Hi,
The official documentation can be found here. It basically explains the architecture behind this with example. By having one fault domain you basically put all your eggs into the same basket so to speak. In such case it is not a very good idea to update two VMs at the same time, especially since in order to take advantage of availability set you need at least two VMs. Even if you have more than 2 VMs it is risky to update more than one VM at the same time when all your VMs are placed on the same hardware. Overall, it is limitation based on logic. If you choose such kind of setup you are agreeing that your VMs are placed on same hardware and you are probably doing it because of some low latency communication you want to achieve.
Please "Accept the answer" if the information helped you. This will help us and others in the community as well.