[DigitalToday reporter Jae-won Choi] As U.S. Treasury yields surge to their highest level in 19 years, financing burdens are growing for companies that have expanded AI infrastructure by mobilising massive debt.

CNBC reported on Sept. 27 that the 10-year U.S. Treasury yield rose to around 5.17 percent. It jumped to about 5.22 percent during the week, the highest level since 2007. When Treasury yields rise, companies must offer higher rates to attract bond investors, raising the cost of new borrowing.

With AI infrastructure investment swelling rapidly, the impact of rising rates could be larger. JPMorgan projected in June that about $5.5 trillion would be invested in AI infrastructure by 2030, and that roughly $4.1 trillion of that would be financed with debt.

Funding conditions are also diverging by company. CoreWeave shares rose about 8 percent this week, but Oracle, which increased its reliance on the bond market during its AI expansion, fell 7 percent and is down about 30 percent so far this year. SoftBank recently raised $11.1 billion through a high-yield bond sale, with the interest rate on its 7.5-year dollar bond reaching 9.75 percent.

Some see the gap widening between large technology companies such as Amazon, Alphabet, Meta and Microsoft, which have high credit ratings, and neo-cloud companies. Large firms can raise funds at relatively low cost based on investment-grade ratings, but companies with heavy dependence on debt have limited capacity to absorb higher rates.

CoreWeave disclosed that, based on its floating-rate debt as of the end of June, a 1 percentage point rise in rates could increase quarterly interest expense by about $30 million. AI demand remains strong, but if high interest rates persist, funding capacity is expected to become an important variable in the competition to invest in data centres predicated on massive borrowing.