NVDA
The money machine is getting creative
AI infrastructure spending has officially entered its “who’s actually paying for this?” era. According to the article, Nvidia is helping backstop more than $100 billion in debt for a massive AI data center in Ohio tied to OpenAI, even as questions swirl around OpenAI’s profitability and slipping market-share momentum.
That’s not exactly the kind of sentence that screams “totally normal corporate finance.” It reads more like a venture-backed fever dream with power bills.
Capex: because apparently $790 billion still isn’t enough
The larger theme here is the capex arms race. Hyperscaler spending is projected to surge 90% year over year in 2026, approaching $790 billion. That’s a giant number — the kind of number that makes even tech giants look like they’re playing SimCity with unlimited cheat codes.
But there’s a catch: AI service growth isn’t rising nearly as fast as the spending. So investors are left with a pretty basic question dressed up in fancy spreadsheet clothes: if the revenue isn’t catching up, who’s holding the bag?
Why investors should care
This matters because the AI boom is increasingly leaning on financing gymnastics, not just clean operating cash flow. If growth slows, margins wobble, or demand doesn’t keep up with the infrastructure buildout, the market could start treating these capex plans less like a growth story and more like a stress test.
Big picture: the AI trade is still alive, but it’s starting to look a lot less like a rocket ship and a lot more like a very expensive relay race.