Amazon, Alphabet Lead AI Debt Surge Linked to 20-Year High Rates
Hyperscaler debt issuance is a key driver behind 20-year high Treasury yields, with Amazon and Alphabet leading a $244 billion bond issuance surge. Off-balance-sheet debt has reached $1.65 trillion, raising concerns about leverage and future returns on AI investments.

*this image is generated using AI for illustrative purposes only.
Long-term interest rates have climbed to their highest levels in two decades, with hyperscaler debt issuance emerging as a primary driver alongside inflation and fiscal deficits. Apollo Global’s chief economist Torsten Slok identified “hyperscaler issuance” as one of three main factors behind 30-year Treasury yields reaching a 20-year high. This surge in borrowing by major technology firms is reshaping credit markets, forcing investors to assess whether artificial intelligence investments will generate returns sufficient to service the expanding debt load.
Amazon.com Inc. and Alphabet Inc. have led the borrowing spree over the past 12 months, while Meta Platforms Inc. and Oracle Corp. also rank among the largest issuers. Goldman Sachs highlighted this trend on July 20, noting that Microsoft, Amazon, Alphabet, Meta, Oracle, Nvidia Corp., and SpaceX collectively issued $244 billion in bonds this year. This volume represents 14 times the 2024 levels and more than double last year’s total, signaling a dramatic acceleration in capital raising across the sector.
The Scale of AI Borrowing
Palumbo Wealth Management described the situation as an “AI Debt Tsunami,” citing Morgan Stanley figures showing high-grade AI debt supply reached $270 billion by early July — more than double all of 2025’s total. Hyperscaler capital expenditures are projected at $650 billion to $800 billion this year and could exceed $1 trillion in 2027. This pace has outstripped free cash flow for most mega-tech firms outside of Microsoft Corp., necessitating significant external financing.
| Metric | Value | Source |
|---|---|---|
| High-grade AI debt supply | $270 billion | Morgan Stanley |
| Collective bond issuance (YTD) | $244 billion | Goldman Sachs |
| Off-balance-sheet debt | $1.65 trillion | Nikkei Asia |
| Projected AI spend through 2030 | $5.8 trillion | Goldman Sachs |
Off-balance-sheet debt, including sale-leaseback structures on data centers, has reportedly swelled to $1.65 trillion, according to Nikkei Asia, further obscuring true leverage. Goldman Sachs projects hyperscalers will spend a combined $5.8 trillion on AI infrastructure through 2030, a figure that will keep pressuring credit markets and long-term Treasury yields for years to come.
Market Reaction and Leverage Risks
The rapid increase in borrowing has altered market dynamics significantly. Hyperscaler leverage ratios doubled from 0.9x to 1.8x in roughly six months, according to Goldman Sachs. Concurrently, the bond market’s pain threshold has shrunk from $75 billion to just $25 billion, indicating reduced investor appetite for large-scale issuances at current yields. Credit-default-swap spreads on major tech issuers have widened sharply as investors assess the risk-reward profile of these massive capital expenditures.
What the Numbers Show
Despite the broader strain on the AI ecosystem, Goldman Sachs singled out Microsoft and Alphabet for retaining “fortress” balance sheets and strong free cash flow. This divergence suggests that while the sector-wide debt load is driving up systemic rates, individual company fundamentals vary widely. The concentration of issuance among a few dominant players highlights how their financing needs are disproportionately influencing global interest rate benchmarks, creating a feedback loop where higher rates increase the cost of servicing the very debt being issued to fund future growth.
How might the widening credit-default-swap spreads for major tech issuers signal a potential shift in investor sentiment toward the broader AI infrastructure sector?
What regulatory or accounting changes could emerge to address the growing opacity of off-balance-sheet debt, such as sale-leaseback structures, in hyperscaler financial reporting?
If AI-driven revenue growth fails to match the projected $5.8 trillion in infrastructure spending by 2030, which specific hyperscalers are most vulnerable to a debt servicing crisis?

































