techweb.com
October 12, 1998, Issue: 1130
Section: Extra: Memory
--------------------------------------------------------------------------------
DRAM market changing? -- It's more complex
Mark Ellsberry
About 70% of all DRAM production is absorbed by PC OEMs. But the DRAM market is evolving rapidly, with traditional DRAM commodity products diminishing in importance as greater emphasis is placed on specialty architectures.
The DRAM market is thus becoming more complicated and fragmented. The number of special architectures for PCs is increasing at a terrifying rate, but at the same time, non-PC OEMs are continuing to use older types of DRAM.
EDO DRAM, for example, is currently sought by server manufacturers, and will be for the next three years. At the same time, virtually all the desktop-PC attention is going to PC-100 SDRAM, to be followed by Rambus. Some DRAM manufacturers, however, regard double data rate (DDR) as a better choice for the next generation of PCs-further fueling the tendency toward fragmentation of the specialty-DRAM market.
Meanwhile, there are the burgeoning datacom, telecom, printer, and consumer-DVD markets. For networking and printer OEMs, it's gotten to a point that 64-Mbit-DRAM granularity is too big for some applications.
Printer and networking OEMs are scrambling to convert from 256-Kbit x 16 to 1-Mbit x 16 devices. However, 1Mbit x 16 is a little too big for the actual memory size they need. To keep performance high, these systems houses will switch to SDRAMs.
Two years ago, some networking and printer OEMs indicated an interest in a 2-Mbit x 32 organization to allow them to take advantage of the better 64-Mbit cost structure. But DRAM suppliers responded by saying a x 32 would cost them a 30% to 40% price premium. So that idea quickly vanished.
Now, they're back revisiting the x 32. As it turns out, the graphics market shows a strong need for 2-Mbit x 32 SDRAM. When this trend takes on greater momentum, several other markets can use this particular specialty device. Each new system challenge seems to be met with a new architecture proposal, making the problem worse.
The PC industry is currently focused on PC-100, and the idea is to go to Rambus in 1999 or 2000. Yet, all along the way, as the mainstream PC players keep dragging DRAM technology forward, a lot of OEMs and the DRAM aftermarket still demand the older parts.
With all of this going on, it's wise for the purchasing community to continue to nurture supplier relationships that have developed over the years. The more fragmented the DRAM market becomes, the more important that partnership will be. One of the reasons is to assure the OEM that a reputable, full-line DRAM supplier will support its product plans.
The savvy purchasing executive will be wary of niche DRAM manufacturers that may not have sufficient resources to adequately supply OEMs, thus leaving them in the lurch when volume deliveries of new products are needed. Quite likely, part of this specialty-market trend will involve the emergence of niche DRAM vendors that will add extra premiums to their parts but can't compete in the overall cost-effective DRAM market.
Also, purchasing executives should make a special point to work closely with their suppliers with respect to product forecasts. We're now approaching a period when we can either experience a balanced supply-and-demand situation or we can have a shortage this year, even though DRAM market analysts aren't predicting it.
The industry has been in overcapacity for three years, and analysts say we're going to stay that way for at least two or three more. But before you believe this, remember that in the middle of 1995 they said we were going to be in a shortage situation for the foreseeable future, and that DRAM business cycles were over. Everyone knows how accurate that prediction turned out to be.
With the onset of this specialty market, we'll need to build the right mix of products. One day we'll build too many PC-66s and not enough PC-100s, or we'll build too many Rambus parts and not enough DDR DRAMs. The result will be that one device architecture may be in oversupply and may even have slight price pressure, while other parts will be in short supply and their prices may go up in the short term.
-Mark Ellsberry is vice president of marketing at Hyundai Electronics America's Semiconductor Division, San Jose.
Copyright ® 1998 CMP Media Inc.
|
