The smartest money in credit is splitting into two camps: those betting AI will mint new winners, and those hunting companies AI can't touch.

The Summary

The Signal

At Bloomberg's Global Credit Forum in London, two of the world's largest credit investors said the same thing in different ways: nobody knows how to price AI risk yet. Mark Jenkins from Carlyle, who oversees global credit and insurance, put it plainly. As credit investors, "we don't really know what the model is yet for AI." That's a remarkable admission from someone managing billions in corporate debt.

Roxana Mirica at Apax sees the fog lifting into two distinct paths. She's watching a "real bifurcation" form between credits facing "potential AI disruption" and those with cashflows that stay resilient "irrespective of what's happening in that AI cycle." Translation: lenders are dividing the world into companies that AI eats and companies it can't.

"The credit markets are becoming a live experiment in which business models survive automation."

This matters because credit markets price risk before equity markets admit there's a problem. When you lend money to a company, you care about one thing: will they pay you back. If AI might replace their revenue model in 18 months, that loan looks different. If they're selling something AI can't automate, that's a different bet entirely.

The split Apax sees isn't about tech versus non-tech. It's about:

  • Cashflow resilience: Can this business still generate money if AI agents do what their employees do now?
  • Moat durability: Does this company own something AI can't replicate, relationships AI can't forge, or regulatory protection AI can't bypass?
  • Capital intensity: Does succeeding in AI require billions in infrastructure, or can software eat you for free?

The Implication

If the people lending money to companies don't know how to model AI disruption yet, that uncertainty creates opportunity. Watch for credit spreads to widen between "AI-proof" and "AI-exposed" sectors in 2025. Boring businesses with physical moats, local presence, or human-trust requirements will get cheaper financing. High-margin knowledge work companies will pay more, even if their current financials look pristine.

For founders: your next funding round will include questions about agent resistance your investors don't know how to score yet. Have an answer ready. For workers: if credit markets think your company's cashflows might not survive AI, your equity might be mispriced too.

Sources

Bloomberg Tech | Bloomberg Tech