h264 is the name of "part 10" of the Mpeg4 encoder specification. ( https://en.wikipedia.org/wiki/H.264/MPEG-4_AVC
Its job is to effectively compress video information in a way that the resulting decompressing is almost exactly the same as the original video.
Your calculation about how much data it is in a "raw" videostream is probably correct, but the Pi camera (together with the GPU) will produce already compressed video.
(Because raw data would be too much for the SD-card to handle btw.)
When we say its raw h264 that means that the videostream is not
inside a container file-format. http://www.pitivi.org/manual/codecscontainers.html
One main purpose of container-formats is to synchronise videostream with one or more audiotracks, subtitles etc.
Since the raw videostream is missing some information (like framerate) means that some videoplayers will have problem playing your file correctly.
A solution is to use the programs "avconv", "ffmpeg" or "mp4box" to "mux" (multiplex) the raw stream into the standard container format for mpeg4 video.
( which happen to be part 14 of the standard )
The camera output is already compressed by a "not lossless" codec (h264), and there is no gain trying to reencode it again