US Market: AI credit surge tests lenders as future revenues remain uncertain

The artificial intelligence boom is increasingly spilling into the riskiest corners of U.S. credit markets, where lenders are demanding higher returns to finance companies with uncertain future earnings, Reuters reported.

AI-related issuance by low-rated firms has reached $88 billion this year, according to Goldman Sachs, with most of the borrowing coming from U.S. issuers. In the first 11 months of 2025, AI-related issuance in leveraged finance, primarily through junk bonds and loans, stood at just $20 billion, according to Neuberger Berman data cited by Reuters.

The surge in borrowing is prompting investors to scrutinize the financial assumptions behind less-established AI companies, including their revenue forecasts, collateral values and ability to service growing debt.

Reuters reported that the scrutiny comes as higher-rated AI companies ramp up borrowing and a selloff in the U.S. Treasury market pushes yields higher across the board.

For credit investors, the concern is that debt offers limited upside participation if an AI project succeeds, while losses can become substantial if projected cash flows fail to materialize.

Reuters reported that lenders are therefore seeking greater compensation for taking on the risks associated with uncertain revenues and rapidly changing technology.

Read more: US stocks: Nasdaq rises while S&P, Dow dip as softer inflation cools Fed hike bets

AI CREDIT DEMAND REMAINS SELECTIVE

Despite the sharp increase in issuance, overall investor appetite for lower-quality AI-related credit remains limited, Reuters reported. Buyers in leveraged finance have generally gravitated toward double-B-rated companies, which sit just below investment grade.

Demand has been strongest for borrowers with predictable revenue streams, long-term contracts, tangible assets and established customer bases, fund managers told Reuters.

Data centers, many of which fall into the higher-quality segment of the market, have accounted for a significant share of AI-related issuance. Erin Brown, head of leveraged finance at BNP Paribas, told Reuters that data centers have provided much of the new supply supporting the high-yield market.

AI infrastructure supply in the high-yield market has reached $40 billion so far this year, compared with $12 billion for all of 2025, according to BNP Paribas data cited by Reuters.

The increased issuance has helped support an otherwise subdued high-yield market. Reuters reported that high-yield volumes were broadly flat year over year and would have been materially lower without new-money issuance from data centers.

Borrowing costs rise sharply as issuers move further down the credit spectrum. Analysts cited by Reuters said even BB+ borrowers are paying yields of roughly 9% to 10%, while lower-rated companies could face borrowing costs of as much as 14% to 15%.

SoftBank Group, which carries a BB+ rating, recently raised debt at yields ranging from 8.625% on 3.5-year notes to 9.75% on seven-and-a-half-year bonds, according to a company filing cited by Reuters.

Read more: Global Market Today: Asian shares mixed as bond yields, oil keep investors on edge

CLO MANAGERS BECOME MORE CAUTIOUS

The growing skepticism is particularly relevant for collateralized loan obligation managers, which are among the largest buyers of leveraged loans.

Unlike investment-grade investors that can absorb large debt offerings from highly rated technology companies, leveraged-finance investors face portfolio restrictions that limit their exposure to riskier borrowers, Reuters reported.

AI companies requiring substantial upfront investment can therefore become harder to finance if leverage increases, cash outflows persist or credit ratings deteriorate.

CLOs generally purchase portfolios of leveraged loans and finance those investments by issuing their own securities to investors. Their demand is therefore an important source of funding for the leveraged-loan market.

Elizabeth Templeton, senior product manager for fixed income and multi-asset indexes at Morningstar, told Reuters that CLO managers are becoming more cautious about certain AI-related borrowers and are conducting greater due diligence before committing capital.

Read more: Global Market Today: Asian shares mixed as bond yields, oil keep investors on edge

ZENITH ARC DEAL HIGHLIGHTS RISKS

The market's caution has already been reflected in the performance of some AI infrastructure debt, Reuters reported.

Zenith Arc, a special-purpose project company, sold $2.25 billion of five-year senior secured notes in August to finance an Oklahoma data center leased to Jane Street. The bonds were issued at 99.50 cents on the dollar and carried an 8.875% coupon.

The securities initially fell almost three points to a bid price of 96.75 after issuance, according to analysts cited by Reuters. By late August, Pender Fund Management said in a letter to investors that the bonds had fallen more than seven points below their issue price.

Zenith Arc had no publicly available contact details, Reuters reported.. A representative for Coatue, the investment firm behind Next Frontier, a joint-venture partner in the Zenith Arc project, said the firm had no comment.

The episode highlights the challenges facing AI-related borrowers as financing moves beyond established technology companies and into projects whose revenues depend on future infrastructure development and demand.

INVESTORS SEEK PROOF OF FUTURE CASH FLOWS

The broader AI financing boom is also unfolding against a tougher backdrop for borrowers. Rising U.S. Treasury yields have increased the benchmark cost of capital, putting additional pressure on corporate debt markets, Reuters reported.

At the same time, AI companies still need enormous amounts of capital to build data centers, acquire computing capacity and develop increasingly expensive models. Much of the expected revenue from these investments remains dependent on future demand.

That creates a difficult balance for credit investors. They must assess whether projected AI revenues will be sufficient to support the debt being accumulated today, while also accounting for construction delays, technology shifts, customer concentration and the possibility that spending will rise faster than cash generation.

Lotfi Karoui, multi-asset credit strategist at PIMCO, wrote in a recent research note that debt investors face an asymmetric risk profile, with returns largely limited to contractual payments while potential losses can increase with excessive leverage, project delays and technological changes. Reuters cited the research in its report.

Capital remains available for AI infrastructure, but the terms are becoming more demanding as lenders move further down the credit spectrum. For borrowers without established cash flows, demonstrating credible future revenue is increasingly important to accessing debt markets at manageable costs.

The shift marks an important change in the financing of the AI boom: Investors are increasingly evaluating not only the growth potential of artificial intelligence, but also whether the cash flows generated by that growth will arrive soon enough and at a sufficient scale to support the debt being raised, Reuters reported.

(Disclaimer: This article is based on inputs from agencies. These do not represent the views of The Economic Times)
Add as a Reliable and Trusted News Source Add Now!
(You can now subscribe to our ETMarkets WhatsApp channel)