For example, Amazon Web Services (AWS) is one IaaS provider and is built as a multi-tenant system, which means that multiple virtual machines (instances) can run on a single physical server. However, each virtual machine (instance) has a dedicated part of the physical machine (e.g., its own GPU, dedicated RAM, and dedicated CPU cores). This part of the server is, therefore, only available to a single instance and is not shared.
The instances are provisioned on-demand, so your app may run on one instance on one piece of physical hardware on one day and on a completely different instance and hardware on another day. Also, you may have your app running on one instance on physical hardware, and there may be another app running on another instance on the same hardware. Note, however, that when running, complete isolation is maintained at the instance (virtual machine) level. If you require dedicated hardware, this can be arranged, but it does come at a premium cost.