AI Debt Starts Moving the U.S. Treasurys Market

Watch on YouTube ↗  |  September 08, 2026 at 20:24  |  9:05  |  Morgan Stanley
Speakers
Vishy Tirupattur — Chief Fixed Income Strategist, Morgan Stanley
Matthew Hornbach — Global Head of Macro Strategy

Summary

Morgan Stanley strategists Matthew Hornbach and Vishy Tirupattur discuss how AI infrastructure financing is broadening across corporate credit markets and shifting from long-dated data-center debt toward shorter chip-financing structures. They argue this supply dynamic is spilling into U.S. Treasurys, especially via dealer hedging that pressures long-end yields. They also flag uncertainty around issuance cadence and potential volatility.

  • Hyperscaler AI capex is large and persistent because compute demand exceeds supply.
  • AI financing is broadening across investment grade, high yield, leveraged loans, private credit, and securitized credit.
  • AI-related debt is shifting from 20-plus-year data-center financing toward roughly five-year chip/component financing.
  • Political pushback and data-center moratoria may slow longer-term AI capex.
  • Long-end Treasury yields have been persistently high, partly on dealer selling of Treasuries to hedge corporate bond supply.
  • New issuers' lack of an established issuance cadence may create market volatility.
Ideas
Vishy Tirupattur Chief Fixed Income Strategist, Morgan Stanley 0:37
AI capex financing broadens across credit markets
AI hyperscalers have enormous capex needs because demand for compute vastly exceeds supply, and as long as that imbalance persists, the capex must be financed across credit markets: public and private investment grade, high yield, leveraged loans, private credit, and securitized credit. Issuance has been broader and deeper than expected, including in seven currencies.
LQD Leveraged Loans AI-related corporate credit HYG BIZD Securitized Credit HIGH
Vishy Tirupattur Chief Fixed Income Strategist, Morgan Stanley 2:44
AI financing shifts to five-year debt
AI infrastructure financing is shifting from financing entire data centers with 20-plus-year debt toward financing components such as chips. Chips face technological obsolescence and need refinancing in about five years, so more structures are becoming amortizing five-year loans. Political pushback against data centers may slow long-term capex while near-term chip replacement continues, focusing more issuance at the five-year point of the curve.
5-year U.S. Treasuries 5-year corporate credit MED
Matthew Hornbach Global Head of Macro Strategy 5:24
Long-end Treasuries pressured by AI debt supply
Long-end U.S. Treasury yields have been persistently high, with 30-year yields making multi-decade highs. A likely driver is that investors in AI-related corporate bonds are preparing balance sheets for expected September issuance; dealers absorb that duration and then sell Treasuries to recycle risk, keeping pressure on the 20-plus-year Treasury sector.
TLT 30-year U.S. Treasuries MED
Up Next
6:46
Stocks Slide as U.S.-Iran Tensions Push Oil Toward $99 | Closing Bell
Bloomberg Markets · Sep 08
10:09
El-Sayed Says Trump, Vance, Rogers Have No Answers
Bloomberg Markets · Sep 08
7:12
Oil Near $100: China’s Demand Story Doesn’t Add Up | Art Berman
Wealthion · Sep 08
1:13
Black Sea Disruption Could Push Food Prices Higher | Presented by CME Group
Bloomberg Markets · Sep 08
1:33
Is Silver Ready to Steal the Spotlight Again?  | Presented by CME Group
Bloomberg Markets · Sep 08

This Morgan Stanley video, published September 08, 2026, features Vishy Tirupattur, Matthew Hornbach discussing LQD, Leveraged Loans, AI-related corporate credit, HYG, BIZD, Securitized Credit, 5-year U.S. Treasuries, 5-year corporate credit, TLT, 30-year U.S. Treasuries. 3 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Vishy Tirupattur, Matthew Hornbach  · Tickers: LQD, Leveraged Loans, AI-related corporate credit, HYG, BIZD, Securitized Credit, 5-year U.S. Treasuries, 5-year corporate credit, TLT, 30-year U.S. Treasuries