Meta is sitting on a cash pile — so why did it just borrow $30 billion?
In 2025 the most profitable companies in the world borrowed over $120 billion to build AI — many of them flush with cash. The reason is the heart of corporate finance: the cost of capital.
- A firm funds investment three ways — retained earnings, new shares, or debt — and of the money raised externally, debt is the cheaper source: lenders are paid first, accept a lower return, and the interest is tax-deductible.
- Equity is the most expensive money a firm can raise, and retained earnings are not a free third option — that cash carries the same opportunity cost as equity.
- WACC blends the cost of debt and equity into one hurdle rate; because debt is cheaper and tax-favored, a sensible amount of it can lower the WACC and make more projects worth doing.
- Borrowing while cash-rich buys financial flexibility, but leverage cuts both ways: past a point the risk of financial distress climbs and the WACC turns back up — and the interest is due whether or not the bets pay off.
In late October 2025, Meta sold $30 billion of bonds in a single day, the largest high-grade corporate offering of the year, a day after Mark Zuckerberg promised investors the company would spend even more aggressively on artificial intelligence. Oracle had just raised $18 billion of its own. Across the year, Amazon, Alphabet, Meta, Microsoft, and Oracle issued roughly $121 billion of US corporate bonds, more than four times their usual pace.
Several of these are among the most cash-rich companies on earth. Why borrow tens of billions when you are sitting on tens of billions? The answer is the central question of corporate finance: the cost of capital, and how a firm chooses to fund itself.
Get the next Insight in your inbox. One email when we publish — a market story or exam idea, made clear. No spam, unsubscribe anytime.
Opens at launch. Start free in the app →
Debt is the cheaper money
A company can fund investment three ways: internal cash (retained earnings), new shares, or borrowing. Of the money it raises externally, debt is the cheaper source. Lenders take less risk than shareholders, because they are paid first and on a fixed schedule, so they accept a lower return, and interest is tax-deductible, which lowers the effective cost further. Equity is the most expensive money a firm can raise, because shareholders bear the residual risk and demand to be paid for it. Retained earnings are not a free third option: that cash carries the same opportunity cost as equity, the return shareholders could have earned elsewhere.
WACC, and why cheap debt helps
The weighted-average cost of capital blends the cost of debt and the cost of equity into the single hurdle rate a firm must beat to create value. Because debt is cheaper and tax-favored, adding a sensible amount of it can lower the WACC, and a lower hurdle rate makes more projects worth doing. For a company convinced its AI data centers will earn far more than the mid-single-digit yield it pays on investment-grade bonds, borrowing is rational even with cash in the bank.
So why not just spend the cash?
Financial flexibility. Locking in cheap, long-dated debt while terms are favorable keeps the cash free for operations and buybacks, lets the firm move fast when an opportunity appears, and spreads the cost of a multi-year buildout across the years it will earn its return. Some of the financing went a step further, off the balance sheet entirely. Meta arranged about $27 billion of data-center funding with a partner so the debt would not sit on its own books, which is where capital structure meets financial-reporting quality.
What the borrowing signals
Leverage cuts both ways. More debt lowers the cost of capital up to a point; past it, the risk of financial distress climbs and the WACC turns back up. And if the AI bets disappoint, the interest is due regardless. A cash-rich company issuing bonds is not a contradiction. It is a bet that the return on the project will beat the after-tax cost of the debt. That is the cost-of-capital decision, made in public, at roughly four times Big Tech's normal borrowing pace.
Thirty questions, no signup — see exactly which topics still need the work.
Take the free diagnostic Start free in the appGet Insights by email
One note when we publish — the concept behind a market story, decoded for the exam. No spam; unsubscribe anytime.
The newsletter opens with the public launch. Prefer to start now? Open the free app →
More from Insights
Sources: Bloomberg — Meta sells $30 billion of bonds as AI frenzy fuels record orders · CNBC — Oracle's AI-fueled debt load has investors on edge