@Pam
"
the thread seems to ignore (or at least not discuss publicly) in their analysis to figure out if SPSN does have a future-
"
We do not ignore some things - bring on the topic and we
could try to clarify some. Its also known, that most facts
(also from NAND) will be hard to come by. So, its always
a bit interpreting and guessing - thats business here (memory
segment).
"
1. What is the total capacity that serves the NOR industry today?
"
You only have to look at the 3 biggest players, SPSN, Nymonix
and Samsung, the rest will have no chance in the future.
"
2. What kind of captive capacity (200mm vs. 300mm) does SPSN have at the present time and where it would be, say at the end of 2008 or 2009? This is the key to their cost reduction efforts.
"
Seems like you have to do some homework. Look at the latest
PDF's and you will get an answer - its that easy.
"
. Samsung is vowing to become #1 by 2009-10 time frame.
"
Some dreams never come true. They are loosing in NAND and
DRAM looks also not that good, politely spoken - NOR is the
same.
"
Do they have a realistic chance and will they pursue it seriously?
"
Loosing even more money - doesn't sound like a good business
idea!
"
4. How does "performance" of MirrorBit compare with FG NOR, now as well as when they move down the nodes? How does MirrorBit ORNAND/2, QuadBit compare with NAND MLC 2 b/c, 3 b/c? Does the performance deteriorate with density? SPSN supposedly has MB QB at 90nm where are the specs? Is anyone using it today?
"
These are good questions, of course, but hard data are very
tough to get. Keep in mind, that SPECs could be so different,
that we couldn't come to this or that conclusions - the
differences could be and are VERY HUGE.
I have looked into the latest specs for Quad with 90nm for
SPSN. Its your choice to question, whether these things
exist, but because you couldn't see it in public, doesn't
mean it doesn't work or noone is/will using/use it. From
the latest paper, cell size for 90nm Quad is 0,036 with
write speed > 4MB/s and read speed around 33 MB/s. I think
we will read much much more in the next time. Why was SPSN/
Saifun using 90nm? Its clear, that for new stuff, "older"
nodes are a good base to start and SPSN was/is using Foundry
and they give new PRs with SMIC in China, which aren't doing
65nm for us yet. So they choose 90nm. The same design could
be established in FAB25 and I assume, that they have the
65nm one already "in the bank", which should make a 8Gbit DIE
around 80-85mm^2 large. Toshibas 8Gbit Design on 56nm is
around 100mm^2, so with 43-45nm, they are in the region of
65-70mm^2 for the 8Gbit DIE, which is only slightly lower than
SPSN on 65nm.
From the SFUN paper:
A total of 256 sense amplifiers are used for program verify of all 3
levels in parallel. During the verify stage, the original programmed
data pattern is transferred from the SRAM to the bit slice and chooses
to verify against one out of 3 reference voltage levels (data from
level 01 should pass the program verify against reference 01, data
from level 00 should pass the program verify against reference 00
and data from level 10 should pass the program verify against reference
10). The combination of highly parallel verify together with the
multi-level programming approach results in >4MB/s program
speed.
...
The 8Gb die micrograph is shown in Fig. 23.2.7 with its major blocks
highlighted. The manufacturing process is based on a wordline halfpitch
of 90nm, and a BL half pitch of 100nm cell array integrated
with CMOS. Divided by the bit number per cell, this device achieves
bit size of 0.009µm², and die size of 155mm².
"
Fabs JV1 and JV2 were sold to Fujitsu for merely 150MM, partly because of extremely mature nodes but what would be the valuation for JV3 + Fab 25?
"
I have posted a link, where Chartered just bought a old
200mm FAB 1-2 generations behind FAB25 for 250Mio. So its
easy to see, that if SPSN voted to sell FAB25, the amount of
money should be in 350-400Mio. region alone.
"
6. When one looks at SPSN's operating model, it is clear that they cannot have positive operating margins unless they have significant increase in Sales as well as GMs
"
Plain and simple wrong, do the math again.
"
7. Do SPSN investors know how much 32Mb, 64Mb and 128Mb NOR sells for today and how much pressure there is from alternative solutions at much lower cost?
"
Why care about "low-end". Just look at the mirrobit progress
in overall SPSN volumes and look at the node which SPSN is
using. Now look at all the smaller guys. Do you see, what
I mean?
"
8. Numonyx deal is still getting finalized but does one know what Fabs are included in the deal from both INTC and STM? How much of that is 200mm vs 300mm?
"
These are no unanswered questions. The answers are in that
board, you have to do again homework, instead of crying out loud.
BUGGI |
