Eric Cantor of Moelis & Co. expressed growing concerns regarding the financial markets, specifically the substantial amount of debt being issued and anticipated to be issued, driven by the AI supercycle. He highlighted that while there's a strong desire to transact on the private equity side and capital is available for sound companies, the preferred financing method for this investment supercycle is currently private debt.
Cantor pointed out a notable compression in credit spreads, with the 10-year corporate investment grade instrument versus the 10-year Treasury spread falling below 100 basis points, historically averaging 150 basis points. He suggested this could indicate either healthier corporations or a less healthy government. This environment signifies a "race for capital" amidst a global trend of increasing populism, which pressures governments to find ways to finance their needs and populations.
His observations align with broader concerns within the financial industry. S&P analysts have also warned about the weakening credit quality of hyperscalers, despite their high ratings, due to massive borrowing for AI-related infrastructure. They project the top six hyperscalers will spend over $7 trillion on data centers and AI capital expenditures through 2030. JPMorgan estimates that six AI giants, including Oracle and Microsoft, issued approximately $320 billion in debt this year, absorbing 68% of new long-duration U.S. Treasury borrowing. Goldman Sachs projects $340 billion in hyperscaler issuance in 2027 and estimates total global AI-related investment to exceed $1 trillion in 2026, with the U.S. accounting for about $581 billion. Much of this AI spending, including Amazon's $100 billion in bond market raises this year, is funded by debt.
