Big Five Tech Giants' Debt Doubles to $350 Billion Over Five Years
N.R. Finch
Alphabet, Amazon, Meta, Microsoft and Oracle have added roughly $350 billion in new debt over five years, doubling interest costs past $10 billion, as the AI data-center arms race turns Big Tech from cash cows into leveraged builders — and the bond market is starting to push back.
$350 billion in new debt — where is the money going?
Total debt across the five companies has doubled in five years. Their combined capex commitment for this year alone: $725 billion, mostly for AI data-center construction and Nvidia chip purchases.
Funding comes from three sources: internal cash, new borrowing, and off-balance-sheet arrangements — Meta, for instance, has moved some liabilities off its books to keep the balance sheet looking cleaner.
This means → Big Tech is no longer just "investing what it earns." These companies are borrowing at scale to bet on AI — wagering that future revenue will cover today's interest bill.
Interest costs have doubled — who can still handle it, and who is straining?
Combined interest expense topped $10 billion last year, more than double the 2019 figure. But the pain is unevenly spread: Google's free cash flow — operating cash flow minus capex — hit $64 billion in Q1, dwarfing its interest burden.
Amazon is a different story. Its Q1 free cash flow turned negative, meaning operations no longer generate enough cash to cover capital spending. The gap must be filled by borrowing or drawing down reserves.
Oracle is under the most pressure: its debt is roughly 2.5× its 2025 revenue. S&P Global Ratings downgraded it this week to the lowest investment-grade tier. In plain terms = one more notch down means "junk" status, and borrowing costs would jump sharply.
Is the bond market already saying "enough"?
Amazon's $25 billion bond offering this week met an unusually tepid reception, according to people familiar with the matter cited by Bloomberg.
This reflects a bond-buyer tolerance for Big Tech's open-ended borrowing that is nearing its limit — not unwillingness to lend, but hesitation at the speed and scale.
This means → if the debt market tightens, Big Tech must either slow the buildout or accept higher financing costs. Both paths squeeze margins.
CEOs say "we're highly confident" — do analysts believe them?
Amazon CEO Andy Jassy said he has "high confidence" AI investments can be monetized, citing AWS customers' advance commitments to new capacity. Mark Zuckerberg told Bloomberg that AI compute demand keeps exceeding supply, adding: "That gives us a lot of confidence."
DA Davidson analyst Gil Luria pushed back: the corporate logic is that expected AI returns exceed borrowing costs, so the buildout makes sense — "but you can tell investors aren't comfortable."
Fitch analyst Jason Pompeii was blunter: "I'm not sure we really know whether Amazon, Google, Microsoft and Meta can get a return on these investments. Right now, a lot of the demand looks more like hopeful hype."
Stock prices are already voting — who wins, who loses?
Year to date, only Alphabet has outperformed the S&P 500. Microsoft and Oracle shares have each dropped more than 20%.
This means → the equity market is already sorting "can afford to spend" from "has overspent" — cash-rich Google earns a premium, while more leveraged Oracle and Microsoft are being punished.
Quarterly earnings are due later this month; capex size and funding mix will be the key focus for bond investors. Put simply = when AI investment starts producing visible returns will determine whether this debt-fueled expansion can keep the market's backing.
Content is for reference only, not financial advice.