AI Bonds Could Push Treasury Yields Higher

The race to fund artificial intelligence has moved into bonds, where tech giants now compete with governments for investors.Key Takeaways: Roughly one-third of investment-grade corporate supply through August 31 was long-dated, with AI borrowers playing an outsized role. T. Rowe Price expects heavy supply to pressure hyperscaler bonds well before any decline in credit quality. Investment-grade data center financings pay for complexity, while AI lab bonds remain a possible future addition. Governments worldwide have been issuing more debt since the pandemic hit in 2020. AI borrowing adds to that “increasingly crowded arms race,” according to Arif Husain, head of global fixed income at T. Rowe Price. That supply should eventually push U.S. Treasury yields higher, Husain wrote in a September report. Debt from hyperscalers, the biggest cloud computing firms, should also offer more yield over government bonds as credit spreads widen. AI could drive about half of U.S. dollar credit market growth through 2029, said Mark Stodden, a T. Rowe Price tech credit analyst. Much of that borrowing is long-term, according to Steve Boothe, the firm’s head of global investment-grade fixed income. About one-third of this year’s investment-grade corporate supply through August 31 was long-dated, with AI issuers playing an outsized role. “Long duration investors are being asked to absorb heavy Treasury supply at the same time that hyperscalers and other AI-related borrowers are issuing large amounts of long-dated debt,” Boothe said. “The clearing price for duration must adjust.” Put simply, buyers will likely demand higher yields to hold long-term bonds. Still, most hyperscalers have “some of the strongest balance sheets in the corporate market,” Boothe said. He expects heavy supply to weigh on their bonds well before any credit problems surface.How Active ETFs Approach AI BondsA portfolio could build sizable AI exposure without ever deciding to, simply by owning individually attractive deals. “The question is not how much AI to own in absolute terms, but how much active AI risk a portfolio should take relative to its benchmark,” Stodden said. See more: Beyond AI Stocks: As Markets Broaden, Look to Active ETFs T. Rowe Price’s two active bond ETFs answer that question in different ways. The T. Rowe Price QM U.S. Bond ETF (TAGG ) seeks to match the risk profile of the Bloomberg U.S. Aggregate Bond Index, or the Agg, according to the firm. Unlike an index fund, which takes on more AI debt as issuance grows, TAGG doesn’t have to copy the Agg’s weights. As of September 30, TAGG had a 30-day SEC yield of 5.16%, T. Rowe Price data show. It held $2.7 billion in net assets as of October 14 and charges a 0.08% expense ratio. T. Rowe Price’s two active bond ETFs answer that question in different ways. The T. Rowe Price QM U.S. Bond ETF (TAGG) seeks to match the risk profile of the Bloomberg U.S. Aggregate Bond Index, or the Agg, according to the firm. Unlike an index fund, which takes on more AI debt as issuance grows, TAGG doesn’t have to copy the Agg’s weights. As of September 30, TAGG had a 30-day SEC yield of 5.16%, T. Rowe Price data show. It held $2.7 billion in net assets as of October 14 and charges a 0.08% expense ratio. T. Rowe Price’s two active bond ETFs answer that question in different ways. The T. Rowe Price QM U.S. Bond ETF (TAGG) seeks to match the risk profile of the Bloomberg U.S. Aggregate Bond Index, or the Agg, according to the firm. Unlike an index fund, which takes on more AI debt as issuance grows, TAGG doesn’t have to copy the Agg’s weights. As of September 30, TAGG had a 30-day SEC yield of 5.16%, T. Rowe Price data show. It held $2.7 billion in net assets as of October 14 and charges a 0.08% expense ratio.Looking Beyond Hyperscaler DebtBoothe sees more interesting opportunities away from straightforward hyperscaler bonds. He pointed to investment-grade data center and infrastructure financings “where investors are being paid for complexity.” Those deals require a close look at tenant quality, contract terms, construction risk and power availability. The T. Rowe Price Total Return ETF (TOTR ) is a more flexible option. It can invest well beyond the Agg, from Treasuries and corporate bonds to bank loans and asset-backed securities, according to T. Rowe Price. The fund also aims to spread risk more evenly than its benchmark. TOTR’s 30-day SEC yield stood at 5.58% as of September 30, the firm reported. It had $569 million in net assets as of October 14, with a 0.31% expense ratio. Stodden flagged one risk. If computing power supply eventually outstrips demand, bonds from infrastructure providers could suffer, he said. Even then, data center deals with high-quality tenants should hold up. One area T. Rowe Price would investigate adding is bonds issued by AI labs. The firm has limited exposure today but could build positions if input costs fall while demand stays high, Stodden said. For more news, information, and analysis, visit our Active ETF Content Hub.