Tommy Parks,
The AMD bulls on this thread believe that most of the analysts are not
dealing with reality.
For example, take Soundview Financial. Here's their estimates for #
of K6 processors in '98
Q297A Q397A Q497A Q198A Q298E Q398E Q498E
486 units 0.725 0.5 .2 0.1 0.05 0 0
K5 units 0.7 0.4 .1 0.1 0.0 0 0
K6 units 0.375 1.0 1.5 1.55 2.1 2.7 3.5
Lets look, for example, at the 2.7 million and 3.5 million estimates
for # of K6's in Q3 and Q4. It takes 9-12 weeks from when a wafer is
started to when a sale of finished processors is booked, so the number
of CPU's produced in Q3 is determined by the wafers being started NOW.
That number is 3,000 per week, half of them on the 0.25u process and
half of them on the 0.35u process. AMD is getting at least 140
good CPU's from each 0.25 wafer -- this figure has been verified by at
least 3 analysts. Lets be pessimistic and assume yields are only 40%
in the older 0.35 process, 60 good CPU's per wafer.
AMD said the first week of May would be the last week they start any
0.35 process wafers. So the total 0.35 wafers will be 4 weeks-worth,
or 6,000 wafers, which is 360,000 K6 (0.35u) CPU's.
For the 0.25 micron output, they have 4 weeks with 1500 wafers
followed by 9 weeks averaging 2500 wafers (allowing for equipment
turnaround time). This is a total of 28,500 wafers, or a total of
3,990,000 K6 (0.25u) CPU's.
So the grand total is 4,350,000 K6's, mostly 300 MHz or faster K6's
and K6-3D's, over a MILLION MORE CPU's than Soundview's estimate.
I didn't even mention Q2 or Q4, but you get the picture.
Petz |
