Growing AI exposure could pressure future debt deals
TLT•Growing AI exposure could pressure future debt deals
It’s not only the U.S. government that issuers tied to the artificial intelligence buildout must compete against in debt markets. They are also increasingly competing for limited portfolio space, a dynamic that could force future borrowers to offer wider spreads and cheaper valuations, said Tim Musial, head of fixed income at CIBC Private Wealth U.S., in a Reuters interview.
Musial said many investors already have significant exposure to hyperscalers and AI infrastructure through corporate bonds and data-center financings, making new purchases increasingly a question of which existing holdings to trim.
“To potentially invest in a new issuance from someone, they have to maybe sell something on the other side,” he said, telling Reuters that may mean reducing Treasury holdings or other corporate bond positions to make room.
Demand remains strong, but portfolio managers are becoming more selective as exposure grows.
But after heavy issuance tied to data-center expansion and other infrastructure projects, investors are paying closer attention to portfolio concentration, relative value and risk management. The surge in AI-related bond issuance, alongside heavy government borrowing, has helped push Treasury yields higher as investors demand greater compensation to absorb the growing supply.
“I think in 2026, many market participants have seen their buckets filled to some degree with hyperscalers,” Musial said. “If you think someone is going to continue to bring new debt to the market throughout the rest of this year, and in 2027, I think they should expect to offer those at cheaper valuations and wider spreads.”
Molly Brooks, U.S. rates strategist at TD Securities USA, also sees limits to investors’ appetite for additional debt, and said issuers may have to reconsider their reliance on bond markets.
“There’s going to be a trade-off,” she said. “If you look at the cost of raising debt versus equities right now in the hyperscaler space, it actually costs more to issue in debt versus equities.”
