AI Hyperscalers Fuel Rise in Long-Term Borrowing Costs
Federal Reserve Chairman Kevin Warsh attributed the recent rise in long-term borrowing costs to several factors, including the growing debt of AI hyperscalers. The 10-year Treasury yield has surpassed 5%, a psychologically significant level that reflects increased competition for capital.
The hyperscalers, which include Amazon, Microsoft, Alphabet, Meta, Oracle, and Coreweave, have issued $121 billion in U.S. corporate bonds so far this year, compared to an average of $28 billion per year between 2020 and 2024. Morgan Stanley estimates that AI-related global debt will reach nearly $570 billion by the end of 2026.
Warsh also cited economic growth and geopolitics as contributing factors to the rise in long-term yields. However, some experts have argued that the impact of AI on treasuries is overstated, with PIMCO attributing it largely to the conflict in Iran and a repricing of Warsh's willingness to hike rates.
The Federal Reserve has set up an internal task force to study the implications of AI for future policy. Chairman Warsh expressed optimism that AI will eventually expand the economy's capacity and be disinflationary, but acknowledged that there are also risks associated with its development.