In a significant wave of investment, U.S. companies have issued over $200 billion in bonds this year to fund extensive artificial intelligence (AI) infrastructure projects. Analysts express concerns that this surge in borrowing may overwhelm the corporate debt market and pose new risks for credit investors. Previously, major tech firms primarily funded their data center infrastructure through robust earnings and strong balance sheets. However, as the costs of developing colossal AI data centers mount—and with returns projected to be years away—these companies have increasingly turned to debt markets.
A recent example is Meta, which raised $30 billion in bonds this week, attracting approximately $125 billion in orders. This level of demand represents the highest ever recorded for a U.S. investment-grade corporate bond. Goldman Sachs has highlighted that companies such as Meta, Alphabet, and Oracle have accounted for more than a quarter of all new corporate debt supply in the U.S. for 2023, indicating a striking trend in AI-related bond issuance.
In the same vein, Oracle secured $18 billion in bonds in September, also witnessing strong demand. This financing is earmarked for the development of data centers leased to power OpenAI. Experts are raising alarms that companies committing to substantial builds for largely single customers may face challenges in raising capital, especially as the recovery may not yield immediate returns.
Furthermore, analysts from Barclays have noted that the current issuance wave, while not significantly impacting market supply thus far, could dramatically alter the landscape. The firm warned that soaring capital expenditure (capex) could lead to an unprecedented surge in issuance, amplifying investment risks for lenders who may face systemic concerns if invested funds do not generate expected returns.
Senior fund managers are increasingly concerned about concentration risks and the sustainability of AI-related spending. They point out that the long duration of tech bonds may heighten the sensitivity of the broader US investment-grade credit market to fluctuations in interest rates.
Goldman Sachs anticipates 2025 will be a pivotal year for AI-linked bond issuance and predicts continued fundraising efforts in 2026 to support data centers and related energy infrastructure. This activity coincides with a strong demand for corporate credit, seen earlier in the year when corporate credit spreads hit their lowest levels this century.
Investment discussions within the industry reflect caution, as some fund managers caution against relying excessively on capital markets to finance capital expenditure, especially if concerns about a potential AI “bubble” materialize. If such a bubble results in a significant amount of bad debt, the negative implications for the financial system could be severe. Additionally, the use of private debt in these ventures raises transparency concerns, complicating the overall financial landscape.
