
Yes FlashBlade is indeed in a league of its own.
A VC once told me "if you have no competitors, you have a problem".
Later this year Micron plans to release quad-level cell flash drives that encroach on the nearline disk drive market. QLC flash stores 4bits/cell, whereas TLC stores 3, adding a third more capacity. A 6TB TLC SSD could become an 8TB QLC SSD using this technology. QLC flash has a shorter working life than TLC flash and is …
Going from 1-bit per cell SLC to 2-bits per cell MLC doubled cell capacity.
Going from 2-bits per cell MLC to 3-bits per cell TLC increases cell capacity by 50%.
Going from 3-bits per cell TLC to 4-bits per cell QLC increases cell capacity by 33%.
Other factors affecting overall NAND storage capacity per chip are Vertical (3D) NAND vs. Planar (2D) NAND, and semiconductor device fabrication node density (i.e., "XX nanometer process").
Higher density V-NAND (i.e., more layers) requires lower semiconductor device fabrication node density to retain reliability, so simply increasing from 48 layers to 64 layers does not automatically mean a 50% increase in chip capacity.
Similarly, with each increase in bits per cell, there is a decrease in both the number of V-NAND layers and node density that can be supported compared to a lower bit per cell.
QLC at best could provide a 33% increase in capacity, all other things being equal. But if it has to implemented at a V-NAND layer level below what current TLC is at, and at an earlire process node compared to TLC, much of the 33% capacity gain could be lost.
The result could be QLC will not, in the near term, provide a significant decrease in cost/GB, at least not at the levels see by the near simultaneous mainstreaming of V-NAND and TLC.
What should drive down the cost of QLC is the NAND industry's ability to use older fabs (a the higher process node and lower V-NAND layers) to very inexpensively manufacture QLC.
For these reasons I think the idea of an archival class QLC V-NAND competitive with NL-SAS is still a number of years away.