AI debt surge forces bond investors to absorb mark-to-market losses
In brief
- Morgan Stanley estimates $570 billion in global AI-related debt issuance for 2026
- Tech sector now represents 10% of Bloomberg Corporate Bond Index, up from 9% in 2024
- $236 billion in AI debt priced by May 2026, four times faster than 2025 pace
- Hyperscaler order coverage ratios fell from 5x in February to below 2x by July
- Rising yields on new issuances create mark-to-market losses for existing bondholders
The AI debt deluge
Morgan Stanley estimates roughly $570 billion in global AI-related debt issuance for 2026. The pace has already accelerated far beyond historical norms. By the end of May, $236 billion of that had already been priced, a pace approximately four times faster than the same period in 2025.
Individual issuances have reached staggering scale. Nvidia issued $25 billion in bonds in June to fund AI infrastructure buildout. Amazon followed with its own $25 billion offering in July. Amazon even had to sweeten the deal with extra yield to attract buyers. These aren't niche transactions—they're reshaping the entire fixed-income landscape.
Demand is cracking
The warning signs are clear in the data. The order coverage ratio for hyperscaler bond issuances sat at nearly 5x back in February. By July, that ratio had dropped below 2x. That collapse matters. It means fewer investors are chasing each new deal, a sign that appetite isn't keeping pace with supply.
Spreads are widening too. Alphabet and Meta saw their bond spreads widen by 0.12 and 0.16 percentage points, respectively, after their substantial issuances. When spreads widen after you issue, it signals the market is repricing risk upward—not a vote of confidence.
The pain for existing holders
The tech sector now constitutes approximately 10% of the Bloomberg Corporate Bond Index, up from 9% in 2024. That concentration shift happened fast. For existing bondholders, rising yields on new issuances translate directly into mark-to-market losses on their current holdings. If you bought a tech bond at 4% and the market is now pricing new issuance at 5%, your bond's value drops—on paper, immediately, whether you sell or not.
This creates a vicious cycle. Investors holding underwater bonds face pressure to rebalance. That selling pressure pushes yields higher. Higher yields trigger more losses. The flood of AI-driven debt issuance is no longer a sideshow—it's rewriting the fixed-income playbook.
