Big Tech debt jumps to $350 billion as AI buildout fuels record bond issuance

Amazon, Alphabet, Meta, Microsoft and Oracle—the so-called hyperscalers—have stepped up borrowing sharply to finance AI infrastructure. The five companies have issued about $121 billion of new corporate bonds in 2025, far above the prior annual average of roughly $28 billion—effectively matching more than four years of typical borrowing in a single year. More than $90 billion of the 2025 total came in the final three months, underscoring how quickly funding needs accelerated. Over the past five years, the group's combined debt has roughly doubled to an estimated $350 billion. AI-linked issuance is becoming a meaningful slice of the broader market. Debt tied to AI now accounts for about 30% of net new issuance in the US investment-grade bond market, raising questions about how diversified high-quality credit exposure really is when so much new supply rests on a similar technology bet. Proceeds are flowing into the physical stack behind generative AI—data centers, servers and GPU clusters. By mid-2026, investor unease around longer-dated AI-related debt had become more pronounced. Some megacap issuers are already seeing sentiment pressure tied specifically to how they are financing large capital expenditure programs. A 2026 Chicago Fed analysis pointed to elevated bank commitments and tail-risk exposure to AI-adjacent industries, even as delinquency rates remained low as of late 2025. The central concern is less about any single issuer's solvency and more about concentration risk: the investment-grade market increasingly looks like a leveraged wager that AI infrastructure investment will pay off on a defined timeline. For investors holding diversified bond portfolios in 2025, exposure to data center economics may be larger than it appears.