A note from Electronic News....

UMC to Double 90nm Shipments
Online staff -- 8/22/2005
Electronic News


Claiming the highest number of wafers shipped in 90nm, Hsinchu,
Taiwan-based foundry UMC said today it has shipped more than 100,000
wafers on 90nm process technology.

In July alone, UMC said it shipped more than 10,000 90nm wafers.

The company also said it would double monthly shipment levels by Q4
based on demand. UMC's 90nm production is taking place at both of its
300mm fabs and one 200mm fab, with over 20 different products now being
manufactured.



--
Del Cecchi
"This post is my own and doesn’t necessarily represent IBM’s positions,
strategies or opinions.”

Del Cecchi wrote:
> A note from Electronic News....
>
> UMC to Double 90nm Shipments
> Online staff -- 8/22/2005
> Electronic News
>
>
> Claiming the highest number of wafers shipped in 90nm, Hsinchu,
> Taiwan-based foundry UMC said today it has shipped more than 100,000
> wafers on 90nm process technology.
>
> In July alone, UMC said it shipped more than 10,000 90nm wafers.
>
> The company also said it would double monthly shipment levels by Q4
> based on demand. UMC's 90nm production is taking place at both of its
> 300mm fabs and one 200mm fab, with over 20 different products now being
> manufactured.
>

All other things being equal, manufacturers would prefer to go after
the highest volume market out there. Intel's desire, I'm sure, is to
get manufacturers to chase their customers and not the customers of
AMD. This is where the effect of the sheer size of Intel really starts
to show. On the average, the same engineering effort goes for a market
that's four times as big and supported by a marketing organization
that's proportionally even larger.

RM

"Robert Myers" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> Del Cecchi wrote:
> > A note from Electronic News....
> >
> > UMC to Double 90nm Shipments
> > Online staff -- 8/22/2005
> > Electronic News
> >
> >
> > Claiming the highest number of wafers shipped in 90nm, Hsinchu,
> > Taiwan-based foundry UMC said today it has shipped more than
100,000
> > wafers on 90nm process technology.
> >
> > In July alone, UMC said it shipped more than 10,000 90nm wafers.
> >
> > The company also said it would double monthly shipment levels by
Q4
> > based on demand. UMC's 90nm production is taking place at both of
its
> > 300mm fabs and one 200mm fab, with over 20 different products now
being
> > manufactured.
> >
>
> All other things being equal, manufacturers would prefer to go after
> the highest volume market out there. Intel's desire, I'm sure, is
to
> get manufacturers to chase their customers and not the customers of
> AMD. This is where the effect of the sheer size of Intel really
starts
> to show. On the average, the same engineering effort goes for a
market
> that's four times as big and supported by a marketing organization
> that's proportionally even larger.

How many of those 100,000 90nm wafers are for chipsets? Are _any_
chipsets now being manufactured at 90nm?

"Felger Carbon" <(E-Mail Removed)> wrote in message
news:4StOe.382$(E-Mail Removed)...
> "Robert Myers" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>> Del Cecchi wrote:
>> > A note from Electronic News....
>> >
>> > UMC to Double 90nm Shipments
>> > Online staff -- 8/22/2005
>> > Electronic News
>> >
>> >
>> > Claiming the highest number of wafers shipped in 90nm, Hsinchu,
>> > Taiwan-based foundry UMC said today it has shipped more than
> 100,000
>> > wafers on 90nm process technology.
>> >
>> > In July alone, UMC said it shipped more than 10,000 90nm wafers.
>> >
>> > The company also said it would double monthly shipment levels by
> Q4
>> > based on demand. UMC's 90nm production is taking place at both of
> its
>> > 300mm fabs and one 200mm fab, with over 20 different products now
> being
>> > manufactured.
>> >
>>
>> All other things being equal, manufacturers would prefer to go after
>> the highest volume market out there. Intel's desire, I'm sure, is
> to
>> get manufacturers to chase their customers and not the customers of
>> AMD. This is where the effect of the sheer size of Intel really
> starts
>> to show. On the average, the same engineering effort goes for a
> market
>> that's four times as big and supported by a marketing organization
>> that's proportionally even larger.
>
> How many of those 100,000 90nm wafers are for chipsets? Are _any_
> chipsets now being manufactured at 90nm?
>
If UMC is sipping 10,000 wafers/month and going to 20,000 then some of
them surely could be chipsets. And 90 nm chipsets will be smaller and
cheaper than 130 nm. And if UMC is adding capacity, TSMC and Chartered
probably are too, making capacity available in contrast to the recent
arguments in this group that there was insufficient capacity to pick up
the slack if Intel stopped doing low end chip sets.

"Del Cecchi" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
>
> "Felger Carbon" <(E-Mail Removed)> wrote in message
> news:4StOe.382$(E-Mail Removed)...
> > How many of those 100,000 90nm wafers are for chipsets? Are _any_
> > chipsets now being manufactured at 90nm?
> >
> If UMC is sipping 10,000 wafers/month and going to 20,000 then some
of
> them surely could be chipsets.

I agree completely. Some of them _could_ be chipsets. Does anyone
know for a fact that production chipsets are being produced today in a
90nm process by _anybody_?

> And 90 nm chipsets will be smaller and
> cheaper than 130 nm.

The implication here is not limited to chipsets; if they are smaller
and cheaper, then _any_ product will be smaller and cheaper at 90nm
than at 130nm. The implication is that _no_ product should be built
on a 130nm process because it will be more expensive than at 90nm. So
all the 130nm fabs are covered with cobwebs, right? No? There must
be a problem with this logic then.

> And if UMC is adding capacity, TSMC and Chartered
> probably are too, making capacity available in contrast to the
recent
> arguments in this group that there was insufficient capacity to pick
up
> the slack if Intel stopped doing low end chip sets.

"Felger Carbon" <(E-Mail Removed)> wrote in message
news:ylvOe.383$(E-Mail Removed)...
> "Del Cecchi" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>>
>> "Felger Carbon" <(E-Mail Removed)> wrote in message
>> news:4StOe.382$(E-Mail Removed)...
>> > How many of those 100,000 90nm wafers are for chipsets? Are _any_
>> > chipsets now being manufactured at 90nm?
>> >
>> If UMC is sipping 10,000 wafers/month and going to 20,000 then some
> of
>> them surely could be chipsets.
>
> I agree completely. Some of them _could_ be chipsets. Does anyone
> know for a fact that production chipsets are being produced today in a
> 90nm process by _anybody_?
>
>> And 90 nm chipsets will be smaller and
>> cheaper than 130 nm.
>
> The implication here is not limited to chipsets; if they are smaller
> and cheaper, then _any_ product will be smaller and cheaper at 90nm
> than at 130nm. The implication is that _no_ product should be built
> on a 130nm process because it will be more expensive than at 90nm. So
> all the 130nm fabs are covered with cobwebs, right? No? There must
> be a problem with this logic then.
>
>> And if UMC is adding capacity, TSMC and Chartered
>> probably are too, making capacity available in contrast to the
> recent
>> arguments in this group that there was insufficient capacity to pick
> up
>> the slack if Intel stopped doing low end chip sets.
>
Of course the folks building stuff in 130 nm will likely continue to
build it until demand goes away. New parts will be in 90 nm, like the
hypothetical replacements for the low end chip sets that Intel is said to
be discontinuing. These hypothetical replacements were alleged to not be
possible earlier in this thread due to lack of capacity in the wafer
fabrication business.

Are you trying to deliberately be argumentitive, for some reason? Why?

del cecchi
>

On Mon, 22 Aug 2005 21:10:45 -0500, Del Cecchi wrote:

>
> "Felger Carbon" <(E-Mail Removed)> wrote in message
> news:ylvOe.383$(E-Mail Removed)...
>> "Del Cecchi" <(E-Mail Removed)> wrote in message
>> news:(E-Mail Removed)...
>>>
>>> "Felger Carbon" <(E-Mail Removed)> wrote in message
>>> news:4StOe.382$(E-Mail Removed)...
>>> > How many of those 100,000 90nm wafers are for chipsets? Are _any_
>>> > chipsets now being manufactured at 90nm?
>>> >
>>> If UMC is sipping 10,000 wafers/month and going to 20,000 then some
>> of
>>> them surely could be chipsets.
>>
>> I agree completely. Some of them _could_ be chipsets. Does anyone
>> know for a fact that production chipsets are being produced today in a
>> 90nm process by _anybody_?
>>
>>> And 90 nm chipsets will be smaller and
>>> cheaper than 130 nm.
>>
>> The implication here is not limited to chipsets; if they are smaller
>> and cheaper, then _any_ product will be smaller and cheaper at 90nm
>> than at 130nm. The implication is that _no_ product should be built
>> on a 130nm process because it will be more expensive than at 90nm. So
>> all the 130nm fabs are covered with cobwebs, right? No? There must
>> be a problem with this logic then.
>>
>>> And if UMC is adding capacity, TSMC and Chartered
>>> probably are too, making capacity available in contrast to the
>> recent
>>> arguments in this group that there was insufficient capacity to pick
>> up
>>> the slack if Intel stopped doing low end chip sets.
>>
> Of course the folks building stuff in 130 nm will likely continue to
> build it until demand goes away. New parts will be in 90 nm, like the
> hypothetical replacements for the low end chip sets that Intel is said to
> be discontinuing. These hypothetical replacements were alleged to not be
> possible earlier in this thread due to lack of capacity in the wafer
> fabrication business.
>
> Are you trying to deliberately be argumentitive, for some reason? Why?

Felg, argumentative? Nah! ;-)

I think many percieve chipsets as the low-class chips they once were and
90nm as being something special. Both were once true, but times change.
Chipsets are increasingly ciomplex and have to operate at speeds rivaling
processors. ...and 90nm is almost mainstream today. It's no lonnger
something to shy away from if it buys something.

--
Keith

All I know is that Intel is currently producing all of its own chipsets
on a 180nm process on 200mm wafers. Obviously its one of its older
plants. If other people are selling 90nm chipsets, then they'd be 4
times smaller in area. They wouldn't have needed Intel to voluntarily
get itself out of the way, they would already have pushed Intel out of
the way as they would have a significant volume advantage to Intel.

Yousuf Khan

"Del Cecchi" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
>
> Of course the folks building stuff in 130 nm will likely continue to
> build it until demand goes away. New parts will be in 90 nm, like
the
> hypothetical replacements for the low end chip sets that Intel is
said to
> be discontinuing. These hypothetical replacements were alleged to
not be
> possible earlier in this thread due to lack of capacity in the wafer
> fabrication business.
>
> Are you trying to deliberately be argumentitive, for some reason?
Why?

Del, I can see no evidence that anyone is using a 90nm process to
build production chipsets. If that's wrong, I want to know about it
because obviously I'm missing something. I got the impression that
you believe chipsets _are_ being built on 90mn, because it's cheaper
to do so. I have always believed that chipsets are built on the
_last_ technology (currently 130nm), because it is cheaper to do so.

I respect your knowledge in this area, and am asking for some evidence
that my opinion in this matter is wrong. I always have something new
to learn.

"YKhan" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> All I know is that Intel is currently producing all of its own
chipsets
> on a 180nm process on 200mm wafers. Obviously its one of its older
> plants. If other people are selling 90nm chipsets, then they'd be 4
> times smaller in area. They wouldn't have needed Intel to
voluntarily
> get itself out of the way, they would already have pushed Intel out
of
> the way as they would have a significant volume advantage to Intel.

Yousuf, my Asus K8S-MX boards use the SiS760GX chipset, with
significant built-in graphics. Others - nVidia, for instance - are
likewise building significant graphics in their chipsets. I thought
these were being built in 130nm. Is Intel building anything with
significant graphics in 180nm?