RAW capacity is not the same as STRUCTURED capacity.
Different ways of structuring the RAW capacity result in different USABLE space.
Similarly, different image content results in different image file sizes when shooting JPG.
If engineers knew EXACTLY how large each and every one of your images would be, the exact full capacity could be available for image storage...however, that is unreasonable.
(I just checked Canon 7d RAW files I shot today...even their sizes vary)
Solid state storage devices may have additional capacity YOU CANNOT SEE, reserved for "bad block revectoring"... on some devices this may be 80% additional capacity.
So... in real life, the RAW capacity allocated for "user stuff...like images" as a collection of fixed size "chunks/blocks" of storage... used to be 512 bytes per block, now 8192 is a common size. When you need storage (for an image), it receives as many "blocks" as necessary, probably not fully using all the space of the last block. Since all files/images are not uniform size, something has to keep track of which blocks belong to each file. This recordkeeping overhead is space that the "structure" must use to make the remaining space USABLE for you.
Be happy... 16Gb cards weren't possible just a few years ago... but are today.
I have cards from 64Mb... to 16Gb)
Just remember - every mem card will fail... they have limited write cycles...and lots of engineering to maximize their longevity... write wear leveling, write cycle monitoring, bad block revectoring, and more