Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/samsung-begins-14nm-contract-chip-production.4946/
        )

    [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] => 2021370
            [XFI] => 1050270
        )

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

Samsung begins 14nm contract chip production

astilo

New member
Samsung begins 14nm contract chip production | ZDNet


Kim Ki-nam, president of Samsung's semiconductor business and head of System LSI business, said production is going very well for its 14-nanometre FinFET foundry. However, Kim did not say which client it is producing the chips for.

Any guess?

The two likeliest candidates for Samsung's first batch of 14-nanometer chips are Apple, which has reportedly begun chip orders for its S1 system in a package that will power the Apple Watch, and Qualcomm, the biggest fish in application processors (AP), which has inked a deal with the South Korean tech giant for the latter to produce next-generation APs. The chips could also potentially be for AMD's new GPUs.

The S1 made at 14nm would be overkill to me. Qualcomm next Snapdragon 810 is manufactured at 20nm. AMD should not show any 14nm product before 2016. Well, maybe it is a Samsung chip :)
 
Most likely, the production is for Samsung’s own Exynos chips.

Apple’s orders are massive. To produce for Apple, Samsung must expand its 14nm capacity in a big way. I’m not aware of any large delivery from equipment makers, such as AMAT, to Samsung. No sign of such ramp.

In contrast, a lot of equipment purchases and deliveries are happening at TSMC. Just do a search on "TSMC" and "equipment".

On Dec 4, TSMC did a supply chain forum and honored some companies with awards, including AMAT and KLA-Tencor.

2014/12/04
TSMC Recognizes Outstanding Suppliers at Supply Chain Management Forum
 
Coincidentally, I spent quality time with the Samsung Foundry people at the GSA awards event last night in addition to many other semiconductor professionals. I don't know the exact attendance number but there were 160 tables of 10 set for dinner. Jay Leno was the headliner. Even more interesting, Morris Chang won the Morris Chang Leadership award. Read into that what you like. :rolleyes:

As I understand it, from multiple sources, Samsung 14nm is yielding much better now than last quarter so yes I think they will hit the 1H 2015 production requirements of the top fabless companies. I highly doubt the iWatch is 14nm but I will check into it. My guess is Samsung 28nm but that really is a guess. Unlike my guess that Apple will use Samsung 14nm for the next iPhone and TSMC 16nm FF+ for the iPad.
 
Another interpretation is that Samsung’s announcement is 14nm risk production. The result of this risk production will determine how many, if any, of A9s will be built at Samsung, while TSMC has started ramping up 16FF+ for the mass production of A9s.

I maintain my opinion that most, if not all, of A9s will be produced at TSMC. See my reasoning in my prior posts:

https://www.semiwiki.com/forum/f2/samsung-strikes-chip-deal-apple-4864.html#post16870

https://www.semiwiki.com/forum/f2/samsung-strikes-chip-deal-apple-4864-2.html#post16918
 
Well, I´m more convinced now that we are really talking about the Apple A9 new chip.

2 things are anyway now very clear to me: the A9 will be built on 14nm (we can close the 20nm speculation) and Apple is developing it at both Samsung and TSMC (no surprise, but not an obvious thing either).
That said, there a couple of reasons why I think Samsung could get the biggest slice of the cake:
First, Apple prefers to have the production in US, and the Samsung Fab in Austin is perfect for that. They could also partially use the GloFo Fab8 in Malta (copy exact process), to mitigate the supply chain risk.
Second, Samsung seems again to be able to deliver the chips earlier than TSMC and probably, on paper, the performance should also be better (at least when compared to the 16FF).

Apple is going to get a 14nm SoC in the market, probably even before Intel, and by far, sooner than QTI, NVIDIA and AMD. Another KO punch.
 
Well, I´m more convinced now that we are really talking about the Apple A9 new chip.

2 things are anyway now very clear to me: the A9 will be built on 14nm (we can close the 20nm speculation) and Apple is developing it at both Samsung and TSMC (no surprise, but not an obvious thing either).
That said, there a couple of reasons why I think Samsung could get the biggest slice of the cake:
First, Apple prefers to have the production in US, and the Samsung Fab in Austin is perfect for that. They could also partially use the GloFo Fab8 in Malta (copy exact process), to mitigate the supply chain risk.
Second, Samsung seems again to be able to deliver the chips earlier than TSMC and probably, on paper, the performance should also be better (at least when compared to the 16FF).

Apple is going to get a 14nm SoC in the market, probably even before Intel, and by far, sooner than QTI, NVIDIA and AMD. Another KO punch.

So you are saying that Apple taped out the A9 design at both TSMC 16FF+ and Samsung 14nm? They have never done that before by the way. What about the A9x for the iPads? Will they manufacture it at both Samsung and TSMC? Do you understand the additional costs and resources associated with this scenario?
 
I can´t confirm it, so take it as my personal opinion or a speculation, but I believe they are doing that, including the A9X. Basically Apple will have the choice eventually to pick the best option, or maybe the best price, and also splitting production between fabs and foundries.
If I´m wrong, then the A9 will be only manufactured by Samsung, since in order to hit the shelves with a mass production in September, TSMC should start with the real production at latest end of Q1-2015. They could be ready by Q2, so they can only get a slice of the A9 or maybe the A9X. The only hope for TSMC was the Samsung low yield issue at 14nm. Now the yield seems to be good enough, so there is no reason for Apple to delay the new Iphone or to stay at 20nm (they need something new to sell phones after everybody got already a bigger screen one). I also believe, that the additional cost of this scenario, won´t be paid only by Apple, but indirectly, also by the foundries. QTI is doing second sourcing since years now and is able to manage very good prices.
 
"First, Apple prefers to have the production in US, and the Samsung Fab in Austin is perfect for that. They could also partially use the GloFo Fab8 in Malta (copy exact process), to mitigate the supply chain risk."

Can you explain a little bit why "
Apple prefers to have the production in US"? I believe all Apple products (at least majority of them) are assembled in China or Taiwan. Making processors in USA doesn't give Apple any benefit in terms of supply chain efficiency. Actually it might increase the lead time needed and transportation cost.
 
Astilo,

Your entire argument, except the Austin fab being in the US, merely repeats Samsung’s PR, which is known to be pervasive and misleading, similar to Intel’s highly dubious claim of density superiority. It’s a sorry situation that PR, instead of innovation, has become the emphasis of Samsung and Intel.

You ignored the more reliable evidences, such as ramping at TSMC, but not at Samsung.

Anyway, no amount of PR can sway the reality. The truth will unveil in 6-9 months.
 
I can´t confirm it, so take it as my personal opinion or a speculation, but I believe they are doing that, including the A9X. Basically Apple will have the choice eventually to pick the best option, or maybe the best price, and also splitting production between fabs and foundries.
If I´m wrong, then the A9 will be only manufactured by Samsung, since in order to hit the shelves with a mass production in September, TSMC should start with the real production at latest end of Q1-2015. They could be ready by Q2, so they can only get a slice of the A9 or maybe the A9X. The only hope for TSMC was the Samsung low yield issue at 14nm. Now the yield seems to be good enough, so there is no reason for Apple to delay the new Iphone or to stay at 20nm (they need something new to sell phones after everybody got already a bigger screen one). I also believe, that the additional cost of this scenario, won´t be paid only by Apple, but indirectly, also by the foundries. QTI is doing second sourcing since years now and is able to manage very good prices.

The press is still saying that Apple split the A8 between TSMC and Samsung, which is not true, so the rumors of Apple splitting the A9 will continue. Unfortunately, so many people touch the SoC so the truth is known in Silicon Valley. Each process has a unique PDK (Process Design Kit) that must be verified with specific EDA tools and IP, the tools and IP must be licensed for each process node, and there are always problems so even the tech support people know who is designing what and where.

As I have mentioned many times before, the fabless semiconductor industry is all about innovation and innovation requires competition and you will not find a more competitive ecosystem. If Apple, QCOM, and the other top fabless companies do NOT use multiple manufacturing sources innovation will be stifled. Pricing is also a deciding factor. The fabless semiconductor company margins are at risk with only one manufacturing source. Bottom line: As long as multiple semiconductor manufacturing resources are available they will be utilized, absolutely.
 
Hi Daniel, just to better clarify your post, what´s your guess about the A9? Single design at Samsung/TSMC, or double release?
 
Hi Daniel, just to better clarify your post, what´s your guess about the A9? Single design at Samsung/TSMC, or double release?

My guess: The A9 (iPhone) Will be Samsung 14nm LP and the A9x (iPad) will be TSMC 16FF+. I do not think Apple will break with tradition and split a design amongst multiple foundries.
 
The 14nm production is for Exynos. See comment from SEC's 2nd quarter earnings call.

"Regarding our 14-nano time schedule, to recap that, our current 14-nano schedule is that we will launch mass production around theend of this year, and going through a ramp-up during 2015, first with internal and then in the second half using external customers for the foundry
operation."
 
The 14nm production is for Exynos. See comment from SEC's 2nd quarter earnings call.

"Regarding our 14-nano time schedule, to recap that, our current 14-nano schedule is that we will launch mass production around theend of this year, and going through a ramp-up during 2015, first with internal and then in the second half using external customers for the foundry
operation."

Thank you.
Well, let's wait for this first 14nm foundry silicon then.
 
Hi user - No, I'm no way associated with Samsung PR. I'm actually a programmer trying to learn more about computer hardware. I thought Semiwiki would be an interesting place to read about it and that's why I joined. Regarding the big boldface...I did a quick copy & paste from the Korean page link...hence its a little bigger than normal. Apologies for the misunderstanding...as I was just being lazy there.

PS - I personally believe Samsung (as a company) lacks morals & routinely screws everyone (society, employees, business partners, customers, competitors) for money. Frankly, I'd much rather see TSMC and Intel kick its butt.
 
Beautiful examples If you know a thread in particular that you could send me in terms of the production numbers would be greatly appreciated
 
Back
Top