Aug 15 2009 1:36PM | Permalink |Comments (0) |

Jim Elliott is Samsung’s VP of Memory Marketing and right now, Samsung’s got the NAND Flash tiger by the tail. Elliott’s keynote title at the Flash Memory Summit was “NAND Continues to Soar (App after App after App...).” There were many valuable nuggets in Elliott’s keynote and I’ll capture a few of them here in this blog post:

“The current NAND business model is not supportable.” Here, Elliott’s talking about the recent period (2007-2008) when the NAND Flash vendors were forced to sell product below cost just to continue recouping the fab investment costs they’d made a couple of years earlier. CAPEX (capital expenditures) for fabs and equipment always lead the effects of those investments (in terms of sales and pricing) by about two years. Well, the CAPEX spigot got shut off and prices have recently risen and stabilized somewhere above cost. Don’t worry though, even though the freefall drop in NAND Flash pricing seems to have abated (at least for now), it’s still predicted to fall by between 20-30% each year for the foreseeable future, as you can see from this graph.

The key applications are migrating up the food chain for NAND Flash. Enterprise devices like SSDs for enterprise systems use and will continue to use single-bit-per-cell (single-level cell or SLC) NAND Flash, which is the fastest, most reliable, most expensive NAND Flash device with the highest endurance and data retention. Enterprise application developers can put a price on speed and data integrity where other applications often cannot. Even so, Elliott sees at least some enterprise applications migrating to two-bits-per-cell (multi-level cell or MLC) NAND Flash devices in the future for cost reasons. This will only be possible if SSD subsystem vendors vigorously adopt the extensive ECC schemes originally developed by hard-drive designers.

Consumer PC applications currently use MLC NAND Flash devices to get a balance among price, performance, and reliability/endurance. In the future, says Elliott, these applications will be migrating to three-bits-per cell (three-level cell or TLC) NAND Flash devices.

Flash drives currently push the cost envelope as far as possible and use TLC devices but will migrate to four-bits-per-cell (4LC) NAND Flash devices as soon as feasible to continue on the cost-reduction curve. As the costs of such devices continue to decrease, Elliott predicts the rise of NAND Flash as a content-distribution media. He points out that music and media players and mobile phone handsets do not have optical disk drives and will not get them. They do however incorporate microSD slots and are therefore good targets for content distributed on NAND Flash.

As far as market trends are concerned, Elliott predicted that the real bit growth for NAND Flash will be in SSDs, as you can see from this graph from Elliott's keynote:.

Although SSD sales volumes will be nowhere as large as the billion/year run rate of mobile phonehandsets, perhaps rising to 100 million units per year by 2013, each SSD will contain far more NAND Flash memory than a mobile phonehandset. Hence the NAND Flash bit growth favors SSDs.