NAND versus Future Memory Technologies

Thank you for the comments to my last blog. To clarify, I believe we are talking about two different topics spread over two different periods, and I would divide the question into the market conditions of today and those of the near future. In the market conditions of today, the issue is the transition from NAND to any of the emerging nonvolatile memory technologies.

I agree with you that such a direct replacement of one technology by another technology in a single application usually only takes place if the cost-per-bit of the two technologies is similar. I also agree that condition is extremely unlikely in the case of any new semiconductor memory technology challenging any existing high-volume memory products.

I have commented in the past that I don’t expect any new memory technology to directly replace another memory technology. To me, the transition takes place when an emerging technology enables a new product, and that new product obsoletes the older product over time.

This may seem like an unnecessary fine distinction to make, and that distinction is probably not as critical in the relatively fast-changing world of logic designs. In the memory world, however, things necessarily move much slower. This is not due to lack of intellectual creativity, but because of the nature of the customer base.

Each leading-edge product design is expected to support a large number of diverse applications. Each step forward is accompanied by an extremely complex support infrastructure, and the transition from one generation to the next is expected to take place with the minimum amount of OEM disruption. The challenge of doing that well, time after time, can be measured by the 30 or so companies that have dropped out of this market while still surviving and thriving in other product markets. When one intends to supplant a product or a supplier under these conditions, attention to the fine distinctions is critical.

For those reasons, I don’t expect NAND technology to “fail” and then be “replaced” by PCM or other technologies. NAND has replaced chemical film technologies in image storage and is rapidly replacing magnetic storage in mobile and personal computing devices, along with a growing number of other applications. NAND bit growth doubled from the first- to the fourth-quarter 2009; that growth rate will likely continue for several more years.

In one sense, it is the success of NAND that has brought the attention to the other semiconductor memory technologies. The success of NAND in competing against smaller scale magnetic storage devices begs the question of whether semiconductor technologies can eventually replace all magnetic storage devices and eliminate all moving parts in large-scale computing.

I see the potential of NAND and other semiconductor memory technologies providing a continuum of cost and performance attributes replacing magnetic data storage in a growing number of applications over time. Other emerging memory technologies with different performance attributes will also enable different applications. Perhaps I am slicing it too fine, but I don’t view this trend as a replacement of NAND so much as a step in a continuing evolution of semiconductor technologies replacing magnetic storage technologies.

The second topic regarding the market conditions in the near future is definitely more subjective and certainly open to a much broader range of opinions.