Surprises me that RAM hasn't increased more, the machine I'm on is 5 years old and it has 2GB of RAM on the GPU (HD6950) which I paid ~200 quid for 5 years ago.
Memory bandwidth/throughput has been a far larger bottleneck on gaming performance than total video memory for the last few years. They've been trying to deal it by using ever faster and slightly wider memory interfaces, but now they've hit the wall, thus the move to HBM.
I think it's due to the heating issue on the boards themselves. HBM promises to make that issue a bit better since the memory chips are closer to the GPU which makes it easier to cool off than before. But there's a limit due to HBM growing vertically. So, I doubt we'll see massive memory chips on consumer graphics cards in the near future.