Listen to the latest episode of Credit Exchange with Lisa Lee
Published in London & New York
10 Queen Street Place, London
1345 Avenue of the Americas, New York
Creditflux is an
company
© Creditflux Ltd 2026. All rights reserved. Available by subscription only.
News Analysis
Debt markets preach caution on AI as bubble worries grow
by Lisa Fu & Kathryn Gaw
The lustre may be wearing off a rapidly growing sector that was once the darling of the credit market. Data centres, the infrastructure backing revolutionary artificial intelligence technology, are seeing their cost of debt move higher amid increasing questions about AI profitability.
The debt markets have helped drive the financing behind the AI boom to date, attracted to the steady revenues and future growth potential offered by AI firms. Hyperscalers — the likes of Google, Meta, Amazon and Microsoft — keep upping the capex spending forecast, and are willing to write trillions of dollars in the next few years to be among the winners of the AI arms race.
“AI infrastructure has become a strategic arms race, and debt markets are a natural source of funding given the enormous capital required,” said John Fekete, head of tradeable credit at Crescent Capital Group.
“I expect hyperscalers to remain frequent issuers, since the market can absorb the new supply with some concession likely.”
Pushback on prices
However, valuations have come under increasing scrutiny in recent months. The vast majority of AI companies are still not profitable, and there has been pushback on the use-case for many consumer-friendly AI products, as well as protests at data centre construction sites, led by local communities. AI data centre credits have widened more than 100bps since June, according to Barclays research.
“If the market starts to recalibrate the earnings power of these companies and what they’re spending now, it could get really interesting,” said TCW’s CIO of fixed income Bryan Whalen on the Credit Exchange with Lisa Lee podcast. “You could have a fundamental adjustment in cashflow and capital spend, and when you combine that with a market that is literally choking on the amount of supply, that could create a really interesting dynamic, not just for this sector, but for the market at large.”
The leveraged loan market is also being whipped by the change. Take AI-focused cloud computing company CoreWeave. At the end of July, a USD 2.6bn leveraged loan deal had to sweeten terms to attract enough investor interest to clear, a sharp contrast to an earlier loan that saw massive demand, according to sources familiar with the matter.
We’re being very cautious on AI
Melanie Hanlon
Head of US credit research
Napier Park Global Capital
While the difference in the collateral backing the CoreWeave loan played a role, so did worries about AI bond weakness and expectations of more volume down the road, said market sources.
“The significant drawdown in the equity market had to go over into some other markets, including the leveraged loan market,” said Adam Boyle, portfolio manager at Elmwood Asset Management. “This rockiness in pricing is probably mostly temporary in the space, especially since the leveraged loan market has not had a lot of supply this year. That being said, CLO managers are still being selective.”
An abundance of AI loan supply affords managers the luxury of being a little more choosy when it comes to loan selection.
“Certain investors, ourselves included, are going to put limits on how much you want to invest in this space,” said Melanie Hanlon, head of US credit research at Napier Park Global Capital. “That will also create a little bit more selectivity and potentially some more price movements, as people trade around in the sector.”
But the higher prices are providing no deterrent to the near-insatiable demand from hyperscalers such as Meta, which appear happy to accept less favourable terms on their financing as long as they are furthering their AI ambitions.
The push-pull between AI hyperscalers and the credit markets may be raising a few red flags for investors and managers who have already been burned by past events, such as the software sector sell-off earlier this year. But AI no longer belongs in just one sector, and the impact of an AI sell-off could have wide-ranging consequences.
Research is ongoing
“We’re being very cautious [on AI],” said Hanlon. “We’re just evaluating it on a name-by- name basis. I think it’s a very powerful tool, and I think it will change the way a lot of us live our daily lives and the way we do business, but it is evolving very quickly. We’re still doing our research and coming up to speed and forming our overall opinions.”
TCW’s Whalen agrees that AI is a transformational technology for humanity, which will be as impactful as the internet, but the capital spend and the momentum is like nothing seen before.
“There are just little things popping left and right,” Whalen said. “It feels like we’re all standing on this metaphorical stool of AI, and it’s got multiple legs. But if you look closely, some of these legs feel like they’re cracking a bit. And that usually is the beginning of the end of the euphoria.”
