The AI shopping spree isn’t cheap
Everyone loves the part of the AI story where companies race to build bigger models, faster chips, and shinier data centers. The less glamorous sequel? The rent.
According to Mona Dajani, the AI buildout has piled up about $1.5 trillion in lease obligations. That’s not pocket change — that’s the kind of number that makes even the most confident hyperscalers pause and check whether they left the stove on.
Why investors should care
These contracts matter because they’re sticky. If the AI boom keeps humming, no problem: more data centers, more capacity, more compute, more revenue. But if demand suddenly wobbles, those long-dated lease commitments can become a financial anchor instead of a launchpad.
What makes this tricky is that the market is still trying to price AI like it’s an endless straight line up and to the right. The reality is messier:
- hyperscalers are locking in capacity years ahead
- companies are signing contracts for their own data centers
- a change in AI enthusiasm could hit the economics faster than the headlines suggest
The big picture
Dajani’s view is that the AI race probably keeps accelerating. Fair enough. But the scale of the obligations is a reminder that every gold rush eventually invoices somebody.
Big picture: the AI buildout still looks powerful, but the hidden liabilities are getting large enough that investors may want to read the fine print, not just the hype reel.