
Rising Bond Yields Put AI Infrastructure Financing Under Pressure
Published by AINave Editorial • Reviewed by Ramit
Treasury yields climbing to their highest levels since 2007 are adding cost to the AI infrastructure buildout, but they are not affecting every borrower in the same way. In a September 27 report, CNBC put the 10-year Treasury yield near 5.17%, about one percentage point above the start of 2026, and said companies issuing debt would need to offer investors more attractive returns. JPMorgan Chase estimated in June that $4.1 trillion in AI-related debt would be issued through 2030. That is a forecast, not a tally of debt already raised. CNBC reported
The more telling divide is between companies with cheaper access to capital and those with less room to absorb financing costs. Amazon, Google, Meta and Microsoft have investment-grade credit ratings, which give them cheaper access to capital than less highly rated borrowers. By contrast, market participants told CNBC that neocloud financing may get harder: one private-credit investor cited these companies’ limited cushion for higher costs, while a Mitsubishi HC Capital America executive said lenders were getting pickier about which projects they would fund, even if borrowers offered higher rates. Those differences in credit access and reported lender selectivity suggest that a higher market rate can affect both the price of a loan and whether a project gets financed at all.
Floating-rate debt makes the exposure concrete
CoreWeave’s filing offers a specific measure of rate sensitivity. The company said that, based on its outstanding floating-rate debt balance as of June, each 100-basis-point increase in rates could add $30 million to its interest expense. That is a company-disclosed sensitivity tied to that debt balance, not a guaranteed increase in future costs. CoreWeave’s filing details the exposure.
SoftBank provides a separate example of what borrowing can cost: the Japanese company raised $11.1 billion in a junk-bond sale, with yields as high as 9.75% for the seven-year tranche. That rate reflects a specific bond sale, not a standard price for AI infrastructure debt. Meanwhile, a KBRA executive told CNBC he expected relatively large issuance to continue despite higher rates. The reporting therefore points to ongoing demand for capital alongside more expensive and selective financing, not a sector-wide halt. The bond sale and market outlook
Project financing also depends on timing
Oracle’s New Mexico Project Jupiter illustrates how financing concerns can intersect with construction schedules. CNBC reported that Oracle’s stock fell after a Bloomberg report said the company sent a force majeure notice tied to the project, potentially protecting it from higher expenses and delaying payment if the campus missed its expected 2028 start. Oracle said the project remained on its planned schedule, so the reported notice should not be read as confirmation of a delay. The report and Oracle’s response
For infrastructure operators, the pressure is not captured by a single Treasury yield. It also depends on the borrower’s credit access, the terms of existing debt and lenders’ assessment of each project. Strong demand may keep financing activity moving, but it does not guarantee that every planned build will secure capital on the same terms.




















