The AI Debt Bubble Is Here, and Wall Street Is Already Hedging Its Bets

The logic isn't crazy on its face. These are among the most profitable companies in human history. Their balance sheets are strong. The debt is investment grade. What could go wrong?

Call me old-fashioned, but I remember when people said similar things about mortgage-backed securities. The underlying assets were houses! Everyone needs houses! The math checked out right up until it didn't.

Now I'm not saying this is 2008 all over again, the scale and structure are different, and these companies actually make money (unlike a lot of the dot-com era players). But the pattern of banks hedging their bets while simultaneously facilitating ever-larger debt issuances, while hedge funds pile into the "easy" side of the trade... that pattern I recognize.

Winners and losers

The AI race is still in its "everyone's a winner" phase. Microsoft is winning. Google is winning. Amazon is winning. Meta is winning. Nvidia is definitely winning. The startups they're funding are winning. It's winners all the way down.

But that's not how technology cycles actually work, is it? At some point, the market figures out who's got the goods and who's been running on hype and borrowed time. When that sorting happens, and it always does, some of this debt is going to look a lot less safe than it does today.

Bloomberg's video analysis puts it pretty plainly: if the AI race starts producing clear winners and losers, the quiet CDS trade becomes much riskier. The hedge funds selling protection will suddenly be on the hook. The banks will be fine (they bought their insurance), but the ripple effects through credit markets could get interesting.

And "interesting" is not a word you want applied to credit markets.

What we don't know

I should be clear about the limits of what we're actually seeing here. The exact volumes of CDS trading tied to hyperscaler debt aren't public, so we're working from reporting and trader conversations rather than hard data. It's also too early to say whether this derivatives activity is prudent risk management or the early stages of something more concerning.

What we do know is that the debt issuance is real and accelerating. The infrastructure spending is real. And the hedging activity is real enough that Bloomberg is writing about it, which usually means it's been going on for a while already.

The kids building AI startups right now probably aren't thinking much about credit derivatives. Why would they? They're focused on the technology, the products, the race to AGI or whatever we're calling the finish line this week. But the money people are thinking about it. The money people are always thinking about it.

And when the money people start buying insurance against their own bets, that's worth paying attention to. It doesn't mean disaster is coming. It means the people closest to the money aren't as certain as the hype suggests. It means there's enough doubt in the system that protection has value.

I've been covering tech long enough to know that most bubbles don't pop dramatically. They deflate slowly, painfully, while everyone insists it's just a correction. The interesting question isn't whether AI infrastructure spending will pay off (some of it will, some won't), it's what happens to the financial structures built around the assumption that it all pays off.

But what do I know. If you want to argue, my email's on the about page.

Aisha Patel · 25 Jun · 8 min