From chip fever to cash hunger
For the last couple of years, the AI trade was pretty simple: buy the companies selling the shovels. Nvidia became the poster child because everyone needed GPUs, data centers, and the whole electrical-grid-meets-sci-fi setup.
Now the vibe is changing. The question isn’t just who builds the AI infrastructure — it’s who pays for it.
Debt is joining the party
Nvidia just raised $25 billion with a bond offering, its first major debt deal in years, and the demand was reportedly a jaw-dropping $85 billion. That’s not the behavior of a company desperate for cash; that’s a signal that even the AI kingpin sees a world where funding this buildout gets expensive fast.
JPMorgan’s back-of-the-envelope math is even bigger: about $5.5 trillion in AI infrastructure spending through 2030, with roughly $4.1 trillion potentially financed through debt. Translation: the AI arms race may be less “who has the fastest chips?” and more “who can keep borrowing without face-planting?”
Who could cash in?
If AI turns into a financing story, the winners could widen beyond semis and cloud names. Think:
- big banks like JPMorgan, Goldman Sachs, and Morgan Stanley, which can underwrite and arrange the deals
- private-capital firms like Apollo, Blackstone, and Blue Owl, which love a chunky, asset-backed financing opportunity
Oracle’s reported plan to raise around $40 billion for its AI buildout and Meta’s financing explorations are more evidence that the tab is getting large enough to require more than corporate checkbooks.
Big picture
This doesn’t mean Nvidia is suddenly yesterday’s trade. It means the AI boom may be entering a more mature, more leveraged phase — and that could pull a very different set of stocks into the spotlight. If the first act was about chips, the sequel might be about credit.