The AI build-out is now a credit story: bond markets are pricing the debt behind it

The debt behind the build-out

The scale of AI spending is the starting point for every credit conversation in this batch of podcasts. Morgan Stanley's Andrew Sheets put global AI-related capital expenditure at $900 billion in 2026, while a Bloomberg Surveillance discussion put investment-grade AI capital expenditure above $1 trillion over the next two years. The Surveillance guests added that fixed income moves over the last two to three months were driven by the sheer volume of AI-related debt issuance, not by a change in underlying credit quality alone. m:…6201 m:…1799

That issuance is already changing prices. Morgan Stanley's discussion said the tech sector's bond issuance has pushed up both yields and spreads in credit markets. In other words, the market is no longer treating AI borrowing as a footnote to the equity story; it is repricing the debt itself, and the supply of new bonds is part of the price. That repricing is part of why dispersion matters: when yields rise unevenly, the cost of funding the build-out differs sharply across issuers, and lenders start distinguishing between balance sheets rather than funding the sector as a whole. m:…6201

Dispersion replaces the single AI trade

The second theme is dispersion. Morgan Stanley's podcast said credit market dispersion has increased on fears of AI-driven disruption in sectors such as telecom, and that the gap between high-growth tech issuers and slower-growth sectors creates openings for active credit management. A Bloomberg Surveillance episode made the related point that AI funding is becoming dependent on both equity and credit markets as AI cannibalises the base businesses that used to generate cash. m:…6201 m:…9716

The equity session on October 30 showed the same selectivity. Bloomberg's Christina Kino reported that the sell-off was broad, with the S&P 500 down 1.8% and the Nasdaq Composite down 2.9% in the session, and that investors are increasingly favouring companies that limit capital expenditure. The VIX closed at 19.4 after spiking to between 24 and 25 earlier in the session, a reminder that the repricing arrived with a jump in volatility rather than a calm rotation. m:…3159

Apple's upcoming earnings report was singled out on Bloomberg Talks as a test of investor sentiment on hardware spending in a K-shaped economy, where different income groups experience very different conditions. The through-line across the shows is that investors are no longer underwriting AI spending in general; they are asking which balance sheets can carry it. m:…3159

What credit investors watch next

For now, the credit backdrop still looks attractive to the people paid to buy bonds. Morgan Stanley's discussion noted that US investment-grade credit yields close to 6% and European investment-grade credit yields over 4%, and that a market with high yields and no central bank intervention matches what investors have long said they wanted. That framing cuts both ways: yields are high partly because supply is heavy and dispersion is wide. m:…6201

One caution came from 42 Macro's quant systems, discussed on Bloomberg Surveillance, which have shifted to a risk-off regime favouring quality and lower-beta exposure. Another came from the same Surveillance episode's observation that the market has already priced in approximately two interest rate hikes from the Federal Reserve, leaving less room for a dovish surprise to bail out stretched borrowers. The open question for credit desks is whether the cash flows from AI arrive before the refinancing wall does. m:…1799 m:…9716