Why bonds won’t protect you from an AI bubble
As more debt is issued to fund AI build-outs, investors need to consider whether they really are diversified.
With the U.S. stock market dominated by companies in the artificial-intelligence business, some investors might want to move a few eggs to another basket. Say, to the bond market.
So what happens if corporate bonds start to become AI-dominated too?
Big tech companies are streaming into debt markets to fund gigantic investments in data centers, chips and other AI-related infrastructure. Other kinds of companies that borrow a lot, such as utilities, are also having to invest to feed the chips’ various energy needs.
As this debt becomes a greater part of bond markets—and funds that track them—investors seeking a broad mix of market exposure will have to make sure they are really diversified.
Within the market for highly rated, so-called investment-grade bonds, the current group of issuers that JPMorgan Chase credit research strategists consider “most closely tied to the AI revolution" represents 14.5% of a JPMorgan index of U.S.-dollar investment-grade corporate bonds.
That is up 3 percentage points from 2020. And the share could be more than 20% by 2030, if a scenario like current capital-expenditure forecasts plays out, JPMorgan’s analysts figure. If considered as a “sector," this group is already bigger than U.S. banks in JPMorgan’s bond index.
To be clear, stocks are way more concentrated. A basket of AI, data center and electrification-related companies in the stock market tracked by JPMorgan represented just under 40% of the S&P 500 as of late November.
And bonds can provide a steady yield that, barring a default by the issuer, can keep paying even if the same company’s stock is tumbling. Plus, the prices of bonds can rise when stocks are struggling in a bad economy, because the Federal Reserve is likely also lowering interest rates. When that happens, bonds’ fixed-rate payouts tend to become more valuable.
But when it comes to AI-related risks, there might be times when bonds and stocks move together.
For one, it’s hard to say yet which AI companies will be the winners that can justify their investment with enormous returns. Those who lose out might be at risk of downgrades to their credit ratings if their future cash flow doesn’t keep up with debt they have issued. Downgrades typically lead to lower prices for a bond.
Past investment booms have also had broad effects on companies issuing debt in those cycles.
In the 2010s shale-oil boom and the 1990s dot-com boom, issuance in related sectors grew by 51% and 312%, respectively, over three-year periods, according to credit strategists at Barclays. And as issuance of bonds ramped up, prices dropped as investors demanded relatively more yield versus the broader bond universe, JPMorgan analysts found.
“You tend not to want to own the things that are the biggest issuers in corporate credit," says Ben Inker, co-head of asset allocation at investment manager GMO. He added: “If I owned a lot of U.S. equities and have a large AI weight, I would be careful what I own on the corporate debt side."
This might have already started to happen with some AI-related bonds. Over the past year, the difference, or spread, between yields on the bonds of so-called hyperscaler companies and the broader investment-grade index has widened by about 0.3 percentage points from its recent average, Barclays strategists said in a recent note. They said they “still see room for further underperformance."
Plus, because high-grade corporate bonds are among the most liquid and heavily traded credit instruments, investors might look to sell them first in a time of stress or concern, Barclays analysts observed in a separate note.
Diversification does come in many forms. Some of the AI issuers are offering long-term debts such as a bond that doesn’t mature until the 2040s. For an investor such as an insurance company, with long-term obligations, that helps with the so-called duration of their portfolios. Retirees might see it the same way.
However, some large investors can only have so much exposure to a particular company or sector, says Dominique Toublan, head of U.S. credit strategy at Barclays. “When the concentration gets very high, there is a lot more scrutiny on that specific name or sector," he says. That itself could prompt some selling, as investors seek to rebalance.
In general, investors might need to move past pure index investing in corporate bonds. Because even within AI, there might be different profiles.
As JPMorgan U.S. high-grade credit strategist Nathaniel Rosenbaum observed in a recent note, among the high-grade AI-related bond cohort, some big tech companies still have relatively little debt compared with their massive cash flow. But some utilities and industrials are relatively more leveraged. That could lead some investors to shift within AI toward those big-tech issuers.
AI might reshape how people invest. But it might happen in unforeseen ways.
Write to Telis Demos at Telis.Demos@wsj.com
topics
