I feel that this is the last generation we'll see with limited VRAM.
I think some forget that making a 12GB GPU as the bear minimum isn't straightforward. Without 3GB modules, the bandwidth has to change, and that changes a lot of how a die is developed. Budget GPUs have traditionally been 128-bits, 256-bits for midrange, and 384-bit for high-end. For a budget GPU to have 12GB of 1 or 2GB modules, they'd need to change the architecture to 'fit' a larger bus width of 192-bits. The 3060 had a 192-bit bus but was more expensive than the 5060 and was quite a bit larger of a die. The 6700XT was 192-bits but was even more expensive and larger in die size. Nvidia and AMD have to make a complete shift in budget cards to fit 12GB as a minimum. So I imagine they'll make 16GB the minimum instead if 3GB modules don't become the norm. They should have done that this generation, but I can see why they didn't. I think this will be last generation to do that. They've pushed it as far as they could. Unless we see a 192-bit 6060 with 12GB of VRAM, I imagine 16GB will be the new baseline. The problem with that is they will have to change the bus width of higher-end cards to fit 24GB (unless 3GB modules) for cards like the 6070Ti. So the same problem arises. It all comes down to early development and planning. I'd say Nvidia did the calculations and felt that a 5060 with a 192-bit bus was not beneficial. But maybe that'll change in the future. It's more nuanced than many make it out to be. Either they boost VRAM by doubling it across the board (6060 16GB up to 6080 with 32GB), they increase/decrease bus width, or they use 3GB modules more readily.