Most people think of the bond half of their portfolio as the boring half. It is supposed to be the part with no story in it.
This month it has one. The companies building artificial intelligence are borrowing on a scale the bond market is having trouble swallowing. Over the past several weeks, the investment-grade corporate bond market struggled to absorb a combined $75 billion of new bonds from just three borrowers: Nvidia (NVDA), SpaceX and Amazon.com (AMZN). The Journal summed up Wall Street’s message to them as: please slow down.
It Isn’t Doubt About Getting Paid Back
This is not a credit scare, and it is worth being precise about that. For the most part, the paper reports, investors say they aren’t particularly concerned about the creditworthiness of the borrowers, or the sustainability of the AI infrastructure build-out. Their worry is simpler and more mechanical: they expect hundreds of billions of dollars of new bonds to keep coming, and nobody wants to be full when the next deal lands. “Everyone knows there’s a lot more coming, and so I think there’s been a hesitancy to jump in with both feet here,” said Travis King, head of investment-grade corporates at Voya Investment Management. “Everyone wants to leave some room for the next deal.”
You can see the flinch in the prices. Many investors were especially caught off guard by Nvidia’s $25 billion bond sale in June, and again last week by Amazon’s issuance of the same size. Alphabet’s (GOOGL) 10-year spread rose 0.12 percentage point last week and Meta’s (META) climbed 0.16, while the average investment-grade bond barely moved. SpaceX, a first-time borrower nobody quite knows how to price, has seen the spread on its 10-year bonds leap nearly half a percentage point since it issued them on June 23. The strain is landing on these bonds specifically, not on the bond market as a whole.
What This Has To Do With Your Bond Fund
Here is the honest version. This reporting is about professional managers, and it does not tell you what any particular fund owns. What it does say is narrower and still worth knowing: tech company bonds now make up an increasingly large share of benchmark bond indexes — the yardsticks most bond funds are measured against. The closer a fund hugs its yardstick, the more of this it will tend to hold. And the further it strays, the bigger the bet. One manager has strayed on purpose: John Lloyd, global head of multi-sector credit at Janus Henderson, has held fewer hyperscaler bonds than benchmark indexes because his team long believed the companies would outspend the consensus estimate. Going forward, he said, “you still have a wide range of outcomes” for how much they will invest in AI — with high-end estimates north of $10 trillion over the next several years.
So the general principle is an old one wearing new clothes. Concentration is not only a stock-market problem, and the safe-sounding half of a portfolio still has holdings in it that somebody chose. Nothing here says a bond fund is broken or that anyone should be doing something today. It says the boring half now has a live question attached to it, and the answer is knowable: what does your bond fund actually hold, and what is it trying to match? That is a fine thing to find out on a calm week rather than a loud one. Bring your statement to your next review and we will look at it together.
Clients believe they own AI on the stock side and safety on the bond side. That deserves a check. Tech bonds are a growing share of the benchmark indexes most core bond funds are measured against, and last week’s supply wave widened hyperscaler spreads while the broad investment-grade market moved 0.02 of a point. Pull the top holdings and the stated benchmark for every core bond fund in the book. Two things to see: how much hyperscaler paper is in there, and whether the manager is tracking the index or deliberately leaning against it — because per the Journal, some are underweight on purpose. This is the same concentration conversation we already have about the S&P 500, just on the half of the statement nobody reads. Do not tell anyone to sell anything; the point is that people should know what they own before the market makes it interesting.
NVDA · AMZN · GOOGL · META · ORCL
Mentioned for context only. A mention is not a recommendation — see the portfolios page for what the books actually hold.