Nvidia Didn't Need the Money. It Borrowed $25 Billion Anyway

Nvidia just raised $25 billion in its first bond sale since 2021. The catch is that Nvidia didn't need the money. The company generated $50 billion in operating cash last quarter, holds $13 billion on the balance sheet, and authorized $80 billion in buybacks. So why borrow?

The order book is the story. Demand reached $85 billion, more than three times the final deal size. Nvidia started targeting $20 billion and raised the offering to $25 billion before pricing. The longest-dated tranche came in at just 65 basis points over Treasuries after tightening 25 points from initial guidance. Investors weren't accepting Nvidia's credit, they were competing for it.

This episode breaks down what that means. The deal is five times the size of Nvidia's 2021 bond sale and over twelve times the 2016 offering. It's split across seven tranches with maturities from 2 to 30 years, which lets Nvidia lock in long-term financing at near-historic low credit spreads. The US-Iran agreement has pulled investment-grade risk premiums back to pre-conflict levels, and high-grade bond funds have logged 13 straight months of inflows.

The broader pattern matters more than the single deal. Alphabet, Amazon, and other AI hyperscalers have been raising similar bond debt to fund data center buildouts. Nvidia joining sets a new credit benchmark for the sector and gives bond investors a way to position around the AI capex cycle without buying equity. For a company with a $5.15 trillion market cap and over $200 billion in projected free cash flow this fiscal year, this isn't a liquidity move. It's a market signal.

We cover what the proceeds are actually for (refinancing, general corporate purposes, and the buyback program), why bond investors wanted more than Nvidia was willing to sell, what a 3x oversubscription tells us about confidence in the AI hardware cycle, and whether this is the top of the cycle or the middle.

Nvidia bond sale, NVDA bonds, AI infrastructure, AI capex, investment grade bonds, AI hyperscalers, Nvidia stock, AI bubble, data center spending, credit markets.