AI Infrastructure Boom Tests Debt Markets as Treasury Yields Surge to 17-Year Highs

Spools of electrical wires outside a series of assembly tents during a media tour of the Stargate AI data center in Abilene, Texas, US, on Tuesday, Sept. 23, 2025. Stargate is a collaboration of OpenAI, Oracle and SoftBank, with promotional support from President Donald Trump, to build data centers and other infrastructure for artificial intelligence throughout the US.

Kyle Grillot | Bloomberg | Getty Images

With Treasury yields climbing this week to their highest levels since 2007, companies dependent on debt financing are confronting rising borrowing costs. The AI infrastructure buildout, already at historic levels, is poised to become significantly more expensive.

JPMorgan Chase estimated in June that $4.1 trillion in AI-related debt will be issued through 2030, as data center operators and other firms tied to the artificial intelligence boom race to expand capacity to meet what industry experts describe as insatiable demand for AI services.

As borrowers return to the market, they face a 10-year Treasury yield near 5.17%, up roughly one percentage point since the start of the year. This means companies issuing debt must offer more attractive returns to attract investors.

The market is not in panic mode, at least not yet. Shares of debt-heavy neocloud provider CoreWeave have held up well, rising nearly 8% this week, while Oracle, which has relied on debt markets for its AI expansion, has struggled, falling 7% for the week and about 30% year-to-date.

Stock Chart IconStock chart icon

CoreWeave vs. Oracle this week

Meanwhile, Japan’s SoftBank, a principal capital provider for AI projects, raised $11.1 billion in a junk-bond sale this week, with yields reaching as high as 9.75% for the 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 an interview. “In my view, a lot of these companies need to be price insensitive. They need to get as much capital as possible to compete.”

At the center of the AI frenzy are leading model developers OpenAI and Anthropic, each valued at close to $1 trillion in private markets. To supply the infrastructure required for their advanced models—as well as those from a host of other companies—tech’s hyperscalers Amazon, Google, Meta, and Microsoft have committed hundreds of billions of dollars in capital expenditures this year, with further increases expected in 2027.

While a significant portion of that investment is funded through debt issuance, those tech giants all hold investment-grade credit ratings, granting them cheaper access to capital. For the rest of the field, however, steeper challenges loom, according to market participants.

Warning signs?

A senior private credit investor, who requested anonymity to speak candidly, told CNBC that neocloud deals will become more difficult to finance going forward, as these companies have less cushion to absorb higher costs.

Riley Thompson, a vice president at Mitsubishi HC Capital America, said in an interview that lenders are growing more selective about the projects they will fund, even if borrowers agree to pay higher rates.

“Instead of a roster of 50 neoclouds, there’s probably 20 that the market’s truly interested in,” Thompson said.

CoreWeave, which went public last year, has flagged rising rates in its SEC filings. In its latest quarterly filing, the company disclosed that as of June, every 100-basis-point (1 percentage point) increase in rates could add $30 million to its interest expense, based on its outstanding floating-rate debt balance.

An early warning sign may have emerged this week when Oracle’s stock slid following a Bloomberg report that the company sent a “force majeure” notice tied to its New Mexico data center project to shield itself from higher expenses. The company is seeking to delay payment on the campus, dubbed Project Jupiter, if it fails to come online as expected in 2028, according to the report. Oracle stated the project “remains on our planned schedule.”

Rising interest rates are not the only headwind. Prior to this week’s yield spike, the CEOs of Anthropic and OpenAI had begun urging a slowdown in the pace of AI development after industry researchers publicly warned that advanced models risk spinning out of human control.

At the same time, a nationwide backlash against AI data centers has emerged as a major issue ahead of November’s midterm elections, with 69% of respondents to a recent NBC News Decision Desk Poll, powered by SurveyMonkey, opposing the construction of such facilities in their local areas. On Monday, Texas Republican Gov. Greg Abbott, who is in a tight reelection race, ordered a temporary halt to all data center-related environmental permits following a moratorium on grid approvals last month.

Still, demand for AI services is exploding. The latest example is Meta’s Muse personal assistant app, which has surged in popularity since its launch earlier in September. Muse recorded more than 2.5 million global downloads in its first two weeks, surpassing ChatGPT atop Apple’s App Store, and Evercore’s Mark Mahaney told CNBC this week that it could reach 100 million users within six to 12 months.

Mark Zuckerberg, chief executive officer of Meta Platforms Inc., reveals the Muse Charm device during the Meta Connect event in Menlo Park, California, US, on Wednesday, Sept. 23, 2026.

Minh Connors | Bloomberg | Getty Images

Andrew Giudici, global head of corporate, project, and infrastructure finance at credit rating agency KBRA, said that even as rising rates may affect future deals, he does not see a major impact on borrower demand.

“In a normal environment, people might take a step back and pause a bit,” Giudici said. “But I don’t think that’s going to happen here. I think you’re going to continue to see relatively large issuance.”

Haim Zaltzman, vice chair of Latham & Watkins’ emerging companies and growth practice, said there is no doubt that as costs rise, “somebody will have to absorb it.”

“But absorbing it in that kind of demand structure, where the demand is so great, is a lot easier,” said Zaltzman, who works on AI infrastructure financing.

The equation is even simpler for Bernie Margulies, CEO of American Compute, which advises on risk management for GPU financing. He said borrowers are eager to secure financing even at higher costs, especially if they have commitments with OpenAI and Anthropic, which have been signing contracts to lock in compute capacity years into the future.

“If you have a deal with Anthropic, will 50 basis points really stop you?” Margulies said.

WATCH: Meta’s Muse will likely soar to 100 million users

Source link