This message was deleted.

It doesn't matter what the platform is, you still need to have adequate licensing to cover your needs. Depending on if you need vDWS that can make a big difference. For example, if you only need virtual apps, you can run a central instance and have a bunch of vApps licenses that a cheap and share the same central server and have the flexibility of not having to have a one-to-one allocation of a vGPU. If each end user needs a dedicated machine, then the vGPU configuration and licensing is needed and that can be quite a bit more expensive. In our case we have a lot of different end points, hence do not use dedicated clients, but still have to purchase enough licenses to cover our maximum number of connections. Some of the servers that contain the apps run on bare metal, some under Hyper-V and some under XenServer. The licensing document from NVIDIA is a bit confusing, but hopefully can clarify some of these points.