Updated
Updated · CNBC · Nov 11
Morgan Stanley Backs $67 Billion of Hyperscaler AI Bonds as Lower-Grade Debt Risks Deepen
Updated
Updated · CNBC · Nov 11

Morgan Stanley Backs $67 Billion of Hyperscaler AI Bonds as Lower-Grade Debt Risks Deepen

3 articles · Updated · CNBC · Nov 11

Summary

  • Morgan Stanley said high-quality hyperscaler bonds now offer an attractive entry point, arguing investment-grade AI issuers are better positioned than junk-rated data-center and neocloud borrowers.
  • Debt concerns have intensified as AI-related issuance flooded credit markets, pushing up yields and widening five-year CDS spreads; CoreWeave’s bond yield topped 16% earlier this month before easing sentiment with a revenue beat and raised guidance.
  • $67 billion of new bonds from Microsoft, Meta, Oracle and Alphabet have been sold across U.S. and European markets since Sept. 30, and Morgan Stanley said spreads have tightened on most of those deals.
  • Morgan Stanley expects little additional AI bond issuance this year, but sees 2027 capex from AI names reaching $1.66 trillion—up 75% from 2026—as post-election permitting and local data-center restrictions may delay projects.
  • That backdrop should widen financing gaps between stronger and weaker AI borrowers, with Morgan Stanley saying hyperscalers’ balance sheets leave them better able to absorb higher borrowing costs and keep building.

Insights

With off-balance-sheet debt surging, are safe hyperscaler bonds masking a massive hidden risk if AI monetization stalls after 2027?
As AI infrastructure costs skyrocket into the trillions, can neoclouds survive the debt-fueled cycle without being swallowed by tech giants?