Fabeyes: here are some "projections" bandied around:
DRAMs -- Manufacturers beginning to climb out of a deep canyon
Mar. 26, 1999 (Electronic Buyers News - CMP via COMTEX) -- After a
three-year plunge, DRAM suppliers are beginning to climb back. Global
sales, which dropped in 1998 to $13.9 billion, compared with more than
$40 billion just three years earlier, will rise this year to $19.1
billion, according to Semico Research Corp. Sales will continue to
recover over the next three years, increasing to $27 billion in 2000,
$37 billion in 2001, and $40.2 billion in 2002, the Phoenix-based
research firm said.
During the down year of 1998, DRAM suppliers shipped 3.4 billion
units; but during this year's expected upturn, unit shipments will
decline to about 3 billion and be in the 3.5 billion range for the next
three years, Semico estimates. A major reason is that ASPs are expected
to rise for higher-performing DRAMs, as demand from PC OEMs exceeds a
restricted supply resulting from lack of sufficient investment by DRAM
makers in newer process technologies of 0.22 micron and below.
Well, they got 1998 more or less right, I am not so sure about the forecast nor the assumption that the industry has not adjusted to design rules of .22 microns and below. It seems to me that by the end of this year MU should be finished with the transition to .21 and is already working on .18 microns. Hyundai is introducing RDRAM in "quantities". That by itself dictates design rules well under.25 microns.
Samsung Semiconductor Inc. "plans" include: transition to 0.18-micron process technology; introduce high-density parts such as 256-Mbit Direct Rambus DRAM and 128- and 256-Mbit PC133 and DDR SDRAM. Samsung will complete construction of Fab 9 in Kiheung, Korea, dedicated to production of 256-Mbit SDRAM and Direct RDRAM.
Mitsubishi plans include ramp up 128-Mbit DRAM and PC133 SDRAM; start Direct RDRAM mass production; introduce DDR SDRAM; move to 0.18-micron process technology; start 256-Mbit mass production; and introduce a new 3D-RAM architecture.
NEC's plans Direct RDRAM, and PC133 SDRAM.
It seems that most of the "leaders" are well on their way to create the next glut. Some of the Japanese are actually starting to hedge their planning, initially they were planning on RDRAM to carry a premium of 50% over PC100, but now they are not that sure.
I do not see how we go from 1998 $14 B with 3.5 B chips shipped to $40 B in 2002 with the same number of chips shipped, even if we assume that these future chips will have higher ASP due to higher performance and bit density. After all, we did go down from $40 B to $14 B in three years increasing chip shipments and performance. In any event, if these numbers are correct and in 2002 we should assume at least 50% RDRAM, then RMBS could by then have some $7/share in earning just from DRAM and another $3 to $5/share from non DRAM business. What can MU best case be? let's even assume that they really succeed and get 25% of the market by 2002, and that they bring to the bottom line an outrageous 10% of sales (when was the last time they did that?) and have by then 300 MM shares, all, IMHO best case scenarios (Earlie, Skeeter, don't yell at me <VBG>), I still come back with only $3.3/share earnings. I would take RMBS over MU under such circumstances any day.
Zeev
Zeev |
