It's the worst time in years to invest in AI credit markets: 'Almost no upside and plenty of downside'

By Jules Rimmer

Andromeda analysts are skeptical about the returns for buyers of AI credit

Researchers at Andromeda are worried about the uncertain return on investment for investors who are financing the AI capex bonanza.

The research team at Andromeda Capital Management isn't holding back: Not only do they think credit markets for artificial-intelligence companies are complacent, but they also assess the current environment as the worst entry point into those markets in years, with historically tight spreads leaving "almost no upside and plenty of downside."

Spending by hyperscalers is expected to run to $1.2 trillion in 2027, according to Bank of America analyst Vivek Arya, and that can no longer be funded out of free cash flow or equity. Most companies that aren't public have to borrow money to expand, and that often means turning to the private-credit market.

In a research note published Tuesday titled "AI Credit: Heads I Win, Tails You Lose," the global investment-management firm expressed its bearish views on this market, but the authors took pains to qualify that stance.

"We are not sceptical of AI as a technology - we are users ourselves," wrote a team led by chief investment officer Alberto Gallo. "However, we are sceptical of the returns" for buyers of AI credit, which incorporates bonds, loans and other kinds of debt issued by companies in the AI sector. They cited the capped return on capital, the exposure to uncertain returns on invested capital and the lack of participation in the upside that's available to equity investors.

Credit-spread annualized returns versus entry spread.

What complicates and darkens the picture from Andromeda's perspective is that a significant chunk of AI credit is concentrated in private markets. This sector has enjoyed phenomenal growth in the last decade and now stands at about $3.5 trillion. That's almost three times the size of the U.S. subprime-mortgage market before the global financial crisis in 2008.

Tech stocks make up 40% of private credit, with software-as-a-service accounting for 25% of that. For Gallo and his team of Aditya Aney, Tao Pan and William Koor, "both the AI and private credit universes face significant volatility."

Private-credit exposure to AI-vulnerable sectors.

What troubles the team is that since problems arose surrounding redemptions at the end of the first quarter, the rise in risk assets has pushed the topic of private-credit risk off the radar. But the Andromeda note highlights that losses are mounting. The distress ratio of software loans is over 25% and one-tenth of all credit loans are now pay-in-kind - whereby issuers pay creditors with new loans instead of interest. This suggests some kind of cash crunch.

Because returns on the capital that AI companies have deployed are uncertain, the chances of capital misallocation are high. The fact that many of those loans or bonds are held in private vehicles compounds the problem of transparency and liquidity. For the Andromeda team, the bulk of any potential losses is likely to be borne by the life-insurance companies that own as much as $35 trillion of these assets. Andromeda also frets about the circular exposure created by the insurers, banks and private-equity vehicles.

CoreWeave (CRWV), the Apollo-owned Athene and ION Group are among the companies whose credit Andromeda points to as looking vulnerable.

-Jules Rimmer

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

08-05-26 1044ET