Why enterprises are ditching hyperscalers for neoclouds

Today, they're $100 million. Okay, and when the Vera Rubens come out, 16 racks is gonna cost you $170 to $200 million. These numbers are mind-boggling. People often ask me, why do we need so many neoclouds? Why don't they just go to the hyperscalers? It isn't necessarily either or, but for inference and training right now, those workloads, we are seeing the neoclouds being chosen. And we're faster. We're a lot less costly. We're more cost-effective as well.