I will have to assume that you with Always On mean "Availability group". (Always On is an umbrella name, which besides AG also includes Failover Cluster Instances.)
One idea is to set up a distributed AG with the servers in the new AG in the other group. Once they have been seeded, you can fail over to the new AG. I believe that you should be able to remove the initial AG once you are done.