>MLC getting the SLC treatment, will almost completely disappear this time next year
>3D TLC will become the norm, with inconsistent performance/crappy write speeds
>Larger NAND dies means less parallelization, controllers getting weaker to drive down costs
What will save us from the looming hell of slow NAND?
>>60926510
In the near term, nothing. They can afford to do this because there is no competition to NAND flash in terms of speed/cost. But NAND has been getting long in the tooth, and people are looking to replace it, but in my opinion, whatever replaces NAND will probably end up replacing DRAM in a generation or two.
Xpoint might make it but it is highly unlikely in my opinion. We have PCM, ReRAM, FeRAM, and MRAM, of which the last I am personally backing having brought stock in it after researching the topic and is the furthest along on the speed front.
http://www.anandtech.com/show/11184/everspin-announces-new-mram-products-and-partnerships
The main problem with MRAM is capacity and the fact that STT_MRAM is two terminal, which saves space but wears out the dielectric barrier between the memory cell, so we need something like spin orbital MRAM to hit to make it. Also, you can make multibit cells with it since the operation of MRAM depends on spintronics, but the durability is also through the roof, so you can have TLC MRAM with no downsides and with the same speed as SLC MRAM.
Of course, the developments I listed unfortunately only appears in academic literature for now, and I don't think I can stay here long enough to quote the papers but you can find them online. It will need a few years to come out onto the market, and that is if it can be scaled, which I am confident that it can.