AI Infrastructure Boom Drives Shift Toward Amortizing Junk Bonds

The race to build out artificial intelligence infrastructure is reshaping the high-yield debt market, with issuers increasingly offering early repayment terms to secure necessary funding. As the demand for compute power surges, the financial structures used to fund the physical layer of AI are evolving to meet investor risk appetites. A borrowing frenzy exceeding $20 billion has hit the junk-bond sector this year as companies scramble to construct the massive data centers required for AI workloads. To make these speculative-grade offerings more appealing, several issuers are implementing a feature known as annual amortization. Unlike traditional 'bullet' bonds, where the entire principal is repaid at the end of the term, annual amortization requires the borrower to return a percentage of the outstanding debt each year. This structure provides investors with a faster return of capital and reduces the overall duration of the credit risk. Companies such as cloud infrastructure firm Core Scientific Inc. and data center operator Cipher Digital Inc. have already agreed to these terms. By offering these rare sweeteners, firms are attempting to maintain access to capital markets despite the inherent risks associated with high-leverage infrastructure projects. This shift indicates a high level of competition for capital in the speculative-grade market. By reducing the risk profile for lenders, these issuers can potentially lower their cost of capital or attract a broader base of institutional investors during a period of intense industry expansion.