Amazon Raises $25B More for AI. That's $89B in Bonds This Year Alone.

News Breakdown · FiscEdge Academy

Amazon priced an eight-part US bond offering on July 7 to raise at least $25 billion, its third trip to the debt market this year to fund AI infrastructure. Investor demand peaked at $62 billion before banks trimmed the final spread, and the deal still closed with roughly $41 billion in orders, more than 1.6x oversubscribed. Amazon told the underwriters managing the sale it does not plan to issue more debt for the rest of 2026.

That "no more debt" line is the tell. Stack this sale on top of Amazon's other two 2026 raises, a record $54 billion global offering in March ($37B in the US across 11 tranches plus a $16.8B euro-market debut, drawing $126 billion in orders) and a $10 billion Canadian-dollar sale in June, and Amazon has now borrowed roughly $89 billion in nine months, almost entirely to fund a 2026 capital expenditure budget it set at $200 billion, up from $131 billion in 2025. The dollar figure isn't the interesting part. The signal under the headline is that the biggest, most cash-generative tech company in the world is choosing to borrow at scale for AI rather than fund it entirely out of operating cash flow, and that choice tells you something about how the smartest capital allocators are pricing the AI buildout.

Why a company drowning in cash is still borrowing

Amazon generated well over $100 billion in free cash flow over the trailing year, so this isn't a company that's out of options. Borrowing at fixed rates, from 40 bps over Treasuries on the 3-year tranche up to 125 bps on the 40-year, locks in today's cost of capital for decades while keeping operating cash free for buybacks, working capital, and the flexibility to react if AI demand shifts. It also keeps equity intact: issuing $25 billion in new shares to fund data centers would dilute existing shareholders and signal the market doesn't trust the return on that spend. Debt says the opposite: management is confident the AI infrastructure will generate returns well above the coupon it just locked in.

The AI capex race is now a bond market story

Amazon isn't alone. Meta, Alphabet, Microsoft and Oracle have all tapped bond markets for AI-scale capex in the past year, and the "AI debt boom" is now large enough that credit desks track it as its own category alongside mortgage debt and sovereign issuance. For founders, the lesson isn't about bond mechanics, it's about what happens when your biggest potential customers and partners are all financing infrastructure with debt rather than equity: their unit economics on that infrastructure now include a fixed interest cost that has to be cleared before the spend is profitable, on top of the depreciation clock on the chips themselves. When a hyperscaler's AI division has to service $89 billion in new debt, expect the pressure to monetize AI features (via price increases, tighter free tiers, or aggressive upsells) to intensify, not ease. If your SaaS product runs on their compute, that cost structure eventually shows up in your bill.

The financing lesson that scales down to your business

Most founders will never issue a bond, but the logic Amazon just modeled scales down cleanly. Debt is the right tool when you have predictable, growing cash flow to service it and you want to avoid diluting ownership for an investment you're confident will pay back more than its cost of capital. Equity is the right tool when the outcome is uncertain and you need a partner who shares the downside. Too many early-stage founders default to equity for everything, because it's the instrument venture capital knows how to write checks with, without ever running the comparison against a venture debt facility or a revenue-based financing line. Amazon's CFO ran that comparison at a $200 billion scale and picked debt for the predictable part of the bet. You should run the same comparison at your scale before your next raise, which is exactly the kind of capital-structure math we walk through in financial modeling.

If you remember one thing

When cash-rich companies choose to borrow instead of spending savings or issuing equity, they're telling you they've done the math and expect the return to beat the interest rate, not that they're short on options. Before your next raise, ask the same question Amazon's treasury team just answered: is this spend predictable enough to finance with debt, or uncertain enough that it needs an equity partner who shares the risk? Getting that call right protects more of your company than any single funding round will.

---

We cover capital structure and raise strategy in FiscEdge's startup strategy course, and the cash-flow modeling behind debt-versus-equity decisions in financial modeling. If you're building a product on top of hyperscaler AI infrastructure, our building SaaS with AI track covers how those cost structures flow into your own unit economics. Browse the full blog for more breakdowns like this one. Follow @fiscedge for daily Business & AI analysis.