The all-important Apple (AAPL) "A" chips that power the iPod, iPhone and iPad are all manufactured by Samsung (OTC:SSNLF) since the first iPhone arrived on the scene in 2007. During the design of the iPhone, per the Steve Jobs biography, Intel was considered for the iPhone application processor. Jobs, having had a successful relationship and transfer from the power PC chip to the Intel (INTC) X86 family for the Mac family of computers, was at least amenable to using Intel has the source for this critical component. The key leader of the application processor design team (name forgotten) put his job on the line in support of an ARM (ARMH) based design for the AP. Hence, Apple used a Samsung ARM based chip for the first generation of the iPhone and has ever since designed their own "A" chip in house and used Samsung as a foundry.
In this timeframe Samsung was building feature phones and was thought by Apple to be no threat as competitor to the iPhone. As we all know that changed very quickly as the success and profit on the iPhone became apparent. Before we knew it, Apple and Samsung were in the unusual position of being customer/supplier and tooth and nail competitors in the smartphone business. The unpleasantness got to the point of engaging in a dozen lawsuits in multiple worldwide jurisdictions.
At the time, Samsung supplied the majority of the dollar value of the components used in the iPhone, including the display, mobile DRAM and NAND, the "A" chip, battery and probably quite few of the minor parts like capacitors and resistors. At one time I calculated that the Samsung dollar content in the iPhone was between 60-70%.
Of course with all the acrimony surrounding the relationship, Apple wanted desperately to get out of the Samsung happy time supply arrangement. They found new suppliers for the display, battery, and NAND memory. Since there was a shortage of mobile DRAM and Samsung and Hynix were the only two non-bankrupt suppliers of mobile DRAM, Apple had to stay with Samsung on the critical mobile DRAM product. As long as Elpida was in bankruptcy, Apple had to swallow their pride and buy mobile DRAM from Samsung.
Basically there are two things that have kept Apple from moving the "A" chip out of Samsung. One was the sourcing issue on mobile DRAM and the other was the sourcing issue on LTE chips. The problem on the mobile DRAM will finally be solved on July 31st with the closing of the Micron (MU) acquisition of Elpida (with serious help from Intel).
The LTE chip problem is a little more subtle. Apple does not have an integrated LTE section on their "A" chip, they use a discrete chip supplied by Qualcomm (QCOM). The Qualcomm part is manufactured, to my understanding, by -guess who- Samsung. From here it gets a little hard to follow, but if Apple moved the "A" chip business to Intel (the logical supplier), both Samsung and Qualcomm would lose business because Intel will eventually supply a 4G LTE discrete chip with the "A" chip. So the danger is that if Apple moved the "A" chip business to Intel, Samsung and/or Qualcomm could conceivably make LTE chips really hard to come by. This is similar to the Korean lockdown on mobile DRAM. Apple got a little taste of this when Apple moved their mobile DRAM and NAND business to Elpida and Toshiba and Samsung came back with a 20% price increase on the "A" chips.
The bottom line is that to move the "A" chip business to Intel, two things had to happen; Elpida had to be saved in order to supply mobile DRAM chips to Apple and Intel's Infineon division had to come up with a discrete 4G LTE solution in order to de-fang Qualcomm/Samsung. The first of those is done, but Infineon is nearly a year behind schedule on the 4G LTE solution. It now appears that Intel/Infineon has a certified LTE solution that Apple can use.
So, now all of the potential blackmail roadblocks are out of the way of Apple's intense wish to end all supply business with Samsung.
Now what do we get? An oft-repeated rumor from the notorious Taiwan rumor rag Digitimes, that Apple has signed a three year deal with TSMC (TSM) for 20nm, 16nm, and 10nm versions of future "A" chips. Does that make sense? TSMC, who would have all the reasons in the world to shout about even a 20nm (planar) test chip, has shown exactly nothing, while Intel is shipping 100s of millions of 22nm Trigate devices and in a matter of months will be shipping production chips based on 14nm, again Trigate.
I really don't have anything against TSMC; as a matter of fact I admire the company for all they have done. The fact remains that TSMC is nearly two nodes behind Intel, and in the case of Apple, a conscious decision to put all their beans in the TSMC basket is suicidal. On the other hand, if I were Apple and the largest and best foundry on the planet promised me 20, 16, and 10nm fab services, I would give them orders for a piece of anything they asked for. In the unlikely event that TSMC were to catch up with Intel, Apple would have two sources of leading edge foundry work, TSMC and Intel.
My bet would be that Intel is all set to run the "A" chips as well as the LTE chips for Apple and has been ready for some time while the two roadblocks, mobile DRAM and 4G LTE chip supply, are being resolved.
How could something like this be kept secret? I honestly don't know, but I remind the readers that the Apple move from Power PC to Intel X86 was kept absolutely secret for the six years that it took to re-write OSX for X86.
If the TSMC rumors were true, the stock price should be up $10. It's not.
Intel helped Micron pull Apple's bacon out of the fire by saving and buying Elpida. Micron's payoff for this move, which seemed like insanity at the time, is acquiring Elpida at 20 cents on the dollar, packed to the roof with Apple mobile DRAM orders. Where is Intel's payoff for its help in getting Apple off the Samsung hook?
It will come in the form of application processors (either "A" chips or x86 chips) and LTE chips. Having Apple in the bag goes some way to explaining the huge apparent overcapacity at Intel.
That's my story and I'm sticking to it.