FAB25 Sale
Thx@Linux@WO for the link.
Analyst report: Spansion seeking a buyer for its Austin chip plant
statesman.com
I have written about this option already months ago and I
hope this is right. There are two main critics for FAB25.
It works on 200mm and still mostly on 90nm, 65nm on its way
and (for me) unknown how far this process is done. When I
look at the fresh SMIC partnership on 300mm down to 4x nm,
it should be clear, which option is better. I said this
week, that Spansion should produce 100% with own capacity
at the weakest quarterly level and the seasonal additional
stuff at foundrys. They could also produce less of course,
but they have to make sure, that own capacity works 100%
utilized, the higher the better. Fixed costs are THE highest
hurdle to jump over in this business and owning FABs these
days isn't that easy.
In a prior post I have written about the transition time. It
should be clear, that once FAB25 is "gone", Spansion could
work through inventory, but they have to plan with a big
foundry to ship necessary products and these will be huge
huge numbers, thats clear. I'm not sure, whether all (and
older) products could be changed to new stuff as well -
probably not, so that will be a monster mission, if they will
do it.
I my opinion, the should go this path - which will mostly
depend on the money, they will get. First, I would equip
SP1 up to 4K the faster the better. This will create much
capacity and will of course decrease our overall cost level
there. Imagine, what 2K extra wafers on 300mm should do for
us! Its relativly easy to see, that these 2K wafers should
nearly reach FAB25 capacity depending on the process. Lets
assume SP1 is 1 node ahead =>
FAB25 - Node N+1
8-10K Wafers/months => I take 10K
SP1 - Node N+0
2K extra Wafers => 2x2,5= 5K 200mm Wafers x better Node.
Instead of 2x for a better node, I will choose 1,5x. This
means 2K SP1 extra wafers will produce ROUGHLY 7-8K FAB25
wafers -> so all SPSN has to do is ship these 2-3K extra
wafers production from inventory + foundry. I will strongly
opt for a decreasing inventory level, as said a few times
already => this will buy some time, probably 1-2 quarters.
The question is, how fast could SPSN buy SP1 equipment, once
a FAB25 is done and how fast they could bring this online?
Thats a big question and its still unknown, how much they
could bring this SP1 project forward? Would be nice, if equip.
is ordered and already installed in Q1/Q2 => if not already
planned => unknown.
What are the implications? Driving fix costs way down.
Getting away from 200mm and "old" nodes. Building SP1 more
out to reach 100% in the near future and be sure, foundry
could deliver products on 300mm and new nodes, the lower the
better. With TSMC reaching low utilisation levels in the
next quarter and some foundrys in relativ "bad" shape, it
shouldn't be that hard to get fresh capacity if needed. Should
be welcomed by foundrys. I strongly opt for some 45nm 300mm
if this could be done. SMIC, TSMC -> take this route. Drive
down fixed costs as much as possible and be sure, SP1 works
as efficient as possible. Keep sure, that foundrys work on
ALL processes which will be needed. Keep sure old customers
on old designs are still happy.
I think we will hear about that in the next time. Its time,
that SPSN reacts to a very tough environment and its finally
the time, to benefit from strong products and the best
process for the future with a complete charge trapping port-
folio.
BUGGI |
