Rising U.S. Treasury yields have pushed borrowing costs higher for the artificial intelligence infrastructure buildout, threatening to make an already massive $4.1 trillion debt expansion significantly more expensive for data center operators and technology companies through 2030.
The 10-year U.S. Treasury yield climbed this week to approximately 5.17%, sitting about one full percentage point higher than its level at the start of the year. The increase forces corporate borrowers racing to construct high-capacity data centers to offer more attractive returns to entice investors back to the market.
JPMorgan Chase estimated in June that $4.1 trillion in AI-related debt will be issued by the end of the decade. While major cloud computing giants such as Amazon, Google, Meta, and Microsoft possess investment-grade credit ratings that cushion them against spiking rates, smaller market participants and debt-heavy operators face distinct financial pressures.
Capital Access and Market Pressures
Japan's SoftBank highlighted the aggressive push for liquidity this week by raising $11.1 billion through a junk-bond sale, featuring yields as high as 9.75% on its 7-year tranche. Mark Malek, chief investment officer at Siebert Financial, noted in an interview that the conglomerate acted as a price taker due to being essentially price-insensitive in its capital raise, driven by the sheer necessity to secure funds to compete.
For neocloud companies, the environment is increasingly restrictive. A senior private credit investor told CNBC that future neocloud financing deals will encounter steeper hurdles because these firms maintain thinner financial cushions to absorb climbing expenses. Riley Thompson, vice president at Mitsubishi HC Capital America, added in an interview that lenders are growing selective, noting that the pool of viable neocloud borrowers has narrowed from roughly 50 to about 20.
CoreWeave, a debt-reliant neocloud that went public last year, stated in its latest quarterly SEC filing that every 100-basis-point increase in interest rates could add $30 million to its interest expense based on its floating-rate debt balance. Despite these headwinds, CoreWeave shares rose nearly 8% during the week.
By contrast, Oracle shares fell 7% for the week and roughly 30% for the year. The decline followed a Bloomberg report indicating that Oracle issued a force majeure notice regarding its New Mexico data center project, known as Project Jupiter, to protect itself from escalating expenses by seeking payment delays if the campus misses its scheduled 2028 opening. Oracle maintained that the project remains on its planned schedule.
Regulatory Headwinds and Local Backlash
The financial strain coincides with mounting regulatory and political resistance to the physical footprint of the AI boom. Prior to the recent spike in bond yields, the CEOs of OpenAI and Anthropic began advocating for a slower development pace following concerns that advanced models risk spinning out of human control.
At the same time, local opposition to data center construction has grown ahead of November midterm elections. A recent NBC News Decision Desk Poll powered by SurveyMonkey found that 69% of respondents opposed the construction of AI data facilities in their local areas.
Political friction materialized directly in Texas, where Republican Governor Greg Abbott ordered a temporary halt to all data center-related environmental permits on Monday, following a prior moratorium on grid approvals issued the preceding month.
