Tech news in 3 minutes

Debt-hungry AI companies face increased risk as bond yields spike

10 h ago

**AI Infrastructure Debt Costs Surge as Treasury Yields Hit 5.17%, Threatening $4.1 Trillion Buildout** Rising Treasury yields are driving up borrowing costs for the AI infrastructure buildout, with the 10-year yield near 5.17%—up about 1 percentage point since the start of the year—threatening to make the already historic $4.1 trillion in AI-related debt issuance through 2030 (per JPMorgan Chase) even more expensive for data center companies and hyperscalers racing to meet insatiable AI demand. CoreWeave, a debt-heavy neocloud, has held up with an 8% weekly gain, but Oracle fell 7% this week and 30% year-to-date after a Bloomberg report said it sent a "force majeure" notice tied to its New Mexico data center project (Project Jupiter) to protect against higher costs—though Oracle says the project remains on schedule. Japan's SoftBank raised $11.1 billion in junk bonds this week, with yields as high as 9.75% for the 7-year tranche, reflecting a "price insensitive" capital grab, according to Siebert Financial's Mark Malek. CoreWeave's SEC filings warn that every 100-basis-point rate increase could add $30 million to its interest expense. A senior private credit investor told CNBC that neocloud deals will become harder to finance, while Mitsubishi HC Capital's Riley Thompson said lenders are getting pickier, narrowing the viable neocloud roster from 50 to about 20. Despite rising rates, demand for AI services remains explosive. Meta's Muse app hit 2.5 million downloads in two weeks, topping ChatGPT on Apple's App Store. KBRA's Andrew Giudici expects continued large issuance, and American Compute's Bernie Margulies noted that borrowers with commitments from OpenAI or Anthropic—both valued near $1 trillion—won't be deterred by small rate increases. However, a nationwide backlash is growing: 69% of respondents oppose local data centers, and Texas Gov. Greg Abbott halted related environmental permits.