Big Tech Doubles Debt Load to $350 Billion in AI Spending Spree

Alphabet, Amazon, Meta, Microsoft, and Oracle have collectively piled on $350 billion in debt over five years to fund an unprecedented AI data center buildout — and credit analysts are starting to ask when the payoff arrives.

The five biggest spenders on AI data centers — Alphabet, Amazon, Meta, Microsoft, and Oracle — have doubled their combined debt load to $350 billion in the last five years, borrowing at a pace that is giving bond traders whiplash.

Hyperscalers issued about $121 billion in new debt in 2025 alone, more than four times their prior annual average of roughly $28–30 billion. Over $90 billion of that landed in the fourth quarter.

Oracle has become the poster child for leverage risk. Its debt now stands at roughly 2.5 times sales, and on Thursday S&P Global Ratings downgraded the company to the lowest investment-grade tier. Oracle took on an $18 billion bond issuance plus a $38 billion loan facility tied to AI expansion.

Amazon's free cash flow turned negative in the quarter ending March 31. One analyst model projects that aggregate hyperscaler free cash flow crosses below zero around Q3 2026. Morgan Stanley estimates roughly $1 trillion in off-balance-sheet purchase commitments and $800 billion in future leases that do not yet appear as formal liabilities — meaning the headline $350 billion understates the full exposure.

"I don't know that we know whether Amazon, Google, Microsoft and Meta are actually going to get a return on investment on this," said Jason Pompeii, a corporate debt analyst at Fitch Ratings. "It seems like a lot of demand hype that is very aspirational at this point."

The market is already voting. Only Alphabet's stock has outperformed the S&P 500 this year. Microsoft and Oracle shares are both down more than 20 percent.

The telecom debt boom of the early 2000s — billions borrowed for infrastructure before demand materialized — is the historical rhyme analysts keep returning to. Intel is the freshest warning: once dominant, debt-laden from expansion, wrong bet on manufacturing, and now projecting its third consecutive annual loss in 2026.

Goldman Sachs projects $5.3 trillion in tech capital expenditure through 2030. If AI demand meets the wager, these five companies lock in control of infrastructure for a generation. If it does not, the result would be the most expensive act of institutional faith in corporate history — with creditors already circling.

Sources: Mercury News, Business Report