Rack GPU vs Virtual GPUs

Hi there, I’m following the ‘process’ to see if there is someone in the NVIDIA universe who can help us with a technical GPU question? It is our understanding that NVIDIA does not make a rack server for GPU and our only option is the Orin or Thor boxes, is that correct? We have a large computer visioning deployment and it covers 9 separate site locations which would require 3 edge Orin boxes per site to handle the number of camera feeds. Can someone point me to where NVIDIA might have a partner who can provide a rack server solution for all 9 sites or elaborate on how Virtual GPU might be able to handle this?

There are more detailed explanations here,

so take a look here.

https://www.nvidia.com/en-us/design-visualization/quadro-vdws/

It doesn’t seem like the virtual PC would work because it’s not going to be large enough to handle multiple camera feeds. Is there a server sizing sheet or something that I can use to help me size up how many GPUs and server size I can buy? thank you so much for the initial reply.

You can look it up here.
[Translated with DeepL.com (free version)]