The Recent Borrowing Spree by AI Giants Is Alarmingly Intense

@No Need to Interpret the Market: Over the past couple of days, the borrowing actions of several AI giants have been so密集 that it's a bit scary.

Oracle reported earnings on Wednesday, beating expectations, but casually mentioned that to fund AI infrastructure, it will issue tens of billions more in debt or stock. Its capital expenditure this year has already reached 55.7 billion, with a backlog of undelivered orders piling up to 638 billion. Prior to this, it had already issued 43 billion in debt for AI. Following the earnings release, its stock actually dropped 5% in after-hours trading.

Amazon is going even harder. It just secured a 17.5 billion credit facility arranged by Citigroup. The stated purpose in the documents is general corporate purposes, but anyone can see it's for AI. Just days ago, it also broke records in the Canadian bond market, issuing the equivalent of 10 billion in Canadian dollar bonds in one go. The previous record was just set by Alphabet last month.

Amazon's planned capital expenditure this year is 200 billion. This figure is larger than the annual GDP of most countries in the world.

What's worth pausing over is the warning from the rating agency S&P regarding Amazon. S&P stated that its leverage will increase significantly, and to build out data centers, its free cash flow will likely be negative over the next two years. Translated, this means that this company, renowned as a money-making machine, will not generate enough cash over the next two years to cover its own spending; the shortfall must be covered by borrowing. Even so, Amazon is still rushing to borrow.

Looking at the events of the past few days together: Alphabet issued stock for the first time in twenty years, raising 85 billion. SpaceX's IPO aims to raise 86 billion. Anthropic just scraped together 35 billion in private credit from Apollo and Blackstone to buy chips. Add Oracle and Amazon to the mix. The debt issued by just these few super-majors this year is already approaching 160 billion, whereas two years ago, this figure was a mere 17 billion.

Money is pouring in from all directions into the same targets: data centers, chips, and computing power. Everyone is leveraging up, betting on the same thing: that the ultimate demand for AI will keep pace with this speed of borrowing and burning cash.