NVDA
Another check, bigger thesis
Nvidia doesn’t seem interested in just selling picks and shovels anymore — it wants to own the whole AI gold rush map. Its reported $6 billion deal tied to Poolside’s model factory is basically a bet that the cheaper AI gets, the more everyone will want in on the action.
Cheap AI, expensive consequences
Goldman Sachs is floating the idea that lower-cost AI could be a gift for adoption. Translation: if the price of building with AI drops, more companies build with AI. That’s great for usage, great for demand, and potentially great for Nvidia’s chips — even if it’s a headache for frontier labs trying to keep their margins looking cute.
Why investors should care
For Nvidia holders, this is the sneaky part of the story:
- More accessible AI can mean more training and inference workloads
- More workloads usually means more demand for accelerators, networking, and the whole Nvidia stack
- But if model economics get commoditized, some of the juicy software-layer profits could get harder to defend
So yes, this is a partnership story. But it’s also a thesis test. Nvidia is basically asking: if AI becomes cheaper and more open, does that shrink the pie — or make the pie so big that everyone still needs more chips?
Big picture: Nvidia keeps betting that in AI, volume can matter more than pricing power. And so far, Wall Street hasn’t exactly been eager to bet against it.