AI bonds sell off as Meta and Microsoft earnings put spending in focus
AI-linked corporate debt widened in July as investors braced for more Big Tech spending plans from Microsoft, Meta and Amazon.
By Frankie Delgado · News Reporter
3 min read
The AI bonds selloff picked up in July, with debt tied to the artificial-intelligence buildout weakening just as some of the biggest tech spenders prepare to show investors their latest numbers.
Alphabet, Google’s parent, told investors last week that it expects full-year AI spending of $195 billion to $205 billion, according to MarketWatch. Microsoft, Meta Platforms and Amazon are scheduled to report earnings on Wednesday and Thursday, while Oracle reported in June and Nvidia is due in late August.
Microsoft declined to comment to MarketWatch. Meta, Amazon, Oracle and Nvidia did not immediately respond to requests for comment, MarketWatch reported.
Why are AI bonds selling off?
AI-linked bonds have weakened because investors are weighing how much money hyperscalers need to borrow to fund data centers, chips and other AI infrastructure. In bond markets, a selloff often shows up as wider spreads, meaning investors demand a bigger premium over Treasury yields to hold the debt.
BofA Global data cited by MarketWatch showed AI-related debt spreads moving wider in recent weeks, while the broader high-grade corporate-bond market held up better. Telecom and technology also showed notable spread moves in July, according to the BofA Global chart.
Lukasz Labedzki, a fixed-income analyst at the Franklin Templeton Institute, said Monday that he expects issuers to turn more cautious. He pointed to a heavy year for U.S. investment-grade bond sales: more than $1.2 trillion was issued in the first half of 2026, the most since 2021, when low rates encouraged large U.S. companies to borrow aggressively.
About $200 billion of this year’s high-grade supply has come from hyperscalers, Labedzki said. He also noted that the borrowing arrives while the U.S. government still has major funding needs.
Corporate bonds trade at a spread over Treasury rates to compensate buyers for credit risk. BondCliQ data cited by MarketWatch showed longer AI hyperscaler bonds, especially those maturing in 10 years or more, taking the heavier hit.
On an issuance-weighted basis, 10-year AI debt from the group traded at roughly 121 basis points above Treasurys, MarketWatch reported. That compared with about 80 basis points for the ICE BofA US Corporate Index, up from about 73 basis points in June.
Bryce Doty, senior portfolio manager at Sit Fixed Income Advisors, told MarketWatch that money used to buy new hyperscaler bonds has to come from somewhere. If investors pull cash from other sectors to fund those purchases, he said, some spillover can follow.
What does this mean for Treasury yields?
The heavy AI borrowing has also added pressure to Treasury yields, according to the MarketWatch report. The 10-year Treasury yield slipped slightly Monday but remained near 4.65%, while the 30-year yield stayed above 5% for its longest stretch since 2007.
Moody’s Ratings said last week that it expects AI-related capital spending by hyperscalers to reach $1 trillion next year. Joe Boyle, head of asset-class specialists at Hartford Funds, told MarketWatch that the supply should keep yields elevated.
Boyle also said many AI hyperscalers are high-quality companies with business lines beyond AI that can support their debt. He added that wider spreads on strong corporate borrowers may pull some demand away from Treasurys.
The backdrop is a U.S. debt pile pushing $39.7 trillion, according to Treasury fiscal data cited by MarketWatch. Boyle said that also adds structural upward pressure on rates.
This story draws on original reporting from MarketWatch.