the end of next year, the DRAM industry will have shrunk to only three
major players, says Howard: Samsung, SK Hynix, and Micron. While there
are other DRAM makers, they don't own the technology and have to license
it from these three, he notes. The DRAM manufacturers in Taiwan, for
example, are still hanging on. "They just refuse to go out of business."
The DRAM vendor in
the best position is Samsung, which has about 40 percent of the overall
DRAM market and 60 percent of the mobile DRAM market, according to
Howard. It also has a stable customer -- itself, in Samsung Mobile.
Hynix SK has 22 to 24 percent of the overall DRAM market and 22 to 24
percent of the mobile DRAM market. Micron historically had about 13
percent of the overall DRAM market and only 4 percent of mobile DRAM. It
is fixing that by acquiring Elpida, which has about 20 percent of the
mobile DRAM market.
With the right
balance of PC DRAM, mobile DRAM, and specialty DRAM, these companies
should be in a more stable market. All three of them also make NAND
flash as well. And even though NAND is made in different fabs, "if they
feel there's way too much capacity in DRAMs, they could transition those
fabs to produce NAND," according to Howard.
In some ways, the
NAND market has the elasticity of the old DRAM market, in the sense that
OEMs can always stuff more memory into their products. That's
particular true for the solid-state drive market, notes Howard. On the
other hand, the mobile device market doesn't have that flexibility. Once
a smartphone or tablet is designed, it's not so simple to just add more
elasticity component of the DRAM market is shrinking [as the PC market
shrinks], whereas the elasticity component of the NAND market is
these factors should help the remaining memory companies build a better
business future. "They aren't out of the woods yet, but long-term I
think the industry is going to be a lot healthier," Howard adds. By
2014, "we anticipate this being a tidy, profitable industry."
Tam Harbert is a freelance journalist based in Washington, D.C. This article originally appeared on EBN, an EE Times sister site.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.