The Thin Margin of Safety in Selling Compute

I came across this funny tweet last week that says “the best way to signal status in the Bay Area is to act noticeably stressed about lack of compute”. Every earnings season, all the hyperscalers and neoclouds remind us how demand for compute is outstripping supply. Given these contexts, one could be forgiven for thinking selling compute must be an easy business. Indeed, selling compute in 2026 is likely to be one of the easier positions to be in the AI value chain.

Of course, what matters for owning compute sellers as long-term investments is what happens beyond 2026, especially if the demand-supply mismatch in compute starts to narrow over time with a deluge of supply coming online in the next few years. So, I wanted to understand the margin of error for owning a more pure-play compute seller such as CoreWeave.

I will share the detailed work, along with a downloadable spreadsheet, behind the paywall.

Subscribers get the daily journal and five+ years of Deep Dives, i.e. full-length analyses with financial models on 70+ companies. The daily is just how I think out loud between the Deep Dives!

Let’s start with how a compute seller such as CoreWeave makes money.