AI Bond Supply Is Raising the Cost of Capital for Small-Caps

Amazon, Microsoft, Meta and Oracle have helped turn artificial-intelligence spending into a bond-market event, not just a technology investment cycle. The five largest hyperscalers are expected to spend more than $1 trillion on AI-related capital expenditure across 2025 and 2026, according to the Bank for International Settlements. Much of that buildout is now being funded with debt, project finance and private-market structures.

That matters for small-cap stocks because credit is not an infinite pool. When investment-grade issuers with fortress balance sheets sell enormous volumes of bonds at yields investors consider attractive, they absorb demand that might otherwise reach weaker borrowers. Smaller companies then face a less forgiving market: wider spreads, tougher covenants and shorter maturities. The Federal Reserve said in July that financing remained accommodative for larger businesses but was still somewhat restrictive for many small businesses.

The pressure is most acute among firms that rely on refinancing rather than internally generated cash. A company with $100 million of floating-rate debt does not need a dramatic rise in Treasury yields to feel the squeeze. A modest increase in its credit spread can consume millions of dollars in annual interest expense, forcing management to cut hiring, delay investment or issue stock at a depressed valuation. Private credit may provide a lifeline, but it often comes with higher rates, tighter collateral requirements and less flexibility.

The risk is not that Amazon or Microsoft suddenly become distressed borrowers. It is that a crowded AI financing trade changes the market’s treatment of everyone else. The BIS has warned that AI-linked debt is creating stronger connections between hyperscalers, nonbank lenders and private-credit vehicles, while a repricing of equity risk could tighten corporate credit more broadly. That transmission channel runs straight through the Russell 2000, whose companies generally have smaller cash cushions and less access to cheap unsecured funding.

So far, the market is not signaling a generalized credit panic. On Sept. 14, the ICE BofA single-B high-yield spread stood at 2.79 percentage points, while the BB spread was 1.56 points. But those relatively calm readings can obscure the divide between strong borrowers and marginal ones. If AI returns disappoint, data-center utilization falls short or investors simply demand more compensation for duration, small-caps may discover that the real cost of the AI boom is their own cost of capital.