Hi there @Mooli and (belated) welcome to our NVIDIA developer forums.
Sadly much of the information you are looking for is proprietary and not shared publicly.
The OS is a custom OS based on Linux that runs the VMs, and the GPUs used are whatever we need them to be. The FAQs on GeForce Now explain more about what kind of GPU performance level you will receive on the client side. But the server side is highly dynamic and hybrid to accommodate a highly scalable system.
I guess the closest you get with public information are our virtual GPU pages, specifically Virtual Workstations from NVIDIA RTX
Thanks!