I'm glad to see that someone is looking at adding flash as something other than a pseudo hard drive through a SATA interface. That connection is very nicely optimized for spinning media and really doesn't take best advantage of flash media. Some companies have done PCIe interface drives, but the price difference is very large. If it were dropped into a standard memory slot then raw flash filesystems in Linux like ubi could use it very effectively.
although I'm always gratified to find flash devices a little cheaper at the corner computer store, the prices are not much like spinning media. (especially now that the industry is mostly recovered from the thaipocalypse...)
yes, moore's law applies to flash. sort of. not entirely. but really the point is that spinning disk also benefits from a density that's the product of two dimensional shrinkage (ie, same class of curves as moore's law!) though really there are more interesting things happening: MLC was pretty fun, as are the almost-step-functions that the spinning disk industry rides (GMR, perp, maybe patterning or laser-assist?). at the moment, disks have some major changes pending, and flash *doesn't*. (though perhaps a non-flash solid-state NV memory will take up the mantle - maybe but not imminently.)
so really, why is flash interesting? it seems to be in a boring phase, with not a lot of innovation being exhibited. yeah, sure, put 20TB in 1U (ie, match spinning disk), but at what price? is it terribly passe and old-fashioned to perseverate about price?