Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/umc-falls-out-of-14nm-race.6097/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

UMC falls out of 14nm race

TSMC should recognize that perhaps its time to join forces with UMC. With Global Foundries joining forces with Samsung (and paying for a portion of their 14 nm development costs), TSMC could license their 16 nm process node to UMC in return for money & perhaps a joint development of the 7 or 10 nm node.

I remember UMC turning down the opportunity to license Samsung's 14 nm node about a year ago. I got to wonder if part of UMC's reason was because of TSMC.
 
Well, while the competitors are already developing their 10nm offers, UMC is moving to 18nm, whatever that means (most likely a kind of 28nm + FinFET).
Bye bye leading edge.
It is clear that their only option to compete is by offering lower prices (that is exactly what happened at 28nm).
 
astilo said:
Well, while the competitors are already developing their 10nm offers, UMC is moving to 18nm, whatever that means (most likely a kind of 28nm + FinFET).
Bye bye leading edge.
It is clear that their only option to compete is by offering lower prices (that is exactly what happened at 28nm).
Click to expand...
From the wording of the article it's not FinFET: i.e. they mention "14nm FinFET process", but "18nm process" (no FinFET)
I remember UMC turning down the opportunity to license Samsung's 14 nm node about a year ago.
Click to expand...
The attractiveness of that agreement depends on the profit sharing. Is it 50-50, or maybe 80-20 profits to Samsung?
 
lefty said:
From the wording of the article it's not FinFET: i.e. they mention "14nm FinFET process", but "18nm process" (no FinFET)
Click to expand...
Neither they stated 18nm planar process. Do you believe that it could really be 2D? No way, without FD-SOI or FinFETs, 18nm is just a suicide.

lefty said:
The attractiveness of that agreement depends on the profit sharing. Is it 50-50, or maybe 80-20 profits to Samsung?
Click to expand...
And by the assumption that UMC is able to make 14nm yielding, that is not trivial at all.
 
Last edited:
lefty said:
UMC has released that it doesn't have the resources to develop it's own 14nm process, instead it'll introduce a 18nm node:
UMC developing 18nm process technology
Thoughts?
Click to expand...

Interesting times for engineers; tougher times for people who have to make the business decisions :).
I assume all layers (except maybe poly cut) will be single patterned then. Good for mask cost, a little bit worse for cost/transistor. If not FinFET I think this process would also then be a perfect candidate to have a FDSOI version of it; e.g. a ST-UMC joint development program if such a thing is possible from company culture stand point.
 
Last edited by a moderator:
prime007 said:
TSMC should recognize that perhaps its time to join forces with UMC. With Global Foundries joining forces with Samsung (and paying for a portion of their 14 nm development costs), TSMC could license their 16 nm process node to UMC in return for money & perhaps a joint development of the 7 or 10 nm node.

I remember UMC turning down the opportunity to license Samsung's 14 nm node about a year ago. I got to wonder if part of UMC's reason was because of TSMC.
Click to expand...

I proposed this a while back:

https://www.semiwiki.com/forum/content/4429-tsmc-umc.html

Unfortunately, from what I was told there is some history between TSMC and UMC that will prevent this from happening anytime soon.
 
Thanks Danni - I remember reading your article...although I didn't know about the "history" between TSMC & UMC.

As you stated times are changing...TSMC needs to recognize that Samsung (and Intel) isn't to be under-estimated.
 
prime007 said:
Thanks Danni - I remember reading your article...although I didn't know about the "history" between TSMC & UMC.

As you stated times are changing...TSMC needs to recognize that Samsung (and Intel) isn't to be under-estimated.
Click to expand...

If you look at the history the same was said about GlobalFoundries challenging TSMC at 28nm. Do you remember the whole Common Platform thing? What a tremendous failure that was. So maybe TSMC had a right to feel overconfident at 16nm.

I'm not sure what UMC has to offer TSMC other than a trusted second source. TSMC called GlobalFoundries the "Samsung accessory" after the 14nm licensing deal. I felt it was truly a big step forward for Samsung in the foundry business but if they do not continue at 10nm and 7nm it was all for naught in my opinion.

My guess is that UMC 18nm, if that is a real thing, will be planar and it will fail. GlobalFoudnries has a 22nm SOI platform that would be much more competitive if you are going for a non double patterning process:

https://www.semiwiki.com/forum/content/4630-asmc-2015-globalfoundries-22nm-soi-plans-more.html

TSMC can keep some 20nm capacity available for the planar people. It is extremely high yield and already paid for by Apple.
 
Last edited:
You must log in or register to reply here.
Back
Top