5 Comments
User's avatar
Jorrit's avatar

Fascinating take on NVIDIA’s verticalization strategy. I like how you frame Vera + Rubin as a move from component supplier to full-stack control, with CoreWeave as the proving ground. The economics of owning the entire bill of materials really changes the game. What do you see as the biggest risk to this strategy: execution at scale or hyperscalers accelerating their own silicon even faster?

I’ve subscribed and would be happy to support each other.

Jorrit

Edge Of Power's avatar

NVDA needs somewhere to sell its chips and build out its own cloud. CoreWeave is a deeply unprofitable project that serves exactly those purposes. NVIDIA funnels its clients there, including META, which allows it to monetize both the cloud layer and the direct chip sales.

Why CRWV chose such a deeply unprofitable and, frankly, financially questionable model is hard to say. Maybe they’ve already cashed out and decided they’ve made it.

Felix's avatar

When a business becomes to powerful they can just buying themselves into anything. Absurd.

Edge Of Power's avatar

That’s exactly what’s happening. I think they wanted to do it from the beginning

Felix's avatar

Probably indeed!