The cost of insuring Nvidia’s debt against default just spiked the most on record. Not because anything is wrong with Nvidia, but because everything is very, very expensive in AI land right now.

Nvidia’s 5-year credit default swap spreads widened to 57.25 basis points, up from roughly 42 basis points in late June. That’s a meaningful jump for a company sitting on approximately $50 billion in cash. The trigger: reports that hyperscalers are gearing up to spend over $750 billion on AI data center infrastructure in 2026, with that figure potentially climbing toward $870 billion in 2027.

The great AI borrowing spree

That kind of capital expenditure is expected to exceed the free cash flow of even the largest hyperscalers. Which means one thing: debt. Lots of it.

Nvidia has announced plans to issue between $20 billion and $25 billion in high-grade bonds around June 15, 2026. It would be the company’s first bond issuance in five years. And the market’s response suggests investors are more than willing to play along. Demand for the offering reportedly reached $85 billion, roughly a 4x oversubscription.

The widening CDS spreads, then, aren’t really about Nvidia’s balance sheet. They reflect a broader market repricing of credit risk across the entire AI supply chain as the sector collectively reaches for its corporate credit cards.

Why the credit market is getting nervous

Credit default swaps are essentially insurance policies on corporate debt. When spreads widen, it means the market perceives higher risk, or at least wants more compensation for holding that risk.

The precedent is already being set elsewhere in the AI infrastructure stack. CoreWeave, the GPU cloud provider, raised $4.59 billion from a junk bond sale tied to Nvidia chips earlier this cycle. That deal highlighted both the appetite for AI-linked debt and the willingness of investors to accept lower credit quality in exchange for exposure to the theme.

What this means for investors

The 4x oversubscription on Nvidia’s planned bond offering tells an important story. Despite the widening spreads, the market’s appetite for high-grade corporate debt linked to AI remains voracious. Investors are essentially saying: yes, we acknowledge the risk has increased, but we still want in.

The broader signal here is that AI infrastructure spending has reached a scale where it’s becoming a macroeconomic variable, not just a sector theme. When four or five companies collectively plan to spend nearly a trillion dollars in a single year, it affects bond yields, credit spreads, and capital allocation across every asset class.