AI Ate the Stock Market. Now It's Coming for Your Bonds.

The trade that ate everything

Here's the number that should make you sit up: nine of the ten best-performing ETFs in the first half of 2026 are tied to AI and chips. Most ride the semiconductor wave. The Invesco Semiconductors fund is up 138%. A broader AI "supercycle" fund is up 124%. And one leveraged single-stock chip bet jumped a barely-believable 959%—though those geared funds are wildly volatile and not built to hold, so treat that number as a curiosity, not a strategy.

The driver, per Morningstar, is "relentless demand for semiconductors, which are in limited supply and highly complicated to manufacture." In plain English: everyone wants the shovels for the AI gold rush, and there aren't enough to go around. The chipmakers—not the app builders—have been the real winners so far, which is exactly how the actual gold rush played out.

So AI has effectively taken over the stock market. That's old news. The interesting question is where the money flows next.

Buying an AI stock is a bet on the upside. Buying its debt is a bet you'll get paid back—and that's the harder call.

Next stop: the debt pile

The argument here is blunt: AI took over the stock market, and the bond market is next. All those data centers, all those chips, all that power infrastructure—someone has to pay for it up front. Increasingly, that someone is borrowing, either through corporate bonds or through private credit (loans made by investment funds instead of banks, with far less public disclosure).

That matters because buying an AI stock and lending to an AI project are two very different bets. Buy the stock, and you're betting the upside is huge. Buy the debt, and you're betting on something narrower and less glamorous: that you'll get paid back. And judging the credit risk of the AI boom is genuinely difficult—arguably harder than pricing the equities.

Why harder? Because the equity story is easy to tell: demand is booming, earnings are growing. But debt is a promise dated years into the future, and the AI buildout is a staggering amount of spending against payoffs nobody can confidently size yet. A data center financed today has to throw off enough cash to service its loans through whatever the AI economy actually looks like in five years. That's a lot of faith baked into a coupon payment.

Why this is the fault line to watch

Stocks can fall 40% and life goes on—shareholders eat the loss, the company survives. Debt is less forgiving. Miss the payments and you get defaults, forced sales, and knock-on effects that hit lenders who thought they were nowhere near the AI trade. That's how a hot theme quietly turns into a bigger problem: not through the flashy equity crash, but through the boring debt nobody was scrutinizing.

And the borrowing isn't about to slow. CFRA expects tech earnings growth to top everything but energy this year and run into next, and cooling energy prices are easing rate-hike fears—both tailwinds that encourage more building and more borrowing. Good for momentum. It also means the AI-linked debt stack keeps getting taller while everyone's still cheering the stock returns.

So what do you do with this? If you own bond funds or private-credit exposure, ask what's inside—how much is riding on hyperscalers (the giant cloud players like Amazon, Microsoft and Google) and data centers. If you're reaching into fixed income for AI yield, know you're making a repayment bet on an unproven cash-flow story, not a growth bet. The tell to watch is credit spreads on AI-heavy borrowers—the extra yield lenders demand over safe government debt. When that gap widens, lenders are telling you the risk is finally getting priced. That's when the party gets quieter.