Memory giant Micron Technology has commenced mass-production of 21 Gbps and 24 Gbps-rated GDDR6X memory chips that will be exclusively used by NVIDIA in its next-generation RTX 40-series "Ada" graphics cards. GDDR6X is a derivative of GDDR6 co-developed by NVIDIA and Micron, which leverages PAM4 signaling to increase data-rates. Depending on the graphics card model, NVIDIA will use 8 Gbit (1 GB) or 16 Gbit (2 GB) density memory chips. We're hearing that 21 Gbps will be the standard data-rate used by SKUs that succeed the RTX 3080 and RTX 3080 Ti; while 24 Gbps will be used by the faster RTX 3090/Ti successors. The part numbers of these memory chips are listed below.
Looks like NVIDIA is ramping up for the next gen GPU release.