Semiconductor Business News, c 1998, CMP Media Inc.
September 15, 1998
DRAMs looking at big shakeout
Four vendors may end up with most of global market
By Jack Robertson and J. Robert Lineback
Once again, the semiconductor industry's most watched market segment is going through a fundamental reshaping. Hit by nearly three years of free-falling memory prices and a sea of red ink, the DRAM business is now being changed dramatically by a series of cutbacks and consolidations that will probably end with just four suppliers controlling more than two-thirds of the world market.
The largest convulsions are coming now from the nation that took over the DRAM business from the Japanese during the last industry upheaval. The government of South Korea is forcing a merger between the semiconductor operations of Hyundai Electronics Industries Co. and LG Semicon Co.
If these two divergent chip makers can survive this shotgun marriage, the merged company would be starting off with a DRAM market share of nearly 18% -- as big as any supplier in the world. It would be roughly equal to the DRAM sales of its Korean rival Samsung Electronics Co. and to that of U.S.-based Micron Technology Inc., which is expected gain an 8% share of the world's dynamic RAM sales when it completes its $800 million purchase of Texas Instruments Inc.'s memory business (see story in the July 1 publication).
The Korean merger and Micron's acquisition of TI's memory business come at a time when many Japanese chip suppliers are now opting to scale back in DRAMs because of their rapidly growing losses and the nation's serious financial problems.
Other DRAM merchants also are finding it difficult to maintain investments as the market size continues to shrink due to low average selling prices for both 16- and 64-Mbit memories. Germany's Siemens AG, for example, is now getting ready to close down its one-year-old DRAM fab in North Tyneside, England, because it has to staunch its losses coming from a weakening memory business.
"It looks like we are entering a period when companies must be either fully committed to DRAMs or they're out," observed analyst Bill McClean, president of IC Insights Inc. in Scottsdale, Ariz. "The old strategies of limiting exposure to DRAMs are no longer viable in the long run," he said, referring to the market share targets set earlier by a dozen smaller suppliers that aimed to gain and hold between 5%-to-10% of the global business.
The way the market looks now, IC Insights predicts that Samsung, Micron, and the proposed Hyundai/LG Semicon combination could each end up holding about the same size market share in DRAMs next year -- or between 16% and 19% of the global business. The market researcher estimated that NEC Corp. would most likely be the fourth-ranked supplier with about 12% of DRAM revenues. Outside the four leaders, other suppliers would most likely each hold less than 10% of the DRAM market, McClean expected.
IC Insights is now expecting a solid DRAM recovery next year after this year's disaster. The market researcher predicts sales to climb by around 23% to $15.9 billion in 1999. This year DRAM sales are predicted to fall 35% to just $12.9 billion.
Other industry observers and some chip managers agreed with McClean that the stage is set for another shakeout that would fundamentally change the DRAM business.
Over the coming year, second- and third-tier DRAM suppliers could find it increasingly difficult to compete against the new "Big Four" memory vendors for large OEM contracts, said analyst Jonathan Joseph with NationsBanc Montgomery Securities in San Francisco. Many of the smaller DRAM merchants, he said, could be forced to play mostly in the highly volatile spot market. And that would not provide them enough stability to continue making the huge investments needed to build next-generation wafer fabs.
"You can draw a parallel between [what's now happening] in our business and the PC industry," commented Avo Kanadjian, vice president of memory marketing for Samsung's U.S. semiconductor subsidiary in San Jose. "All of the recent downsizing [of production activities and investments] suggests that if you are not committed to being a big player, you might not want to participate at all," he said.
Agreeing somewhat with the Samsung executive is Kipp A. Bedard, vice president of corporate affairs at Micron in Boise, Idaho. "To have four large suppliers will make a difference from the buyer point of view -- fewer phone calls to make -- and if you're not doing business with one of the big suppliers it could limit your ability to get parts in a major way," Bedard speculated. "If there were 10 or 15 suppliers [relatively equal in size] a buyer could always call the next one."
While Bedard believed that fewer, larger DRAM makers could help to bring some stability back to the memory business, he isn't sure that will be enough to completely change the troubled marketplace. "Consolidation alone will not change anything," he said. "It will take a change in business philosophy to not spend capital dollars without a payback - something we've not yet seen out of Korea."
Other DRAM managers, particularly those from the smaller suppliers, aren't so sure that the big mergers and acquisitions will necessarily have a major impact on the marketplace. "Every time the water goes down, all of the rocks start showing," said Ron Bechtold, vice president of DRAM operations at Hitachi America Ltd. in Brisbane, Calif. He figured that it would be the overall reduction in DRAM fab investments that would cause the big changes in the DRAM business.
"People are trying to squeeze more bits out of current capacity through engineering resources and adding a new piece of equipment here and a process improvements there," Bechtold said. "That will work to increase capacity over a period of time, but you will reach a point to where that won't work and then you need three years to bring on new fabs."
What is taking plenty of capacity out of the market now, according to many DRAM managers, is a wave of decisions over the past month or so by producers to close down their aging 4-Mbit and even some of their 16-Mbit fabs rather than upgrading them. "The investments are being watched very, very carefully by all the suppliers now, and companies are not putting the money into the new equipment for advanced die shrinks like they were in the past," maintained Bart Ladd, senior product marketing manager at NEC Electronics Inc. in Mountain View, Calif., the U.S. subsidiary of the Japanese chip giant.
Even NEC, which has already committed to expand its Roseville, Calif., fab for 256-Mbit and other future DRAMs, continues to scrutinize its investments in the memory segment while it waits for clear signs of a recovery. "We are not immune to these problems," Ladd added.
For months, industry analysts and chip managers have agreed that the most potent cures for what's now ailing the DRAM business would be a reduction in fab capacity and a major market shift to higher density products capable of serving faster computers. Over production in standard 4-, 16-, and 64-Mbit memories has caused the average price-per-bit to fall by more than 60% a year since 1996, said Samsung's Kanadjian. The normal decline of the price per bit is 20-25% annually, he noted.
"It's hard to believe that so many companies hung in there with DRAMs," the Samsung executive said. "But there was an expectation that a recovery had to be around the corner." There were good reasons for feeling that way. "Historically, an oversupply condition has not lasted for more than a year or two," Kanadjian said. "So everyone has been tightening their belts and trying to hang on."
While more than a dozen DRAM makers tried to stick it out, global market revenues kept shrinking even though unit market growth continued. The number of bits shipped by memory makers is expected to rise 71% this year, but DRAM revenues in dollars are expected to fall 35%, according to IC Insights. Its 1998 forecast assumes a modest recovery in the third quarter followed by a 16% rise in sales in the fourth quarter compared to the same quarter last year.
DRAM prices reportedly have remained flat since the middle of July, which has caused some suppliers to predict optimistically that the worst has past.
IC Insight's McClean is still waiting for signs of a full DRAM recovery, but he's encouraged by what he is hearing now from chip companies. Like others in the industry, McClean believes that market conditions couldn't have gotten much worse than they were in July. "It sure felt like a bottom; it was certainly painful enough," he quipped.
It was during that time that a number of companies began deciding to pull back and close down some of their aging DRAM fabs rather than upgrade them for higher density chip production. Hitachi Ltd. opted to shut down its 4-Mbit fab in Irving, Tex., while Matsushita Electric Industrial Co. Ltd. announced in early September that it would close its U.S. fab in Puyallup, Wash. That factory was making 4-Mbit DRAMs and had been slated for an upgrade. Fujitsu Ltd. also said at the same time that it was closing its UK fab because of glut in 16-Mbit chips.
At the same time that these older plants are being mothballed, new facilities are being delayed. But some analysts maintain that as long as companies attempt to remain in DRAM markets, new capacity could be added just as soon as the slightest hint of a recovery is seen.
"We don't expect to see revenues to grow significantly again until at least two more suppliers pull completely out of the market," said Sherry Garber of Semico Research Corp. in Phoenix.
But some optimistic DRAM managers believe that many existing 16-Mbit fabs will be soon be made obsolete when demand emerges for high-speed memories needed for next-generation personal computers. "If you look at the available bits -- adding 16- and 64-Mbit capacity -- it looks pretty hefty, but when you take out the older 16-Mbit fabs, it's a different story," declared Samsung's Kanadjian.
Some chip managers aren't convinced that the current wave of mergers and combinations will result necessarily in four big DRAM giants dominating the market so thoroughly.
"There are questions about whether the marketplace will allow merged companies to keep their combined market share," said Jim Sogas, director of Hitachi's U.S. DRAM business unit. "For example, if one customer is buying 12% of its DRAMs from one supplier and another 18% from the other, I doubt the customer will would be willing to buy 30% from one large supplier," he maintained. "That's too much exposure."
|
