Quite astounding considering the current GM200 "Maxwell" flagship model has just 8 billion transistor but we assume this includes the transistor count of the HBM2 memory. FUD Zilla claims TSMC's 16nm FinFET process will be used by NVIDIA's and AMD's next-gen GPUs and notes the size of the gate is almost identical to that of Samsung/GlobalFoundries' 14nm FinFET process. Interestingly, the site also heard that the size of the 14nm Pascal GPU will be significantly smaller than that of the 28nm based GM200.
Consumer versions of Pascal will reportedly get up to 16GB of HBM2 while the professional and HPC versions may get as much as 32GB HBM2.
Nvidia will use second generation HBM for its Pascal GPU to get to a 32GB on the highest end card, This is 2.7 times more than the already impressive 12GB used on Titan X. The second generation HBM or HBM 2.0 will enable 8Gb per DRAM die, 2Gbps speed per pin and 256 GB per second Bandwidth/ stack.
The first generation offers 2Gb Density per DRAM die, 1Gbps speed per pin, 128 GB/s GB per second Bandwidth and maximum of 4 Hi stack chips with 4GB per HBM card. You saw this with Fiji cards.
HBM2 enables cards with 4 HBM 2.0 cards with 4GB per chip, or four HBM 2.0 cards with 8GB per chips results with 16GB and 32GB respectively. Pascal has power to do both, depending on the SKU.