I wish I could do this, but I already have more than four billion As, so my A IDs need 64 bits. I use composite keys for my Bs (A ID + sequence number per A), which allows me to search on the pkey index using just the A column; but by the time I get to C, that gets unwieldy. So my Cs have UUIDs, my Bs also have UUIDs (secondary unique ID, not the pkey) and my Cs have a foreign key pointing to their B. While also having an indexed A ID field.

I'd love to just just have "A-B-C" as my Cs' IDs... but it'd only work for my use-case (i.e. be performant) if it was running on a computer with 256-bit registers.