Hyperscalers flood bond market with $93 billion in AI debt

Hyperscalers including Alphabet, Amazon, Meta, and Microsoft issued $93 billion in debt in 2025 to fund artificial intelligence infrastructure. This volume dwarfs the $35 billion annual average recorded between 2020 and 2024. The companies are using the capital to purchase chips and construct data centers and power plants.

Fixed income experts describe the current environment as a generational opportunity. With the 10-year Treasury yield hovering near 5% and the Federal Reserve expected to hold rates steady at its upcoming meeting, investment-grade corporate bonds are offering attractive returns. Steven Purdy, co-head of global credit at TCW, noted that investors rarely get the chance to buy AA-rated companies at such favorable yields.

The VettaFi AI Hyperscaler Bond Index, which tracks debt from major tech firms, yielded 6.51% at the end of September. Joseph Matus of Cerity Partners called this a unique window where above-average credit quality meets above-average yields. He said investors are currently giving these tech giants a pass on their large-scale issuance because their AI projects are proving profitable.

However, cash flow pressures are emerging. Meta reported negative free cash flow for the first time since its 2012 IPO. While Purdy stated the debt levels are not dangerous, he acknowledged they are a significant amount. Investors remain cautious, particularly regarding CoreWeave, whose bonds plummeted in October as spreads in its credit default swaps widened due to concerns over its debt-fueled expansion.

Oracle offers another point of caution. S&P Global Ratings downgraded the company's debt to the lowest tier of investment grade in July. Jordan Chalfin of CreditSights described Oracle as the most important investment grade tech credit to watch, noting it trades more like a high-yield credit due to investor concern. Analysts will closely monitor earnings reports from these tech giants this week.

Future issuance is expected to remain high. Brian Barnhurst of PGIM Credit predicts continued debt linked to the AI ecosystem over the next year. JPMorgan forecast in June that over $4 trillion in AI-related debt could be issued through 2030. To fund these projects, companies are diversifying their financing. Meta announced a plan to sell up to $95 billion in stock offerings alongside debt to fund its AI expansion.

ProShares has filed for an exchange-traded fund focused specifically on AI hyperscaler debt, expected to launch by early 2027. The fund will track a VettaFi index designed to isolate debt from core AI infrastructure companies. Todd Rosenbluth of TMX VettaFi described the proposed fund as a way to marry high credit quality with meaningful yield generation.

Advisors recommend diversification within the bond market. Joanna Gallegos of BondBloxx pointed out that BBB-rated bonds, the lowest investment-grade tier, are yielding over 6%. Bob Shea of Dynasty advised a focus on active management over passive index funds in the current rate environment. He highlighted the iShares Flexible Income Active ETF, which has a 30-day SEC yield of 5.65% as of October 23.

Edison Byzyka of Credent Wealth Management suggested extending portfolio duration to seven or eight years on average. In the investment grade space, hyperscalers now represent over 20% of the 10-year plus index. More than 40% of hyperscaler issuance in 2026 has occurred in maturities of 20 years or more. The Vanguard Long-Term Corporate Bond ETF offers a 30-day SEC yield of 6.39% with an average duration of 11.6 years.

Byzyka predicts equity markets will correct, making junk bonds more volatile than high-quality corporates. He is seeking bonds rated BBB or better to secure yields between 6% and 6.5% without undue risk. Matus cautioned that while hyperscaler bonds offer value, investors must consider if the extra yield is worth the loss of diversification against risks in the AI equity story.