Goldman Sachs reported this week that six major tech hyperscalers — Amazon, Alphabet, Meta, Microsoft, Oracle, and Nvidia — have already issued ~$194 billion in investment-grade bonds in 2026, on track to hit $250 billion by year-end, more than double the $108 billion they issued across all of 2025. Borrowing spreads are widening as supply floods the market.
Amazon, Meta, Alphabet, Microsoft, Oracle, Nvidia. The most cash-generative companies on earth. And they're flooding the bond market.
Here's the question a CFO would ask first: why are companies swimming in operating cash flow borrowing at all? The answer is in the math. These six firms are on track to spend $600–725 billion on AI capex in 2026 alone. Even $577 billion in combined operating cash flow doesn't fully cover that. So the shortfall goes to debt markets — and Goldman Sachs now expects hyperscaler bond issuance to hit $400 billion in 2027.
This is a capital structure decision, not a desperation move. A CFO looks at the weighted average cost of capital (WACC — the blended cost of a company's debt plus equity, weighted by how much of each you use). Debt is cheaper than equity because bond investors get paid before shareholders, and interest is tax-deductible. As long as the return on invested capital (the AI data centers, the chips) beats the WACC, adding more debt is rational. It's value-creating leverage.
The stress signal to watch: spreads. For Amazon and peers, the median spread on short-term bonds has already widened from 30 to 40 basis points over risk-free rates. The market is saying: we'll lend, but it'll cost you more. If AI returns disappoint, the leverage that looked smart on the way up becomes painful on the way down.
Capital structure is never just about today's interest rate. It's a bet on tomorrow's return.
📚 Learn the concept: Capital Structure & WACC
▶ Play the 90-second CFO game All daily posts