Tony & Intel Investors - Intel Modifies their Flash Manufacturing Strategy
Here's a report from Dataquest describing a modified (AGAIN) Flash Manufacturing strategy -
which may be MORE AGGRESSIVE than the previous strategy.
Note the final comment about AMD.
Paul
{===============================}
Intel Shifts Flash Manufacturing into Overdrive
Intel has quietly made another major shift in its flash memory
manufacturing strategy. In case readers missed the last
episode, here is a re-cap: In 1996, Intel announced with great
fanfare the first "memory only" fab in years, to be constructed
with government assistance in Israel for a total cost of about
$1 billion. Late last year, this plan shifted with the
acquisition of Digital Equipment Corporation's chipmaking plant
and a more realistic forecast of PC unit sales. Intel said it
would defer a new fab in Texas and use Fab 18 in Israel for
microprocessors instead of flash chips. Intel still needed the
flash capacity, so it decided to invest the floating $1 billion
in Fab 9, located in New Mexico, the United States, using a
slightly modified version of its 0.25-micron logic process.
Now, Intel is saying that it has changed its mind again. It
will ramp up another New Mexico fab, Fab 11, currently making
Pentium processors with a 0.25-micron process. Since 0.18-
micron logic capacity is increasing more rapidly than forecast,
Fab 11 at a quarter micron will have space for flash, but it
still needs a capital infusion of $1 billion. Since Fab 11 is
more advanced than Fab 9, Intel gets "better bang for the buck"
and can use the money in the processor wars. Speaking of 0.18-
micron technology, the company says the internal development
samples of flash devices with this lithography should come by
the end of this year.
As seen in Table 2, Intel is ramping Fab 11 quickly and
believes it will beat internal goals to have 30 percent of bit
be delivery with the new facility, well ahead of its nearest
competitor, the Advanced Micro Devices Inc. and Fujitsu joint
venture, FASL, in Japan. Although unlikely, on paper Intel
could supply the entire worldwide demand for flash memory from
Fab 11 if it were converted exclusively to memory! What
probably would happen is drafting off another logic fab, say in
Ireland.
Table 2
Intel Flash Fab Capacity (Millions of Megabytes)
==============================================================
1996 1997 1998 1999 2000
Fab 9/Fab 11 13 36 62 167 395
Sharp 18 38 49 51 51
Fab 7 54 56 90 97 82
Total 85 131 200 315 528
Difference (%) 55 53 58 67
==============================================================
Source: Intel, Dataquest
The bottom line here is that Intel is serious about staying a
major player in the flash market and knows that the only way to
do this is reduce its chipmaking cost so it can outprice the
competition in key deals. Doing this at the fab level, in
addition to new technologies like multilevel cell (MLC), will
give it an advantage in a very competitive market. Watch out,
AMD! |
