Lower-Rated AI Borrowers Face Financing Hurdles as U.S. High-Yield Market Demand Concentrates on Higher-Quality Assets
nashnova research
US AI-related leveraged finance has surged to $88 billion this year, but lower-rated borrowers are being turned away — the asymmetric payoff of debt leaves risk squarely with creditors.
$88 billion poured in — where did it go?
Goldman Sachs data shows lower-rated AI leveraged issuance hit $88 billion in 2026; the first 11 months of 2025 totalled just $20 billion (Neuberger Berman data).
This means → volume quadrupled, yet risk appetite did not follow — more money, not more courage.
Buyers cluster around double-B rated issuers with long-term contracts, hard assets, and stable client bases. Data centres absorb the bulk of new supply.
BNP Paribas puts AI infrastructure supply in the high-yield market at $40 billion this year, versus $12 billion for all of 2025.
How expensive is the borrowing?
Even a relatively high-rated BB+ issuer pays steep rates — SoftBank priced bonds this month at 8.625% (3.5-year), 9.25% (5.5-year), and 9.75% (7.5-year).
This means → those yields normally correspond to lower-rated borrowers; SoftBank is paying that price anyway.
For lower-rated AI firms, borrowing costs can reach 14% to 15%.
In plain terms = the lower the rating, the higher the price; at the bottom tier, every $100 borrowed costs nearly $15 a year in interest.
Why are creditors so cautious?
Larry Holzenthaler, senior portfolio manager at Catalyst Funds, names the core tension: "If everything goes right, you just get par back; if something goes wrong, the creditor bears the loss."
In plain terms = win and you collect principal plus coupon; lose and you may lose everything — capped upside, uncapped downside. That is the "asymmetric structure."
PIMCO strategist Lotfi Karoui wrote in a recent note that returns come mainly from coupons and principal, while risks span high leverage, project delays, and rapid technology turnover.
This reflects a fundamental creditor concern: AI projects burn cash before generating any, yet all the risk already sits on the balance sheet.
Why are CLO managers tightening standards?
CLOs — collateralised loan obligations, vehicles that buy baskets of loans and issue their own securities to investors — are the largest structural buyer in the leveraged-loan market. Their stance directly shapes liquidity.
Many AI borrowers need heavy upfront spending before stable cash flow arrives — leverage is rising, cash is flowing out, and downgrades loom.
Elizabeth Templeton, a senior fixed-income product manager, told Reuters: "CLO managers are becoming increasingly cautious on certain names, checking every detail before investing."
This means → the biggest buyer is reluctant to take on the paper, and liquidity for lower-rated AI loans is narrowing.
Can the boom last?
The test comes down to one question: can AI capital expenditure ultimately convert into stable cash flow sufficient to cover the cost of debt?
The current picture is split — higher-rated borrowers with contracts and assets fund smoothly; lower-rated, cash-burning firms are shut out.
This reflects a high-yield market that is using price and access thresholds to tier AI risk: not refusing to lend, but lending only to those with visible cash flow.
市场有风险，内容仅供研究参考，不构成投资建议。