On Wednesday evening, June 10, Oracle announced it had spent $55.7 billion building AI data centres in the past year alone — and that it planned to raise another $40 billion in debt to keep going. By Thursday morning, its stock had dropped 12%, wiping roughly $72 billion off its market value in a single session. Somewhere in Cupertino, someone probably slept very well.
This was the week the AI spending story changed tone. Not dramatically, not with a crash — but with a creeping, unsettling question that investors had been quietly sitting on for months: at what point does all this spending on AI infrastructure actually start paying back? Oracle's results were good. Revenue was up 21%. Cloud computing, the business of selling companies access to computing power over the internet, grew nearly 50%. And the market still punished the stock, because the spending plan that came with those results was simply staggering.
Morgan Stanley expects AI-related global debt issuance to more than double to nearly $570 billion in 2026, with hyperscaler spending — that is, spending by the world's largest technology companies on computing infrastructure — set to exceed $1 trillion by 2027. This is a number so large it barely registers. It's the kind of number that, once it does register, makes you wonder who's going to be left holding the bill.
Apple, as it happens, has not been in the room where that bill is being run up.
