The artificial intelligence boom is driving an unprecedented surge in corporate debt, with Morgan Stanley estimating approximately $570 billion in global AI-related debt issuance for 2026. By the end of May, $236 billion of this had already been priced, a rate four times faster than the previous year. This substantial borrowing, particularly by hyperscalers like Nvidia and Amazon, is flooding fixed-income markets and prompting investors to demand better terms, including higher yields.
Major tech companies are leading this borrowing spree, with Nvidia issuing $25 billion in bonds in June and Amazon following with its own $25 billion offering in July. Amazon even had to offer extra yield to attract buyers, signaling a shift in market appetite. Alphabet and Meta also saw their bond spreads widen by 0.12 and 0.16 percentage points, respectively, after significant issuances. The tech sector now accounts for approximately 10% of the Bloomberg Corporate Bond Index, up from 9% in 2024.
This influx of debt is raising concerns about the financial health of some AI-focused companies. While not traditional debt, "hidden" or off-balance-sheet debt, primarily lease commitments, for Alphabet, Amazon, Microsoft, Meta, and Oracle has reached a combined $2.5 trillion. This figure now exceeds their transparent balance sheet debt of $1.35 trillion. Oracle's credit default swap (CDS) spreads have soared to 218 basis points, a level typically associated with junk bonds, and S&P even downgraded Oracle's long-term credit rating to 'BBB-', just one notch above junk status. Nvidia's five-year CDS touched 82 basis points, double its level in mid-June.
Investors are becoming increasingly cautious as the market's appetite for this new debt appears to be waning. The order coverage ratio for hyperscaler bond issuances, which was nearly 5x in February, dropped below 2x by July, indicating fewer bids for the bonds offered. For existing bondholders, rising yields on new issuances directly translate into mark-to-market losses on their current holdings. This situation suggests that the AI sector's reliance on debt funding may face significant challenges, especially with a 5% Treasury yield serving as a benchmark.
