I have created a lab using the lab plan with Windows Server 2022 Datacenter (Gen2) template. I need to create two VMs with hyper-V on the template. I followed this article. We are teaching basic Windows Server administration tasks such as installing Active Directory services, File Server, GPO, etc.
After I published the lab, I realized all users are connected to the same DHCP server named something like "xyz.fx.internal.cloudapp.net". So, If VM1 changes the DHCP configuration, VM2 is affected by the same configuration even though the scope is bound to the Virtual Switch (192.168.0.1) which should be a Hyper-V internal interface. Meaning that other VMs in the network should NOT see that process.
In the article you mention, the DHCP role is installed on the host machine for convenience, so the client/nested VMs will automatically be assigned an IP address when they boot up. However, you don't have to do this. You could manually assign the IP addresses to the client/nested VMs.
The *.internal.cloudapp.net is the default FQDN for all Azure VMs. It's not coming from the host machine.
The network being setup in the article you reference is a NAT network. This means that will proper configuration, the client/nested VMs can see the internet and each other. However, anything outside of the host VM will not be able to access the client VMs.