The Neoclouds Series, Part 2: CoreWeave

This is the second article in my series working through the neocloud sector, the fast-growing AI infrastructure companies that buy Nvidia systems, assemble data centers around them, and rent the resulting compute to hyperscalers, AI labs, and other companies that need compute for their AI usage. Part 1 covered Nebius.

Part 2 covers a company that was one of the first in the “neocloud” sector, but this is not about how early it was. CoreWeave invented and defined many of the terms and creative accounting practices that are now widely used by other competitors in the sector. Not only that, but it has taken the model further and faster, with more leverage than any other public neocloud. Its aggressive approach to stockpiling as many GPUs as possible and renting them out as computing capacity appealed to Jensen, who adopted CoreWeave and made it the model, both then and now, for how neoclouds should do business, how they should manage their relationships with Nvidia, and how quickly they need to grow because Nvidia needs to sell. That is why CoreWeave is considered the most important neocloud in the sector and the most important independent SPV for Nvidia.