If you're planning to grow your network, then I'd suggest sizing your VM based on the expected size. I like to do this ahead of time to save on the future hassle, but there are older forum threads that describe the VM sizing increase process (requires VM instance restart). So for VM of AMP 1000, you'd scale the same as the AMP 1000 server scaling in the sizing guide, but then increase RAM and CPU by 20% to account for VM overhead. You also want to definately make sure that your disks are 15k RPM or better (SAS or SSD) or even better than that (SAN).
If you have Failover, then that instance can also be on the same VHost, but again - it'd have to have the same sizing as the AMP instance - because if the AMP fails, then the Failover server needs to meet the same server requirements to process the data.