I wondered the same thing about what justifies the 50x cost, and I do believe it has everything to do with the actual manufacturing costs rather than the materials. The fact that the DIMM shown above is only 64MB is pretty telling of this.

There’s an answer to this though: the possibility of developments in the field of hardware encryption. At the speeds ST-MRAM operates, if we get into a stage where hardware-based encryption becomes a reality, we may have for the first time in the computer history a truly functional non-volatile random access memory. The performance impact will be negligible.

What is more interesting is that with this solution, we could approach the greatest grail of Storage: A unified storage solution. All types of memory being served by the same technology.

We need however to understand exactly what determines the 50x price tag on ST-MRAM. If this is exclusively attributed to manufactoring costs, I suspect ST-MRAM will not see its price come down anytime soon so it could become ubiquitous on our computers. If however there’s production costs involved here, it’s possible.

As a replacement to NAND, I doubt we’ll see any major improvements just yet. Remember, it’s the controllers and interface that are the limiting factor at the moment. At least we have SATA Express and SFF-8639 to look forward too.