AI Infrastructure Borrowing Is Meeting a More Selective Bond Market

Bond investors are becoming more selective about the companies financing the artificial intelligence buildout. Reuters reported September 22 that AI-related corporate borrowers are facing wider spreads and pricing concessions, while some traditional industrial, insurance, and pharmaceutical issuers continue to draw strong demand.

The pressure reflects the scale and uncertain timing of infrastructure spending. Goldman Sachs data cited by Reuters put expected gross debt issuance by hyperscalers at $420 billion in 2027, 60% above estimates for 2026. Investors are weighing that supply against the still-uncertain returns on data centers, chips, and related capacity.

A risk premium for volume

Reuters reported that spreads on AI-related issuers were around 115 basis points, compared with 78 basis points for the broader investment-grade market. Portfolio managers interviewed for the story said the difference largely reflects supply, concentration, and uncertainty about investment payback—not a claim that the largest technology companies are near default.

That distinction matters. The bond market is still open to AI firms, but a strong balance sheet may no longer guarantee the same easy financing terms when several large borrowers need capital at once. Investors also have to consider exposures bundled through data-center financing vehicles and related companies.

For AI businesses, the signal is practical: clear borrowing plans and credible explanations of how capacity will earn a return could become more important. For the broader industry, financing costs are another check on the assumption that every announced infrastructure project will be funded on equally favorable terms.

Source: Reuters, September 22, 2026.