Seeking Alpha
Profile| Send Message| ()  

Six reasons why the iPhone will soon have processors built by Intel:

  1. Apple (AAPL) needs to have the best technology to continue to provide cutting edge products and to be able to charge the premium it has charged in the past.

  2. Intel (INTC) needs to put its processors in smartphones. The smartphone market is growing rapidly.

  3. Intel will make more money by building the iPhone processor than not.

  4. Intel will soon have excess capacity in its fabs (semiconductor chip factories).

  5. Apple will pay less per processor buying from Intel, compared to its current supplier Samsung (GM:SSNLF), or its other alternative TSMC (TSM).

  6. Apple and Intel would both find strategic advantages in such a deal.

Despite the widespread belief that Intel is a microprocessor company, its distinguishing characteristic is that it is a leader in the manufacturing of integrated circuits. Intel's microprocessor design division is really just a way to ensure its fabs are full of high value semiconductors.

I have listed six reasons why this will eventually occur, but I have not specified the processor. Two scenarios exist:

  • Scenario 1. Apple could put an Intel designed Atom processor in the iPhone, or
  • Scenario 2. Intel could act as a foundry for the A series, and produce those Apple designed A series (A7 or A8) for the iPhones.

The six reasons above apply to both scenarios. More details on the six reasons are listed below.

Apple needs to have the best technology to continue to provide cutting edge products and charging the premium it has charged in the past. Apple invented the high end of the smartphone market. Its average selling price of $640 (along with a $400+ profit on each phone) is providing piles of cash for its balance sheet. However, Samsung is growing rapidly in the smartphone market and encroaching on Apple's dominance in this area. To complicate things, Apple is currently relying on Samsung to produce several key parts in the phone, including the central processing unit.

Intel needs to put its processors in smartphones. Smartphone sales are currently growing at a rate of 40% annually, while the growth of more traditional PC CPUs is much lower. Of course, Intel has its own product line of processors that it wants vendors to build phones around. But under either scenario discussed here, Intel would suddenly be manufacturing 150 million CPUs/year for smartphones. Under Scenario 1 these CPUs would be Intel designed, while in Scenario 2, these CPUs would be Apple designed.

Intel will make more money by building the iPhone processor than not.

If Intel fabricates the A-series chips for Apple, either this new work is filling empty capacity, or it is displacing Intel Atom chips the company would be putting into smartphones. If this is filling empty capacity, the decision is a no brainer. If Intel is manufacturing near capacity, then it just needs to charge Apple an amount that exceeds what it would have made profit on the Intel Atom chips.

Intel will have excess capacity in its fabs. Last year, Intel was running its fabs at 50% capacity, although more recently it has been back to almost full capacity. However, next year Intel will open Fab 42 in Chandler, Arizona. Fab 42 runs on 450 mm wafers. All other modern plants run with 300 mm wafers. Also, Fab 42 will be the first 14 nm node. When Fab 42 begins production, probably in 2014, Intel will suddenly have the capacity to make twice as many transistors. See the chart below:

Fab 42 will essentially double Intel's capacity.

Apple will pay less per processor. A foundry sells wafers, not devices. If Apple goes to TSMC and buys 300 mm wafers running on a 28 nm process, it pays about $5000 per wafer. The A6 die has an area of about 100 mm2, which means that Apple will get about 500 physical dies per wafer. TSMC has a yield that I will estimate at 50%, meaning that only 250 of those 500 will be functional devices. So for a price of $5000, Apple gets 250 devices, for a cost of about $20 per device.

At the Intel foundry, Apple will buy 450 mm wafers on a 14 nm process. Because the process is now half the size, the same number of transistors will fit in one fourth the area, multiplying by the increase in area of the wafer results in a huge number, 4500 physical devices, on a wafer. Intel has historically had the highest yield in the industry. It is rumored to be in the high 80% range. Let's call it 80%. Apple will have 3600 functional devices per wafer. Compare that to the 250 devices per wafer it would get at TSMC. If Apple still pays $20 per processor, Intel would be collecting $72,000 per wafer. Costs for a 450 mm line are going to be higher than a 300 mm line, maybe twice as high. The exact number depends on assumptions. TSMC makes a gross margin of about 40%, with these estimates Intel would be making a 93% gross margin on the Apple business, while Apple's costs would remain the same. Probably, Intel will charge Apple a premium for access to this leading edge technology.

Apple and Intel would both find strategic advantages in such a deal. Doug Freedman of RBC Capital wrote in a report last fall that initial talks occurred between Apple and Intel about the iPhone processor. The basic framework was that Intel would produce Apple's A series (ARM processors) for the iPhone, while Apple would switch the iPad to Intel's x86 architecture. The primary concern for Intel is that the iPhone competes with Intel's other customers. However, Intel's customers already compete against one another. As long as Intel makes more money from an iPhone than an Intel based Android or Windows phone, Intel should be selling Apple whatever it wants.

For Apple, this deal would satisfy a number of strategic goals. First, it would reduce its dependence on Samsung, its chief competitor in the high end smartphone market. Second, it would get access to leading edge manufacturing technology, potentially a year ahead of what competitors have access too. The 3 month lead time that Apple's A6 processor had on the competition from Samsung, was a multi-billion dollar advantage. Third, it would put the new x86 iPad on the market, which should be binary compatible with its MacOS laptop computers. Fourth, it would eliminate capacity constraints that have limited Qualcomm's sales last year.

Conclusion

Intel should soon be producing processors for Apple's iPhone. It might be an Intel architecture, or it might be an Apple architecture, but either way Intel will be making a nice profit.

Source: 6 Reasons Why The iPhone Will Soon Have Processors Built By Intel