As Treasury yields climb to their highest levels since 2007, artificial intelligence companies heavily reliant on debt are facing steeper borrowing costs, threatening to make the already expensive AI infrastructure buildout even more costly. JPMorgan Chase estimated in June that $4.1 trillion in AI-related debt will be issued through 2030 as data center firms race to meet insatiable industry demand.
The 10-year Treasury yield currently sits near 5.17%, approximately one percentage point higher than at the start of the year. This shift forces companies issuing corporate debt to offer more attractive returns to secure investors, a trend illustrated by SoftBank’s recent junk-bond sale. The Japanese investment firm raised $11.1 billion this week, with yields reaching as high as 9.75% for its seven-year tranche.
“They basically are price insensitive to that raise, which means they’re price takers,” said Mark Malek, chief investment officer at Siebert Financial. “In my view, a lot of these companies need to be price insensitive. They need to get as much capital as possible to compete.”
Market reactions have been mixed. CoreWeave shares, a debt-heavy neocloud provider, rose nearly 8% this week, while Oracle fell 7% for the period and has dropped roughly 30% year-to-date. Oracle recently made headlines after reports emerged that it sent a force majeure notice for its Project Jupiter data center in New Mexico to protect against rising expenses, although the company maintains the project remains on schedule.
Lenders are becoming increasingly selective. Riley Thompson, vice president at Mitsubishi HC Capital America, noted that while borrowers may agree to higher rates, financing for neocloud deals will become more difficult due to thinner cushions for absorbing costs. “Instead of a roster of 50 neoclouds, there’s probably 20 that the market’s truly interested in,” Thompson said.
The financial pressure is quantifiable for public companies. In its latest SEC filing, CoreWeave warned that every 100-basis-point increase in rates could add $30 million to its interest expense based on outstanding floating-rate debt. Meanwhile, tech giants like Amazon, Google, Meta, and Microsoft continue to commit hundreds of billions in capital expenditures, bolstered by their investment-grade credit ratings that provide cheaper access to capital.
Despite the headwinds, demand for AI services remains robust. Meta’s new Muse personal assistant app garnered over 2.5 million downloads in its first two weeks, surpassing ChatGPT on Apple’s App Store. Evercore’s Mark Mahaney predicted the app could reach 100 million users within six to 12 months.
Andrew Giudici of KBRA observed that while rising rates might cause caution in a typical environment, he does not expect a significant pause in borrower demand. Haim Zaltzman of Latham & Watkins agreed, noting that absorbing higher costs is easier given the current demand structure. Bernie Margulies, CEO of American Compute, added that firms with contracts from major developers like OpenAI and Anthropic are eager to secure financing even at elevated rates, questioning whether a 50-basis-point difference would stop a deal.
Beyond finance, the AI expansion faces political and social friction. A recent NBC News poll indicated that 69% of respondents oppose local data center construction. Additionally, Texas Governor Greg Abbott recently ordered a temporary halt to data center environmental permits, following a previous moratorium on grid approvals, as the issue becomes prominent ahead of November’s midterm elections.
Wonder if locals will keep opposing data centers? The political pressure in Texas seems like the next big hurdle.
SoftBank’s 9.75% yield is wild. It shows how desperate these companies are to build, regardless of the cost.