Is credit spread widening a harbinger of future stock returns?

Credit markets have a track record of flashing warnings before equities catch on. High-yield spreads widened well ahead of both the Dot-Com Bubble and the Great Financial Crisis, giving bond investors an early read on corporate distress that stock investors got later.

Today, a historic wave of investment-grade debt is funding the buildout of AI data centers, and AI-related credit spreads are starting to widen. The question worth asking: is this the same early warning, or a different signal entirely?

The debt boom and widening spreads

The scale of AI capital spending is showing up directly in the bond market. IG corporate issuance is up more than 40% year-over-year in duration-equivalent terms, and AI-related debt now makes up roughly 6% of the overall IG index.¹

That supply is moving prices. The hyperscaler credit index recently touched new wides of 145 basis points.² Stress is showing up further down the quality ladder too, with CCC-to-BB spreads decompressing to 705 basis points, their widest level since November 2023.³

The bear case: Uncertain ROI

The open question is timing. During a recent earnings call, American Express leadership expressed skepticism about immediate returns, stating they are "not so sure how many companies have really committed to actual P&L savings [from AI] for 2026 and probably not even for 2027.”⁴ That uncertainty echoes the dot-com telecom buildout: billions poured into fiber-optic infrastructure created overcapacity that took years to absorb, and returns fell well short of the original pitch.

The bull case: Three reasons this cycle looks different

The market can absorb the supply: Hyperscaler credit quality remains strong enough to take on this debt without much friction. Pricing suggests Google could issue up to $50 billion in new bonds at only a 10 basis point yield concession.⁵

The spread widening so far looks like the market adjusting to new supply rather than hitting a real constraint.

The economics pencil out: If $6 trillion is spent on AI capex through 2030, AI model providers would need $1.8 trillion in revenue by 2030 to justify it.⁶ This equates to roughly 8.5% of the projected S&P 500 total revenues that year.⁶ That's a demanding target, but not unreasonable.

This may be rotation, not a warning: Rather than a crash signal, the widening could simply reflect investors rotating toward AI adopters and away from pure infrastructure plays, representing a normal mid-cycle shift rather than a red flag.

Dot-com telecom or durable foundation: the credit market will likely tell us which one this is before the stock market does. That's exactly why it's worth watching now.

Contact my colleagues at Farther Institutional to discuss opportunities we see in AI credit, equities, and alternatives.

—

Disclaimer: This material is for informational purposes only and does not constitute personalized financial advice or a solicitation to buy or sell any security. Forward-looking statements are based on assumptions and current market conditions, which are subject to change.

References
¹ Morgan Stanley Research, "Markets On the Move: Signal or Noise?", July 27, 2026.
² J.P. Morgan Research, "US Market Intelligence", July 28, 2026 (Index methodology notes hyperscaler index widened to 145bp).
³ J.P. Morgan Research, "JPM High-Yield and Leveraged Loan Morning Intelligence", July 28, 2026.
⁴ American Express corporate earnings call commentary, cited in J.P. Morgan Research, "US Market Intelligence", July 28, 2026.
⁵ J.P. Morgan Research, "US Market Intelligence", July 28, 2026.
⁶ J.P. Morgan Research, "US Market Intelligence", July 28, 2026.