You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!

Apple A10, Samsung & TSMC

Robert Maire

Moderator
Does TSMC win of Apple A10 hurt Samsung chip biz? What does it say about Samsung's Capex plans? PLAB pre-announce is worse for semis than itself...

Evidence keeps mounting that TSMC has captured all of Apple's A10 business for the fall. What does this do to the only part of Samsung (Semis) that is doing well? How bad will the trickle down of reduced CAPEX spend be on the semicap companies as a result? Article in Korean Electronic Times News appears to support Samsungs loss of A10.

We have been talking about the Apple A10 processor ever since the A9 was released and turned out to be split production between Samsung and TSMC. We have been saying that the rumor mill has TSMC capturing 100% of the A10 processor business for Apple's Iphone 7 this fall which would imply a production of processors starting in June. In past controversies, Taiwanese trade rags supported TSMC while Korean trade rags supported Samsung, as would be expected. But yesterday a Korean trade rag, Electronic Times, reported that TSMC has won 100% of the A10 production which seems to be pretty good confirmation of what we already know.

Apple Entrusts TSMC with All of APs for iPhone7

Is it good or bad for Samsung? Perhaps its a blessing in disguise?
It certainly can't be good news for Samsungs only strong business unit, the semiconductor division, to lose its biggest customer, Apple and the all important for bragging rights for the A10 processor. However, given that Apple beats the heck out of its suppliers perhaps it could be a blessing in disguise IF Samsung can replace Apple's business with other business such as Qualcomm (which appears to be the case).

Is it really a zero sum game?
There is only so much 14nm foundry capacity in the world and its split between TSMC and Samsung (we won't count Intel as their foundry efforts have been less than stellar...). So one would imagine that if it isn't at TSMC it will show up at Samsung and vice versa. However that assumes that demand for all products remain the same in the face of the global economic weakness in front of us. Depending upon your point of view you could argue that Apple business will hold up better than other smartphone businesses with less cachet. Qualcomm has been under pressure and perhaps more so of late so depending on their business to fill your fab at good prices may be more "iffy", than producing Apple's A10 which will likely sell gazillions when the Iphone 7 launches

How does this trickle down to capex spending plans and semicap companies fortunes?
The capacity for the A10 processor already exists because it is being built on the same 16/14nm technology and fabrication platform already used for the A9, so TSMC or Samsung don't need to add a lot of tools if any for the Iphone 7 ramp.

However 10nm and the Apple A11 , which also appears to be going TSMC's way, may be the break point where TSMC outspends Samsung because TSMC will need more capacity than Samsung at 10nm . Both companies spent heavily on 16/14nm as both got Apple's business.

We think this implies that while both TSMC and Samsung will be racing for 10nm, that when it comes time to do "capacity" buys after the process and pilot lines have been "nailed down" that TSMC will buy in higher capacity for 10nm as it ramps to satisfy Apple's A11 needs in June of 2017 than Samsung which will have a less certain demand outlook for 10nm and may see its semiconductor business unit under pressure due to reduced revenues from the lack of Apples A9. This likely trickles down to order patterns for the latter part of 2016 which is when the companies have to do capacity buys for 10nm as the process should under control by then, and they have to order equipment towards the end of 2016 to ramp for a June start of A11 production (only more accurately only TSMC needs to....)

This then would suggest that hoped for/wished for recovery in the second half of 2016 which everyone seems to be counting on may not be all that spectacular if only TSMC is ordering in big volume as compared to the 14nm ramp when both ordered in volume.

All in all this does not bode well for Samsung spending in 2016. Having lost their "anchor" customer that paid back their capex investment in the past makes spending big just that much more risker and a poor bet when the rest of the company is suffering.

Impact on the stocks
We would be careful about Samsung specific exposure. Lam which has Samsung as one of its biggest customers is very lucky indeed to be have KLAC in the bag with monster orders just as Samsung continues to be weak with an uncertain outlook. Without the KLAC upside to look forward to, Lam may not have otherwise been as attractive. As I have stated before, Lam's timing couldn't have been better.

Small companies, such as Mattson, with high Samsung exposure, have already reported that business is falling off a cliff in Q1, but they are lucky?? enough to be falling into the arms (or jaws...) of a Chinese dragon (investor...) to rescue them.

Applied traditionally has had better exposure to TSMC and will likely benefit if capex spending shifts away from Samsung to TSMC. TSMC is the foundry that AMAT built.... Not much impact on ASML as it appears EUV isn't going to be used at 10nm anyway....

PLAB pre announcement casts a long shadow on the industry
Photronics, which manufactures photomasks for the industry, pre-announced a much weaker than expected start to the new year with business expected to be below what is already a seasonally slow period.

As you will recall, photomask sales are a good leading indicator of where the semiconductor industry is headed as the first thing you buy before production can start is a set of photomasks. This weakness is basically saying that chip production will be down more than usual in the near term, confirming what we already know about macro economic conditions but putting into black and white the view that chip companies will likely be reporting poor Q1 performance.

This is certainly not good news for the entire semiconductor industry and of course could trickle down to weaker capex spend as chip makers tend to put the brakes on capital spending very, very quickly these days. It probably has less of an impact on technology spending on 10nm but may slow spending on expanding 28nm capacity which has been helping semicap companies all along (28nm is the gift that keeps on giving for the industry....)

While certainly not great news for Photronics, it is probably worse of an indicator for the industry as a whole than Photronics specifically which well still view as a good company with excellent management that happens to be stuck with a crappy demand environment in the near term.

Our "macro" concerns have certainly been correct
Since the beginning of the year, in virtually every newsletter we have written and we have commented that the semiconductor and semiconductor equipment industry specific news and earnings are overshadowed by the negative macro economic environment which "swamps" any other news be it positive or negative.
All the stocks have been whacked by this as technology has taken it on the chin with semiconductors being no exception to the rule.

We continue with our view that the macro news flow will rule the day with little reason to look positively on the semi sector in the near term as it remains sensitive to consumer spending trends and confidence.....it may take a while to get out of this funk...
 

hist78

Member
The relationship between TSMC and Applied Materials is deep and wide. Take a look at one biography among TSMC eight board of directors':

Michael R. Splinter

Independent Director
Joined TSMC Board: June 2015
Member, Audit Committee
Member, Compensation Committee

Mr. Splinter served as Chief Executive Officer of Applied Materials from 2003 to 2012 and as Chairman of the Board of Directors since 2009 and retired in June 2015. Prior to that, he served at Intel Corporation as Executive Vice President of Sales and Marketing from 2001 to 2003, and Executive Vice President of Technology and Manufacturing group from 1996 to 2001. Mr. Splinter currently serves as Director of The NASDAQ OMX Group, Inc. and Pica8, Inc. He is also a General Partner of WISC Partners LP. Mr. Splinter obtained his Bachelor and Master Degree in Electrical Engineering from the University of Wisconsin Madison, and he was also bestowed an honorary Ph.D. in Engineering from the University of Wisconsin on May 2015.
Source: Taiwan Semiconductor Manufacturing Company Limited
 
Last edited:

benb

Member
Good analysis Robert. We seem to have limited upside in good years (like the last few years?) and deep, deep downside in bad years (like 2016?). We have consolidation to thank for this. I have a few thoughts to add:
1-What happens to Intel's New Mexico operation? That's a lot of capacity that could be converted to something, perhaps by one of the foundries?
2-While the Samsung/TSMC horserace goes back and forth every 12 months, Intel's ongoing collapse from technology leader to ??? (fab-lite?) is the real story in my opinion.
3-What is the minimum amount of fab capex worldwide required to sustain a viable semiconductor equipment industry? And what will happen if we flirt with a capex level that can't sustain Applied and ASML? I don't mean downturns, which the industry has always had, I mean full-business-cycle capex being insufficient.
 
Last edited:
Top