The Atmos format is built on modular speaker systems. Its object based. Meaning, the studio engineer will not place the sound in a particular channel like conventional 5.1/7.1 system. Rather, the recording engineer will place the sound in a 3d location, XYZ co-ordinates.
The audio processor/receiver will have the info about the room layout and speaker locations in 3D. Then it will look at the object, where its placed, map it to the current room + speaker layout and send it to the appropriate speaker/s in real time. The system has gone from dedicated 5.1/7.1 tracks to real time rendering of tracks.
It does not matter how many speakers the theater/room has and how they are placed. The processor will analyze it as part of setup and then use the audio info it has to render accordingly.
DTS also has similar system in development. Tha'ts why they acquired SRS recently.
Update:
To add more to this post - we will not be able to replicate the same sound effects with less no of speakers. Having more no of speakers in appropriate locations will help in rendering the sound more accurately and it will help the processor place those sounds at appropriate locations. In my opinion, the current 7.1 speaker layout does very good job in horizontal plane and meets most of the needs. So, to take care of vertical plane, I think addition of 2 or 4 more speakers on ceiling can take care of most of the home rooms. Although, most of this depends upon the room layout and size too. Most people will be able to do 5.1 + 2 for heights quite comfortably though.