This article is more than 1 year old
FLASH! Aaa-aaah. 3D NAND will save every one of us
19TB flash cards? Not as far off as you might think
Comment We have 3D NAND flash chip supplier timing, courtesy of financial analyst whiz Aaron Rakers. The four foundry operators will all be pumping out mass production volumes of the high-capacity stuff in 18 months' time
3D NAND is a fresh way of getting out of the NAND scaling trap. The traditional way, of increasing NAND chip capacity through decreasing the cell size, runs into a dead-end of cell state unreliability and short working life.
By retreating from the cell size shrink route and stacking layers of cells one atop the other manufacturers can increase capacity without increasing the footprint of the cell.
Samsung is taking the lead here, having reached the 1TB capacity level.
Rakers, managing director of analyst outfit Stifel Nicolaus, has produced a chart comparing the four foundry operators' timescales:
Samsung announced mass production of its 24-layer 3D V-NAND dies in August last year, and its 32-layer, TLC (triple-level cell) 850 EVO V-NAND SSD yesterday.
Rakers notes:
- Samsung’s 850 PRO / EVO SSDs have performance and endurance but not cost superiority
- Intel says its (and Micron’s) 3D NAND will be industry’s first disruptive 3D from a cost perspective in 2H2015
- SanDisk will have 3D NAND in 2H2015 that will be less expensive than its current flash technology
We think that SSD capacities could likely double and/or the cost/GB of 3D NAND will be less that than current 2D or planar NAND.
That implies that PCIe flash cards could go up to the 12TB or more area. SanDisk's SX300-6400 is rated at 6.4TB. Double that to 12.8TB and then, if TLC comes in as well, add on 50 per cent to reach 19.2TB.
Server flash DIMMs could also benefit, with the 400GB capacities available now rising to 800GB and then 1.2TB with a TLC boost if that's feasible.
The net effect here is that server flash capacity and all-flash array capacities should see a substantial jump in 2016. The active archive flash array product category could see – indeed, should see – a significant lowering of its $/GB number making it more attractive from acquisition and OPEX (power draw) points of view, relative to disk and tape archives. ®