As I have mentioned in another thread, the issue with cost/size isn't going to go away anytime soon. Sure SSD's will drop in price, but not greatly when you compare SSD generations to hard drive generations. The biggest issue is simply about NAND memory densities and production capabilities. With a hard drive, they could find ways to pack more data into a small space by refining the surfaces of the platters, heads, and other such stuff. With SSD's it's about the number of NAND chips and the densities of each chip.
Unlike HD's, you cannot really do small-step refinement increase of storage amount per size. For example, some engineer might come up with a way to make platter substrate more smooth (just pulling an example out of my ass). Bam! That's an immediate possible increase in storage capability, since bits can now be packed more tightly together.
NAND memory is pretty much set in size by the particular process (35nm, 25nm, whatever). Thusly for a certain size chip, a particular amount of memory is all you will ever get from that process. The best you can really do is increase the amount of yield per wafer (percentage of good chips from the original silicon wafer) through process improvements, but you will never be able to pack more data storage hold ability into the chip unless you change the chip completely.
Unfortunately to change the chip completely, generally means new fab equipment. All the equipment in the existing fab process is built around that process. You can't just "tune them up" to make higher density chips. It doesn't work that way. Fabs are DAMN expensive and only a couple of memory makers actually have the cash to to do this currently. Micron is one and I think Samsung (which gets a LOT of help by the Korean gov't) might be the other. The other issue with fabs is that they take a long time to get online. So even "if" a bunch of the memory makers (the majority of which are in lots of financial trouble after the last few years of economic crap) could get loans to open a new fab, it would be a couple of years before anything rolled off the lines, and as you know in the electronics/computers business, two years is a lifetime.
In addition there is a LOT of competition for NAND memory to be put into different devices, so each chip has a certain market value outside of SSD useage (e.g. cell phones, etc, etc). Like SSD's, these other device market prices are constrained by NAND supply/pricing. While a company like Micron can see the benefit of having an SSD division, it's primary goal is to sell memory, and if somone like Apple comes along with some new super-Ipad or something and wants to buy a all of Micron's NAND at better prices/profits than Micron would get selling them in SSD's, where do you think the memory is going to go? There aren't that many memory makers making large supplies of NAND chips, so if something popular comes out that uses NAND, it affects all the pricing of all the other devices that use NAND.
PCIe type boards are going to come down in price compared to regular SSD's even slower. As I mentioned NAND chips have a market set price for each. The more you pack on a board (regardless of the type of board), overall price goes shooting up. There is no quantity discount type deal on these type of boards and the market is much smaller, thusly the development costs are much higher compared to the return.
For SSD performance, the current biggest issue is SATA chipsets. Other than Intel, they pretty much all suck. With hard drives, you couldn't really tell how bad these SATA chipsets/drivers sucked because the limitations were generally the hard drives. With decent SSD's and a few benchmarks/compatibility test, you can see how bad some of these non-Intel SATA chips and drivers SUCK! And not that Intel's SATA controller is super incredible, it's just better than everyone else's by far. Intel needs to get off it's ass and integrate some SATA3! SATA3 is currently a gimmick for a couple of benchmarks currently.
Facebook
Twitter