To: Norrin Radd who wrote (4721 ) 2/27/1999 3:44:00 PM From: DJBEINO Respond to of 9582
DRAM margins to rise modestly Vadim Zlotnikov; Sanford C. Bernstein The dramatic underinvestment in DRAM capacity during the past three years has clearly caused prices to stabilize. This will drive margins upward for vendors, particularly those with advanced manufacturing capacity. DRAM capacity above 0.25-micron, however, currently accounts for nearly half of total wafer starts. That is not economically viable for these companies, since the $10 price tag on a 64-Mbit device is slightly below the variable cost of production in the older fabs. The cycle upturn will last at least through 2000, because it will take at least two years of aggressive investment to accelerate capacity growth. Peak margins for the DRAM industry during this cycle will approximate those of the 1984 and 1988 upswings, but will fall short of the extraordinary profitability achieved in 1995. This view may be counter-intuitive, given the higher operating leverage of the DRAM companies and a likelihood of much greater market-share concentration. But the rationale is twofold. First, there are fewer opportunities for DRAM vendors to displace the value-add of other suppliers. DRAM vendors' high margins in 1994 and 1995 were due to an increase in DRAM content per PC from $140 in 1993 to $289 in 1995. Today, DRAM accounts for roughly $80 per PC. If DRAM prices are flat in 1999 and bits-per-PC growth is 45%, nearly $40 will be added to a PC OEMs' bill-of-materials. Since growth in the number of new PC users is occurring in the price-sensitive portions of the market, PC ASPs are likely to decline by 6% to 7% in 1999-or $100-vs. a 10% to 11% decline in 1998. Thus, the DRAM content increase of $40, plus a price decline of $100, would need to come out of PC vendors' margins or other components. But pricing on flat-panel displays, microprocessors, and operating systems is relatively fixed, and PC vendors do not have the option of sacrificing their razor-thin margins. So, the combination of stable DRAM pricing and 45% bit growth is unlikely to occur. In fact, discussions with PC vendors suggest that most intend to keep DRAM and disk-drive content from increasing as a percentage of total PC cost. Bit-demand growth will be modulated to avoid DRAM shortages and price increases. In fact, DRAM dollar growth is highly unlikely to exceed PC unit growth. This is not the 1995 scenario; the result will be lower margins. The second reason for lower industry margins is the absence of a marginal high-cost producer. Since pricing is set by the marginal vendor, industry consolidation and the "higher" level of competition among leaders will actually be detrimental to vendors' margins. Copyright ® 1999 CMP Media Inc.techweb.com