AI borrowing costs rise as Treasury yields surge

With Treasury yields rising this week to their highest level since 2007, debt-reliant companies are poised to see borrowing costs rise. That means the development of AI infrastructure, which has reached historic levels, will be more expensive.

JPMorgan Chase It was estimated in June that $4.1 trillion in AI-related debt will be issued until 2030, as data center companies and others related to the artificial intelligence boom race to expand capacity to meet what industry experts consider an insatiable demand for AI services.

When borrowers return to the market, they are looking at a 10-year Treasury yield which sits close to 5.17%, up about 1 percentage point since the beginning of the year, which means that companies issuing debt are going to offer more attractive rates of return to lure investors.

The market is not in panic mode, at least not yet. Share of debt-heavy neocloud CoreWeave has held up fine, rising almost 8% this week, while Oraclewhich has counted on the debt market for the expansion of AI, has had a more difficult time, down 7% for the week and about 30% this year.

Meanwhile, Japan SoftBankthe main provider of capital for AI projects, raised $ 11.1 billion in a junk-bond sale this week, with a yield of up to 9.75% for a tranche of 7 years.

“The truth is that prices are not sensitive to these increases, which means they are price takers,” Mark Malek, chief investment officer at Siebert Financial, said in an interview. “In my opinion, many 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 ​​craze is a leading model developer OpenAI and Anthropiceach of which is worth close to $1 trillion in the private market. To provide the necessary infrastructure for advanced models, as well as models and services from a host of other companies, technology hyperscalers – Amazon, Google, Meta and Microsoft – has committed hundreds of billions of dollars this year in capital spending, with an increase expected to come in 2027.

While a healthy dose of these investments are funded through rising debt, the tech giants all have investment-grade credit ratings, providing cheaper access to capital. But for other packages, bigger challenges will be there, according to some market participants.

A senior private equity investor, who asked not to be named to speak candidly about the matter, told CNBC that, moving forward, neocloud deals will be harder to finance, as companies have less of a cushion to absorb costs.

Riley Thompson, vice president at Mitsubishi HC Capital America said in an interview that lenders are more selective about which projects to finance even if borrowers agree to pay higher rates.

“Instead of a roster of 50 neoclouds, there might be 20 that the market wants,” Thompson said.

CoreWeave, which went public last year, warned of rate hikes in an SEC filing. In its most recent quarterly filing, the company said that, in June, every 100 basis point (1 percentage point) rate increase could result in a $30 million increase in interest expense, based on its outstanding floating-rate debt balance.

An early warning sign may have landed this week, when Oracle shares slid after Bloomberg reported that the company had filed a “force majeure” notice related to a New Mexico data center project to protect itself from higher costs. The company wants to delay payments on the campus, called Project Jupiter, if it fails to come online as expected by 2028, the report said. Oracle said the project “remains on schedule as planned.”

Rising interest rates aren’t the only problem. Ahead of this week’s results surge, the CEOs of Anthropic and OpenAI have begun urging a slowdown in AI development after industry researchers went public with concerns that advanced models could slip out of human control.

At the same time, the national backlash against AI data centers has emerged as a major issue in November’s midterm elections, with 69% of respondents to a new NBC News Decision Desk Poll, powered by SurveyMonkey, saying they oppose the construction of such facilities in local areas. On Monday, Republican Texas Gov. Greg Abbott, who is in the midst of a tight race for re-election, ordered a freeze on all environmental permits related to data centers following a grid approval moratorium last month.

Still, demand for AI services is exploding. The latest example is personal assistant app Muse Meta, which has been popular since its launch in September. Muse scored over 2.5 million global downloads in its first two weeks, surpassing ChatGPT at the top 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.

Andrew Giudici, global head of corporate, project and infrastructure finance at credit rating agency KBRA, said that while rate hikes may affect future deals, they are unlikely to have a major impact on borrower demand.

“In a normal environment, people can kick back and rest,” 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 chairman of Latham & Watkins’ emerging companies and growth practice, said there’s no doubt that as costs rise, “people have to absorb it.”

“But absorbing it in that kind of demand structure, where there’s a lot of demand, is much easier,” said Zaltzman, who works on financing AI infrastructure.

The equation is simpler for Bernie Margulies, CEO of American Compute, who advises on risk management for GPU financing. He said borrowers are keen to secure funding even at higher costs, especially if they have commitments with OpenAI and Anthropic, which have signed contracts to secure future computing capacity.

“If you had a deal with Anthropic, would that be 50 basis points off?” Margulies said.

WATCH: Meta’s Muse is likely to reach 100 million users