This is tricky as its going to depend on the device capabilities and the load on the device.
Many mobile devices for example will have HW solutions to decode and play video, especially important for HD and more complex codecs.
Once you go beyond one, or a small number at best, video you will likely start having to do the decoding in SW which will be much slower and will put a heavy load on the available processing power. If the device does other heavy processing during the playback this will impact the videos also.
If you need to display a grid of videos for example then sometimes it is easier to actually merge the videos on the server side and then send it to the device as a single video.