Complex chip shift from Samsung expected to take Apple 12-18 months

Posted:
in Future Apple Hardware edited January 2014
For Apple to move production of its iPhone and iPhone and iPad chips away from Samsung would be a long, difficult process with only three realistic alternative companies.

As speculation grows that Apple plans to take its chip business away from Samsung, analyst Amit Daryanani decided to take a closer look at Apple's options. He views it as a long-term project that may not occur until 2014.

"Shifting chip manufacturers isn't easy and requires a complete redo of production and manufacturing process," Daryanani wrote in a note to investors on Friday. "Hence, these changes will take 12-18 months at minimum and won't be commercially sold 'til 2014."

In his eyes, there are two most likely companies for Apple to turn to: Intel and Taiwan Semiconductor Manufacturing Company. TSMC has long been rumored to be a likely partner for Apple if it moves chip production away from Samsung, while Intel has indicated it would like to build custom processors for Apple's iPhone and iPad.

A6X


The case for Intel is that it has three foundry partners and is currently ahead of ARM in development by 1.5 generations. But Apple's iOS platform is already heavily invested in ARM, which would make a transition to Intel's chips more difficult.

That's where TSMC comes in, which Daryanani said could begin manufacturing chips for Apple in 2014, as soon as they transition to their 20-nanometer production process. He believes TSMC would need capital expenditures of between $1 billion and $3 billion, which would give Apple an opportunity to co-invest and ease the company's burden.

Another potential ? but less likely ? partner for Apple would be Global Foundries, a company that is already working with ARM to build 20-nanometer chips based on ARM's reference designs. Daryanani said it's possible that Apple could co-invest to ramp up dedicated fabrication plants with them for production, though TSMC and Intel are seen as more likely options.

Finally, Daryanani said Apple could "go vertical," and opt to build their own chip fabrication plants. He views this as an unlikely option, as Apple mostly keeps design elements in-house and hands off manufacturing to outside partners.

One report this week claimed that technology industry insiders in the Far East believe TSMC could begin building custom chips for Apple as soon as late 2013. Such a transition away from Samsung is expected to have shockwaves through the industry.

Apple's transition has become expected as the company faces a growing rift with rival Samsung, which currently builds all of the custom processors found in the iPhone, iPad, iPod touch and Apple TV. While the two companies partner on supplies, Apple and Samsung are fierce rivals in the smartphone, tablet and computer markets, among others.
«13456

Comments

  • Reply 1 of 101
    MacProMacPro Posts: 18,372member
    I hope the truth is Tim is much further along the planning path than those guessing realize.
  • Reply 2 of 101
    So, IOS might have a chip redo, hopefully Apple does the chip, If another company does it,IOS on the chip might cause errors.
  • Reply 3 of 101
    Not all that surprising given the current Samsung/Apple relationship.
  • Reply 4 of 101
    sr2012sr2012 Posts: 896member
    Now we know why Steve never wanted to do this vindictive move even though Samsung was gnawing away the hand that fed them... Apple still needed the chicken!

    Let's see the options...

    TSMC, not the smoothest experience whenever fabbing with them. Ask Nvidia over the past several years.

    Intel... And Intel is suddenly going to make ARM how? Is it even an option?

    Not looking good.

    More production problems await.

    And you thought the past six months were "off"...
  • Reply 5 of 101
    jnjnjnjnjnjn Posts: 588member
    Amit is as credible as his suggestion about Intel taking over chip production.
    As credible as a factor 10 price hike and a factor 5 in power dissipation.

    J.
  • Reply 6 of 101
    sr2012sr2012 Posts: 896member
    jnjnjn wrote: »
    Amit is as credible as his suggestion about Intel taking over chip production.
    As credible as a factor 10 price hike and a factor 5 in power dissipation.
    J.

    Bingo. Intel. LOL. Nonsensical, isn't it.
  • Reply 7 of 101
    lkrupplkrupp Posts: 7,318member

    Quote:

    Originally Posted by sr2012 View Post



    Now we know why Steve never wanted to do this vindictive move even though Samsung was gnawing away the hand that fed them... Apple still needed the chicken!

    Let's see the options...

    TSMC, not the smoothest experience whenever fabbing with them. Ask Nvidia over the past several years.

    Intel... And Intel is suddenly going to make ARM how? Is it even an option?

    Not looking good.

    More production problems await.

    And you thought the past six months were "off"...


     


    Apple is doomed. Yes, we know that already. Tell us something we don't know.

  • Reply 8 of 101
    haarhaar Posts: 563member
    blah,blah,blah... the 20nm ipad chips will be in the new ipad in 2013 (october) or 2014 (march)...

    i wonder what samsung going to do... they have to come up with a non-copied phone/tablet for the next 18 months (due to the 2014 patent case)... or perhaps samsung will just "Rest on their laurels" LOL...
  • Reply 9 of 101


    Originally Posted by haar View Post

    blah,blah,blah... the 20nm ipad chips will be in the new ipad in 2013 (october) or 2014 (march)...

    i wonder what samsung going to do... they have to come up with a non-copied phone/tablet for the next 18 months (due to the 2014 patent case)... or perhaps samsung will just "Rest on their laurels" LOL...


     


    All they have to do is make "N" models of their existing crap that also look exactly like Apple designs (just different designs than the ones in the case). 

  • Reply 10 of 101
    Going vertical can't be the option, as it's not in Apple's philosophy.
    They have always followed a "make AND buy" paradigm, which means that they develop the core competencies internally and acquire secondary competencies outside. That is: development of products is done inside the company, while manufacturing is done outside by partners (the "specialization socialization" logic).

    Anyway, Intel may produced chips for Apple using the ARM architecture too! Intel is a licensee of ARM, and used to make ARM chips in the past...
  • Reply 11 of 101
    tbelltbell Posts: 3,146member


    It is a shame Apple can't get Intel to do foundry work, which would be profitable for Intel and beneficially for Apple. No way Apple is dropping ARM. It has too much invested, and it gives Apple greater control over its product.


     


    Nonetheless, I am guessing Apple will keep Samsung on board for current chips, but will try to switch for the next major release of the A series processors. 

  • Reply 12 of 101

    Quote:

    Originally Posted by sr2012 View Post



    Now we know why Steve never wanted to do this vindictive move even though Samsung was gnawing away the hand that fed them... Apple still needed the chicken!

    Let's see the options...

    TSMC, not the smoothest experience whenever fabbing with them. Ask Nvidia over the past several years.

    Intel... And Intel is suddenly going to make ARM how? Is it even an option?

    Not looking good.

    More production problems await.

    And you thought the past six months were "off"...


     


    Intel could easily make ARM.  All they need are the process steps (recipes) and with a bit of development could be banging out ARM chips inside 6 months.  I think it's unlikely though.  Intel's manufacturing processes are really first class - industry leading, but they are not really suited to being a foundry, where cost is a key driver.


     


    TSMC are a quality company.  They have had some glitches with NVidia, but that could (and probably is) be a problem with NVidia and their design as much as it is TSMCs manufacturing capability.


     


    Global Foundries would be interesting.  They've build a hell of a big fab in New York state, and will need something to run through it, although they come with the baggage of what used to be Chartered Semi, and they are a joke.  If Apple work with GF, they would want to insist on their devices not being made in the Singapore fab.


     


    As for the option of Apple doing it all themselves, I just don't ever see that happening.  It would cost them far too much money to acquire the expertise to even make a start on building a fab and in reality, the foundry market is so competitive that it's unlikely Apple would ever be able to make it cost effective.

  • Reply 13 of 101
    gazoobeegazoobee Posts: 3,754member


    OMG what a load of crap.  Why would anyone choose to include the ridiculous speculation about Apple moving to intel chips in an article that was ostensibly about moving to a different foundry or fab???  Is anyone even thinking at all when they write this tripe?  

  • Reply 14 of 101
    melgrossmelgross Posts: 31,902member
    This isn't as difficult as some believe. Look at the way Apple moved to 32nm. They gave Samsung chips to make for the older iPad 2 first, to work out any bugs. That went well, so they then moved over the rest of their production for the next generation. So that can give TSMC time to get production to where it is needed. Apple just gives them a chip for one model first, say, an older iPhone and iPad, which could be the A6 at that time. Then, they could transition over to the newer model chips when ready.

    This isn't an all or nothing move. Samsung could continue producing until TSMC is in full production.

    Unlike what some people have been saying, Samsung isn't going to throw Apple under the train. They would get in too much trouble for that.

    As far as Intel goes, yes, they'd love to produce a custom x86 chip for Apple, who wouldn't? But if Apple came to them and said that they would want Intel to produce 300 million custom ARM chips for them, Intel might be willing to do that as well. If the alternative is nothing, that could look like. Very good deal.

    Global Foundries is another case entirely. I don't believe they could produce enough chips for Apple's needs. Though, they will be building a new foundry in upstate NY. It's not a company I would consider.
  • Reply 15 of 101
    Intel would be my choice.

    Intel is much more aggressive in building up performance. It's new x86 chips for smartphones are close to ARM's low energy needs yet are more powerful.

    Intel is a licensee of ARM. It has built ARM chips before.

    Currently, its x86 chips have gained ZERO interest in the smartphone and tablet computing market.

    Thus Intel's choice is either to make nothing and keep losing or to start making ARM chips also in partnering with Apple to gain marketshare back. The profits may be slimmer on ARM but they are present.

    Intel certainly wouldn't stab Apple in the back like Samsung does.

    Apple can certainly invest with Intel in building new chip foundaries dedicated to ARM production - just as Samsung invested in building a foundary in Texas. This way, Intel doesn't have to lose production on its current x86 chip foundaries.

    By producing ARM chips with Apple and continuing to sell x86 chips to Apple, an Apple-Intel partnership would be lucrative to Intel.

    If anything, Apple has pushed Intel to be the best it can be and forces it to continue to be competitive.

    The only fly in the ointment is that Intel has been a slug in developing new versions of its XEON processors. This is why the new Mac Pro has been delayed delayed delayed.
  • Reply 16 of 101
    nhtnht Posts: 4,496member


    I dunno why this analyst is so clueless.  UMC also has 28nm, 20nm and 14nm fab plans AND needs capital to the point they are willing to sell 10% interest in exchange for guaranteed capacity.  UMV licensed IBM's 14nm FinFET just like GlobalFoundries and Samsung.


     


    The kicker is whether Qualcomm has already locked up all the near term UMC capacity.  


     


    No shit 2014 for 20nm.  However, it is possible that these fabs can start producing for Apple on 28nm in 2013 given no one besides Apple and the fabs know how far along they are in the transition process.

  • Reply 17 of 101

    Quote:

    Originally Posted by EgoAleSum View Post



    Going vertical can't be the option, as it's not in Apple's philosophy.

    They have always followed a "make AND buy" paradigm, which means that they develop the core competencies internally and acquire secondary competencies outside. That is: development of products is done inside the company, while manufacturing is done outside by partners (the "specialization socialization" logic).

    Anyway, Intel may produced chips for Apple using the ARM architecture too! Intel is a licensee of ARM, and used to make ARM chips in the past...




    Apple may have no other choice but to start manufacturing its own chips.


     


    If they move away from Samsung and then are f*cked over by their new partner... what then?!


     


    I believe Apple would have a long hard fall from which it might never fully recover.


     


    The devil you know.

  • Reply 18 of 101
    It might take 12 to 18 months... and then again, they might have started 12 months ago!
  • Reply 19 of 101
    melgrossmelgross Posts: 31,902member

    Apple may have no other choice but to start manufacturing its own chips.

    If they move away from Samsung and then are f*cked over by their new partner... what then?!

    I believe Apple would have a long hard fall from which it might never fully recover.

    The devil you know.

    It takes time to come up to speed in process technology. Apple has no recent experience in this. Intel has the best fabs. Everyone else has many more problems. If Apple began their own, it could take two years before the fab was complete. But they would have to hire a large number of people expert in this away from others. What is the guarantee they could get a talented and cohesive team together in that time? Then the fab would need to undergo testing. It could be more than two years down the road, possibly three.

    But they would have to upgrade their fab with every process generation. It's estimated from Intel's numbers, that a new 22nm fab costs about $8 billion. Is this really the best way for Apple to be spending its money? Could they break even on this venture? I don't see it.
  • Reply 20 of 101
    I think 24 months is optimistic... Apple demands extremely high quality on a very low margin. It take a lot of time and scale to achieve a result that can meet Apple's demands and still make any kind of profit for the component supplier.

    Case in point the iPad Mini. Apple shifted away from Samsung for the displays and now they are having supply issues due to low yields. This caused a revision in the sales forecast from 10 million to 6 million. So these types of moves are not without risk...

    Many companies drool over a big OEM contract. However, once they get it, they find out that it is hard to even break even on the contracts. Samsung was able to do this for Apple because of their huge and diverse manufacturing base.

    It will be interesting to see how this all works out....

    P.S. Just got a Verizon LTE iPad Mini....Easily the best damn iPad yet!! As for the people whining about the display, get over it...It looks fine...

Sign In or Register to comment.