Debt Can Plug the AI Hole in Big Tech’s Deep Pockets
We’ve moved from marveling about AI’s potential to puzzling over how to pay for it. The price tag seems to grow every day. On Thursday, Tesla CEO Elon Musk suggested Tesla will need so many chips for its robots and self-driving cars that it might build its own chip factory. He also said Tesla would have to spend “tens of billions” to train the AI in its robot. Meanwhile, his archrival, Sam Altman, is pursuing so many investment initiatives that his finance chief (briefly) opened the door to the idea of the federal government helping guarantee the financing for AI chips.
One group of companies that may have more funding options are the hugely profitable tech firms, such as Google, Microsoft, Meta Platforms and Amazon. These deep-pocketed giants so far have been mostly financing their AI expansion through the cash their businesses throw off, although that’s changing, as this deal done by Meta demonstrated. Last week, Tony Kim, who runs BlackRock’s global technology funds, pointed out in an appearance on The Information’s TITV that given the strength of their balance sheets, the biggest big tech companies could afford to borrow trillions of dollars between them. Imagine what that would mean: Instead of being deep-pocketed giants able to spend freely on whatever struck their fancy, these companies would suddenly be forced to think about every penny they spent. That might not be such a bad thing in the long run, at least for shareholders.
In the TITV interview, Kim estimated that the 10 biggest tech firms between them generate roughly a trillion dollars in annual earnings before interest, taxes, depreciation and amortization—and have no net debt (meaning their cash reserves are higher than any debt they carry). Given the debt levels companies in other industries often take on, he asked why tech firms couldn’t borrow at “one, two, three times net debt to Ebitda” to help fund capex for AI. In this (very speculative) scenario, Google, which now has $100 billion in cash and only $21 billion in debt, could end up with, say, $450 billion in debt and perhaps just $20 billion in cash.
There is of course enormous risk associated with borrowing money for something that has an uncertain return, which certainly describes AI. But this isn’t like borrowing money to buy crypto, which has no underlying value. AI has the potential to create real benefits for businesses, so the debt could enhance long-term returns. As their AI spending increases, big tech companies could become net consumers of cash rather than net producers—at least until AI revenues ramp up. Such a bet may pay off, but life in tech won’t be the same.