Madshrimps Forum Madness - Slow DDR-II adoption makes Dramurai fret

DDR-II memory will finally begin to take over from DDR-I as the DRAM mainstream by the end of this year, manufacturers say, but the transition period could be the final straw for some beleaguered companies.
Adoption of the new memory technology has been far slower than expected: DDR II is about six months behind the schedules predicted a year ago.

Blame high prices and a lack of perceived customer benefit for the slow growth. "At the moment, what's the point of using DDR-II instead of DDR-I," says Jason Han, Sales and Marketing Department Manager at motherboard and SFF manufacturer, Biostar, "there's no difference for most customers."

In addition, industry sources say, AMD, which controls about one fifth of the global desktop CPU market, will not support DDR-II until early next year, further slowing its adoption.

So manufacturers have to time their shift to the new memory standard very carefully. By maintaining dual production lines, offering DDR-I and DDR-II products simultaneously, companies try to keep feet in both camps. But if they misread the market, they could find themselves with one foot in the grave.

"If you produce a lot of DDR-II mainboards, but customers cannot get the supporting DDR-II modules, that will be a nightmare - your mainboard cannot sell", says Lawrence Chang, sales and marketing VP at memory module manufacturer, Kingmax.

DDR-II chips require new testing machines, at a cost of US$6 million each and – in many cases – expensive new SMT machines for BGA-packaged chips. These extra costs are hurting chip package and test companies as well as memory module manufacturers, but the slow growth in DDR-II demand means they have so far seen little return on their investment.

DRAM manufacturers themselves have to absorb the costs of low yields while they ramp up production. It takes months to fine tune lines to produce optimum yields.

Don't expect things to get better in the short term either. DDR-I prices are actually expected to fall compared to DDR-II, due to oversupply in the coming slow season, further decreasing the attraction of DDR-II for end users.

Further complicating things, particularly for motherboard manufacturers, is the ongoing introduction of PCI Express and BTX. Of course, motherboard makers have always had to worry about making their CPU socket choice match market demand. But there will be at least three more ways they can cut a new product off from the market sweet spot during 2005. These include the wrong choice of memory, graphics, and form factor.

What does this all mean for companies that are already overstretched by cutthroat competition amidst an ongoing downturn? As usual, those few that are smart enough or fortunate to time their transition just right will reap windfall profits and come out ahead. However, those that are too early or too late with new products will suffer serious losses in addition to the costs of the transition itself. Expect to see products dumped at low prices – end users, at least, will see some bargains.

In the DRAM market alone, insiders tell me that as many as four significant chip manufacturers could be lost, to mergers or worse, by the end of the year. And the motherboard and memory module manufacturing sectors will also see casualties. Nobody wants to name names yet, but watch for shifting industry alliances, as companies try to choose suppliers who will still be around in a year's time. µ

http://theinquirer.net/?article=21333