The artificial intelligence and data center buildout boom is one of the most important drivers of the U.S. economy in 2026. It is also essential for investors. Technology companies’ massive debt issuance to fuel the boom has had a sharp, swift impact on corporate credit markets.
According to Vanguard research, from 2020 through 2024, the five biggest technology companies behind the boom—Alphabet, Amazon, Meta, Microsoft and Oracle—together issued roughly $35 billion per year in debt.
Last year, these “hyperscalers” issued $93 billion, and that figure has already ballooned to $132 billion through July 2026. Estimates for the full year’s AI-related debt issuance—including chip makers, utilities and others in the wider AI-adjacent theme—range from $300 billion to $570 billion. This wave of net new issuance is also longer-dated, with 20- and 30-year bond sales common.
The supply flood has pushed up yields and heightened concerns that credit quality is falling. In mid-September, Torsten Slok, chief economist at Apollo, pointed to widening credit-default swap spreads for hyperscaler bonds as a sign. Swap spreads for the biggest issuer by industry, banks, remain steady, so Slok suggested in a blog post that wider spreads for AI-related debt are a sign that “what the market is repricing is hyperscaler credit fundamentals.”
Bond investors are keeping a wary eye on potential troubles, yet they note there are opportunities in the AI boom for investors that can dig into the fundamentals.
Tale of Two Markets
John Queen, a fixed-income portfolio manager at Capital Group, says hyperscaler spreads are wider than U.S. Treasurys, but most other investment-grade corporate spreads remain historically tight, reflecting a strong economy.
“What it suggests is you have really a tale of two markets,” he says.
As an example, Queen says he has seen some issuance from hyperscaler-type companies with AA or A credit ratings trade at BBB or BB levels. Given how much supply is expected to hit the market, he expects these spreads to stay wide for a while.
“You have to come to grips with the idea of: If they view AI as this 30%, 40%, 50% [return on investment] compounding investment, do they really care about 50 or 100 basis points on their spreads?” Queen asks. “Maybe not. If that’s the case, even if you think they’re a good deal now, they’re going to be a better deal down the road.”
Matt Brill, head of investment grade credit at Invesco, admits it is overwhelming for buyers: “I kind of describe it as being too much of a good thing,” he says, which can still cause yields to go higher.
There are good opportunities for institutional investors, but they still must be careful, Brill says, and may opt to wait for better future buying opportunities if they believe the AI buildout will go on for a few more years. Investors only have so much money to spend, and they may have portfolio concentration limits as well.
Ally Betancourt, head of U.S. credit strategy at J.P. Morgan Private Bank, wrote in a research note that the concerns about credit quality deterioration may be overblown. Much of the debt comes from investment-grade companies with low leverage and rising earnings that are well-positioned to support that debt.
Brill says the AI issuance also supports sectors such as banks, utilities and industrials.
“A lot of industries are benefiting from it,” he says. “If you take that benefit away, what will their … credit profile be like at that point? We don’t really know. It’s something to be aware of,”
Being choosy matters, even with plenty of generally high-quality supply. Queen says he avoids companies with a single source of revenue, as the lack of diversification will hurt them more if AI’s promises do not work out. He also passes if he does not like a company’s plan for debt issuance or how its managers think about credit markets.
Because AI is so new, Queen says Capital Group is having many cross-group meetings and debates that go beyond the firm’s traditional sense of risk.
“You have to have good risk systems and models, but they will never capture the next thing very well,” he says.
Potential Risks on the Horizon
Lindsey Stewart, director of institutional insights at Morningstar, says institutional allocators are concerned about concentration in both public and private markets on both the equity and credit sides when it comes to AI and AI-adjacent themes.
In the bond market, concentration is not an issue, as hyperscaler debt makes up less than 10% of the total investment-grade market, Brill and Queen say. However, Brill says, because so much of the debt is longer-dated—with maturities of anywhere from 20 to 40 years—there are few companies outside of technology issuing long-maturity debt.
“The ability for us to get access to large liquid 30-year credit is pretty limited right now, other than in technology. As these companies have issued more and more, it’s been negative to the credit curve,” Brill says, adding it has likely crowded out demand for some U.S. Treasury issuance.
Hyperscalers are also using off-balance sheet leases and project finance to raise capital for their projects, which has the effect of obscuring the specific revenues backing the debt, another commonly cited source of investor concern. Brill says that, unlike previous instances when off-balance-sheet leases have covered fraudulent activity, the current AI buildout has been disclosed.
There are many unknowns about the AI/data center boom. Brill and Queen list several, including untested lease and contract documentation; political backlash that could slow the capital expenditure boom; demand fading because end-users do not find a good return on AI spending; technology obsolescence; and other issues. Because of the range of unpredictable factors, it takes more due diligence by analysts to weigh the risks. Those unknowns are also why buyers want higher ,yields despite the issuers’ credit ratings.
As credit investors, “we always expect the worse,” Brill says. “That’s how we’re underwriting this. We’re assuming that this doesn’t last forever, and we’re not going to assume that it goes to 0 next year. … We want to make sure that if the capex boom does stop, we don’t get hurt.”
 |
Data Center-Related Investments Available Across Most Asset Classes |
 |
How Investors Approach the Utilities Sector in the Age of AI |
 |
Energy Transmission Emerges as Bottleneck for Digital Infrastructure |
Tags: Artificial Intelligence, credit markets, data centers, electricity, fixed income strategies, Infrastructure, Technology