1/9/03 article in business on 300mm
STREET WISE
By Olga Kharif
Unequal Bounces in a Chip Rebound
An expected 2003 recovery would likely just widen the gap between strong memory-chip makers and weaker players
Creditors of Hynix Semiconductor, a money-losing South Korean maker of basic computer memory, approved yet another debt-relief plan on Dec. 30. It's the latest evidence that key players in this highly competitive industry expect to hang on until a rebound materializes. Hoping that information-technology spending will revive in 2003's second half, they're dreaming of a return to profitability. The recovery has been so far elusive, with some of these memory makers booking losses for two and a half years. This particular downturn has been twice as long as any they've ever experienced.
Still, the Big Five DRAM suppliers that control 82% of the market -- Samsung, Micron (MU ), Hynix, Infineon (IFX ), and Elpida -- all say they're in for the long haul. That doesn't mean they would benefit equally from any upturn. The industry's growth, when it returns, is expected to be slower than in the past. If a rebound does arrive, chipmakers' revenues are expected to rise just 6% in 2003, estimates Brian Matas, an analyst with chip consultancy IC Insights. With less growth to go around, the cutthroat competition would intensify.
TWO GOOD BETS. Already, the gap between the haves and the have-nots is widening, says Mario Morales, an analyst at tech consultancy IDC. And only businesses enjoying the lowest costs and offering the most services to their customers will gain in this go-round. Analysts are betting on Samsung and Infineon, both of which are able to pour resources into new chip plants that cost $2 billion a pop -- and some more into R&D. Some of the other players, specifically, Hynix and Micron, will likely be hurting for the cash to keep pace.
This is a bad year to be short on cash: The DRAM industry, already among the most capital-intensive around, is becoming more so. The new fabs are a must, since they're capable of twice the production volume of the old plants by stamping chips out of much bigger wafers -- 300 millimeters across vs. the old 200-mm wafers (see BW, 11/11/02, "Chips on Monster Wafers".) The 300-mm technology already allows Infineon to produce chips for 30% less than it used to in its old fabs, says Peter Schaefer, vice-president for memory products at Infineon North America.
That's a key advantage in an industry with a highly commoditized product. It means Samsung and Infineon could lower their prices and still make a profit in an effort to increase market share while less technologically advanced rivals' losses deepen.
MOST LIKELY DROPOUT? Other competitive pressures are on the rise as well: A decade ago, manufacturers could offer one standard-type DRAM. Today, customers require a broader range of products, such as chips with low power consumption for mobile applications or those able to withstand high temperatures for industrial use. Clients also demand more accommodation: They want DRAM makers to set up warehouses inside their own plants so they can grab parts whenever they're needed.
These demands can stretch an outfit thin and make it difficult to recoup losses incurred in the past few years. For example, Hynix would have a tough time moving into the black even in an upturn, says Matias. The chipmaker disagrees and says it's already near profitability.
For two years now, analysts have predicted that financially strapped Hynix would fold. And they still consider it to be the most likely dropout if things don't improve. In December, Hynix creditors restructured $4.1 billion of Hynix's $6.25 billion debt, but they may not feel up to making new cash infusions. The creditors "simply decided not to pull the plug on life support," says Bear Stearns analyst Charles Boucher.
TECHNO-LAGGARDS. Not so, says Hynix. It hopes to raise $300 million through sales of businesses. The money will go to upgrade fabs in 2003 to make smaller chips, says Farhad Tabrizi, vice-president for worldwide marketing. Hynix also plans to start trying out the new, high-output technology -- which many of its rivals already have -- in early 2004, he says. But it's years behind, say analysts. Infineon began testing the 300-mm technology in 1997 and is already using it to mass-produce chips.
Hynix isn't the only struggling chipmaker. No. 2 Micron, commanding 20% of the market, is also behind on the new technology. And while no one doubts its staying power, the lag could cost Micron market share in 2003, analysts say. It has a pilot product line based on 300-mm technology, and it could be starting mass production of chips based on it soon. Still, analysts say cash pressure could cause further delays. The chipmaker burns through about $300 million a quarter. Yet it had just $656 million in cash and short-term investments as of Nov. 28, according to Micron's quarterly report.
To stay ahead in the game, it needs a major cash infusion -- to the tune of perhaps $500 million to $750 million, estimates Adam Parker, a Sanford Bernstein analyst. Fortunately, that shouldn't be a problem and will likely happen in the current or next quarter, Parker is quick to add. "Micron's business is solid," says a company spokesperson. And he adds that it has the money to make necessary technology upgrades. But "we still don't believe [moving to the new technology] makes sense," he notes, citing the high costs of materials and uncertain demand.
"SCALE IS EVERYTHING." A wait-and-see approach could be prudent if demand continues to disappoint. But it could also put Micron at a disadvantage if IT spending revives in the second half, as now forecast. Meanwhile, Elpida, a joint venture between NEC (NIPNY ) and Hitachi (HIT ), may also find itself at a disadvantage. It's adapting the 300-mm technology and expanding its manufacturing capabilities fast: In April, it will take over Mitsubishi Electric's (MIELY ) DRAM operations.
Still, new technology alone won't solve Elipda's problems. Even with the acquisition, its market share hovers around 7%, estimates Morales. "That's not going to be enough. In this industry scale is everything," he says. Morales believes that players with market share smaller than 10% will be at a major cost disadvantage. Elpida couldn't be reached for comment.
Samsung expanded its market share to more than 30% in 2002, from 28.7% at the end of 2001. And on Dec. 31, it announced that it would invest $1.2 billion to build another high-capacity fab in South Korea, due to begin production in 2003. "We invest in good times and bad," says Tom Quinn, vice-president for marketing of memory products at Samsung Semiconductor.
TOUGH YEAR AHEAD. The same holds true for Infineon, which has the most new fab capacity of any memory maker, estimates George Burns, president of chip-capacity consultancy Strategic Marketing Associates. Infineon should gain several more points of market share in 2003, say analysts, up from more than 11% today. "We're one year ahead of the closest competition" when it comes to lowering costs, claims Schaefer.
Any way you look at it, DRAM makers face another tough year ahead. Even if a rebound occurs, this could be a time when the stronger players make big gains in market share, while the weaker players confront even more daunting challenges to their survival.
businessweek.com
Jack |
