is far more dramatic. The AI cloud computing company's CDS spread surged past 1,000 basis points this week, a level that implies roughly a 63% probability of default over the next five years, according to credit market pricing. A year ago that same math put the odds closer to 30%. CoreWeave's 2032 notes now change hands around 78 cents on the dollar, yielding north of 16%, well above what its 8.5% to 9.75% senior bonds were priced to pay when issued. The stock fell about 15% in Wednesday's session, slipped another 6% after hours, and was down roughly 7% again in Thursday's premarket.

What Changed

Credit markets are a slower-moving cousin of the stock market. They tend to confirm a worry rather than announce one, and this week they confirmed something Pulse24 readers have been watching build for months: a growing gap between how much the AI buildout costs and how fast it's generating cash to pay for itself. Oracle's widening matters precisely because Oracle isn't CoreWeave. It carries real revenue, a long enterprise software history, and a credit rating that's supposed to put it in a different risk category entirely. Its CDS moving at all, let alone to a record, tells you the caution is no longer confined to the most leveraged names in the AI trade.

SoftBank's spread rose 24 basis points to 424, and even Amazon, Alphabet, Meta, and Nvidia, companies with fortress balance sheets by almost any measure, saw their own default insurance costs climb to records this week. The moves there are tiny next to Oracle's or CoreWeave's. But the direction is the same, and direction is what credit investors watch first.

Why It Matters

This didn't happen in a vacuum. September's hotter-than-expected inflation print pushed up the market's expectations for how much longer the Fed keeps raising rates, and higher-for-longer policy rates flow straight through to corporate borrowing costs. For a normal company that's an inconvenience. For an industry that's financing a buildout measured in the hundreds of billions of dollars largely with debt, it's closer to a structural problem.

CoreWeave is the clearest example of why. The company's own executives have described a financing model built around special-purpose vehicles, where lenders look through to the credit of CoreWeave's customers rather than CoreWeave itself, letting it borrow at rates closer to a company with an A-minus rating than its own balance sheet would justify. That only works as long as those customer contracts hold. When OpenAI's own reported revenue came in well below what some backers had been told earlier this month, it raised exactly the question credit markets are now pricing: what happens to the lenders standing behind CoreWeave's roughly $50 billion in debt if the AI revenue backing it grows more slowly than the spending it's funding.

Zoom out and the scale gets bigger fast. Hyperscalers have issued roughly $220 billion in bonds so far this year, and Wall Street desks have pushed their 2026 forecasts for tech-sector debt issuance as high as $540 billion, putting that borrowing in direct competition with the US Treasury for the same pool of buyers. Spreads on that debt have already widened across every maturity bucket this year. Thursday's CDS move is the sharpest signal yet that the market is starting to ask for compensation, not just accept the issuance.

What to Watch Next

The number to track isn't Oracle's spread in isolation. It's whether Oracle keeps widening alongside CoreWeave, or whether this stays contained to the handful of pure-play AI infrastructure names with the thinnest cushions. Nebius has no public CDS market, but its stock fell more than 10% on Wednesday and kept falling Thursday morning, leaving it almost 40% below its 2026 high, a sign the selloff is already spreading well past the names whose CDS spreads are easiest to track. Earnings season will matter more than usual this quarter. Anthropic reported third-quarter revenue of $16.8 billion, up 45% from the prior quarter, and investors barely blinked. That's the tell. Strong AI revenue numbers are no longer enough on their own to calm credit markets that are now asking a harder question: not whether AI revenue is growing, but whether it's growing fast enough to service the debt piling up behind it.

The Pulse24 Take

Stock markets and credit markets are having two different conversations about AI right now, and that gap is worth paying attention to. Equity investors have mostly stayed focused on growth. Credit investors get paid to focus on survival, and this week they started pricing in meaningfully more of it. Oracle isn't CoreWeave, and a 288 basis point spread is nowhere near distressed territory. But the fact that caution has spread from the most leveraged AI infrastructure names to a company with Oracle's balance sheet is the kind of signal that tends to show up in credit markets before it shows up anywhere else. Nothing here says the AI buildout stops. It says the market is starting to price the possibility that some of the companies funding it with debt won't all make it to the other side on the terms they borrowed under. That's a very different question than whether AI itself works, and it's the one worth watching through the rest of this earnings season.