Apple and Foundries – Lose Lose.

RFM AvatarSmall

 

 

 

 

 

Switching foundries is not necessarily great news for any of the players. 

  • Apple and Intel are certainly talking and while a foundry deal is not unlikely, I don’t see Apple switching to x86 in the iPhone any time soon.
  • Intel has long been and is likely to remain one to two generations ahead of everyone else in terms of process node.
  • Combine this with its lead in 3D transistors and one can start to see how its x86 chips may come within spitting distance of ARM when it comes to power consumption.
  • However, the real excitement would be to make an ARM processor on Intel’s best in class fabs.
  • If Apple were to switch to Intel for its in house applications processor, one can see how it would be a leap ahead of everyone else taking advantage of both ARM’s efficiency and Intel’s world beating manufacturing.
  • This sounds great but winkling business away from Samsung LSI will be much harder than expected as I suspect Intel and TSMC are finding.
  • This is because Samsung accepts much lower margins on its business from Apple than Intel, TSMC and their investors are used to from other customers.
  • There is no doubt that Apple wants to diversify away from Samsung but how much extra is it willing to pay to do so?
  • Intel, TSMC and others are very keen to get Apple on board but if one customer gets a discount then they will all want one and one can see just how margins might collapse.  
  • This will be a very delicate balancing act where I suspect some ground will be given on both sides.
  • For Apple investors, it’s a negative development as I am a strong believer in Samsung’s ability to keep both in-house and outside customers at arms length.
  • Hence I see no real benefit other than piece of mind for Apple, whereas investors in the shares will have to put up with pressure on gross margins.
  • When Apple and Intel are mentioned in the same breath, the question is always asked whether Apple will switch from ARM based chips for its iOS devices?
  • At the moment, this is extremely unlikely as apples for apples, Intel is still miles behind on power consumption.
  • It can really only make up the gap with its superb manufacturing, thinner line widths and transistor technology.
  • Hence, I think movement on the foundry relationship is far more likely than a processor switch.
  • However, I suspect that a foundry switch will cause negative sentiment for both parties, once the euphoria has worn off. 

RICHARD WINDSOR

Richard is founder, owner of research company, Radio Free Mobile. He has 16 years of experience working in sell side equity research. During his 11 year tenure at Nomura Securities, he focused on the equity coverage of the Global Technology sector.

Blog Comments

While I agree with you that an Apple-Intel foundry relationship makes the most sense, I think that you could be underestimating Intel’s margins for ARM chips on 22nm. Depreciation has been largely paid for with x86 production (with the possible exception of chip packaging), Intel’s foundries have excess capacity, production problems of moving Apple’s ARM chips to 22nm should largely have been solved from the x86 production on the same lines, Apple’s chip orders are (unlike any competitors) for large quantities of the same design, Apple has in-house designers who can work around any problems with Intel.

More thoughts on this and why Samsung is the loser in a Jan article ‘ Intel making Apple’s ARM chips will leave Samsung behind’
seekingalpha.com/article/1090941

Thats a possibility but a node shrink is usually used to lower prices to keep the march of technology going. I suspect that Intel would use lower costs and power consumption to keep up with ARM meaning that margins for ARM products could me lower than on x86. The bigger problem is that Samsung takes MUCH lower prices on its Apple foundry business than TSMC and others do from other customers. Hence to win Apple from Samsung there needs to be meaningful movement on price to get the business done. .

I am curious about the source of your claim regarding the “MUCH lower” price Samsung is willing to charge compared to others. To make a name for itself in the high end foundry business where TSMC is king, Samsung may have wanted to attract a high profile customer known for its demanding nature, as a third party validation of its fab and service quality, leading to some discount, but “MUCH lower”?

Regarding your “if one customer gets a discount then they will all want one” argument: Every customer gets a considerably different price already from TSMC. I am sure volume makes some difference, but the credibility of the threat to wholly or partially move to another foundry makes a big difference, too. This is nothing new. If TSMC thinks you are unlikely to undertake the effort to port your design to a rival fab, it will give you one price. If it wants you to port your design from a rival to its fab, you get another price. Of course, if there is a capacity crunch, prices jump. If there is overcapacity, prices drop. Giving some discount to a big customer is about as problematic as an airline offering deeply discounted tickets on its website every now and then. Besides, the price will be a secret anyways.

Its in the margins. TSMC has much higher gross margins than Samsung LSI. Some of this is down to scale but a ot down to price. Yes the pricing is not transparrent but when you see one lot with much fatter margins than another you start to ask questions. The analysis I have done points to higher prices being carged at TSMC die to its market leading position and reliability.

Much, if not all, of Samsung’s lower pricing will come from Apple’s foundry related CapEx and the large volumes of a single chip design Apple can commit to. As Intel knows, it’s easier to get high yields from long runs.

Totally agree…but Samsung LSI also has pressure to fill its fabs as a lot of business may disappear…furthermore it had to get Apple in the door in the first place. The analysis suggest strngly that there is meaningfu difference between margins at TSMC and Samsung LSI, a large part f which is down to price. Its not transparrent and so people look at profiability as a steer.

Richard, as someone who once worked in the chip industry, I think Apple basically has no choice but to drop Samsung.
Foundry work normally requires approximately four years of collaboration between the two parties, which means a designer like Apple must share their chip roadmap for the next four years (give or take) with the foundry. Not coincidently, Samsung’s recent designs for their own chips is at least heavily inspired-by the designs Apple shared with them. Given that Apple is ahead of Samsung on custom CPU design technology among other things, this is very problematic for Apple.
Richard, I don’t know why you are “a strong believer in Samsung’s ability to keep both in-house and outside customers at arms length”, because AFAIK, not a single major chip design house shares the same belief. In the foundry business, volume is king, if Samsung accepts lower margins than TSMC on their Apple business, they must be willing to give a similar deal to Qualcomm when Apple drops them, right? Yet I don’t think Qualcomm will let Samsung make any of their advanced chips, if at all. Of course it’s still early, but Apple drops Samsung and goes with TSMC has been an open secret in the industry for a long time, yet we haven’t heard any credible information on a potential Qualcomm-Samsung deal. Some may say Samsung will be busy making their own chips, but Samsung is always looking to expand their foundry business, they will build another foundry if the demand is there.

Hi James…
I hold this view because if the way that the semi business treats in house and external customers. It is the biggest seller of flash. If any of its external customers thought they were being disadvantaged compares to the internal customer, they would run a mile. There are plenty of other choices, adnd yet they keep coming back. The same has to be true otherwise there would be no external customers. Samsung does not offer anything (other than some screens) that you cant get anywhere else. It just makes those items more cheaply. If all its external customers left, it would be a disaster…it cant ket that happen hence I believe that it treats customers fairly whether or not competitors choose to believe it.

actually Qualcomm have admitted in public that they have been siffing around Samsung LSI