I feel like I may have communicated poorly. I mean that, ultimately, any recursively self improving intelligence would be limited by the power cord, and as such, the idea that it would permanently escape human control seems somewhat silly.

But you can imagine an agent trading crypto and using its profits to rent a rack of H200s using a cloud provider. There are definitely ways for one rogue orchestrator to expand their compute without anyone being aware.

Overnight? Yes, silly. Given many years of seemingly disconnected actions taken throughout the world? Not silly at all. An intelligence capable of self improvement can also design robotic chassis.