Update: Sept 20th: It has been confirmed that the A7 is still made by Samsung, most likely on their 28nm High-k dielectric process (same as the Exynos in the Galaxy S5). The M7 has also been confirmed to be an off the shelf NXP LPC1800 ARM Cortex-M3, running at up to 180MHz, nothing spectacular, and fairly common for sensor interface. What does this mean for Apple? It means if they can get that much performance out of Samsung’s 28nm process, when and if they Do switch to Intel, the possibilities are quite interesting.
However, its still interesting to play what if, so the below analysis remains.
It has been rumored that Apple’s new A7 processor may be fab’d by Intel, rather then TSMC and Samsung. Previous generations of the Ax have been fab’d by Samsung and in July it was announced that TSMC had picked up an Apple contract. Intel has in the last year begun to market, albeit quietly its excess fab capacity. This is an entirely smart move by Intel. It will help them use their multi-billion dollar fabs to the fullest capacity, as well as test and experiment with other designs.
Apple using Intel as a contract fab makes sense, for Apple. Intel has the best fab technology in the world, bar none. Apple is less concerned with competing on price, than they are making the best devices possible. To have the best devices you need the best (fastest and lowest power) chips. To have the best chips you need the best processes, and that means Intel. None of this is in question. If the A7 is fab’d by Intel it will greatly help Apple attain its market leading position. However, what is in question is whether this is a ‘huge win’ for Intel. One blog even referred to Intel making Apple chips as “That’d be a hell of a score for Intel.” In reality this will have little benefit to Intel. Certainly not financially. Lets look at why.
The Apple A6 (which probably would make a good candidate for TSMC to fab) is estimated to cost $17.50 each (at market introduction of the iPhone 5, prices tend to go down). Since the release of the iPhone 5 Apple has sold over 57 million of them (in a year). Intel’s net profit margin is around 15-20% depending on the quarter. Margins obviously are lower on less expensive devices so we’ll estimate a 10% margin on the A7 and a $20 price to Apple, that may be on the low side but it works for comparison sake. If Apple buys 50 million of them, then its a $100 million profit for Intel. Its unlikely that Apple will sell that many iPhone 5s’s due to the fact that they split the market with the new iPhone 5c but we’ll use 50 million anyways. Last quarter Intel’s Net Income was about $2 Billion. That makes Apple’s $25 million per quarter contribution roughly 1% of an increase. A one percent revenue gain is certainly not a “hell of a score” for Intel. Add into it Apple’s very strong history of pressing suppliers on price, decreased cost over time, and the iPhone 5c and the profit will be even smaller.
For comparison sake, lets look at another Intel foundry client, Altera, which Intel picked up early this year, to make their high end FPGAs. Altera chose Intel for the same reason, Intel has the best processes and Altera needs the best to compete with Xilinx. Intel will be making the Altera Stratix lone of FPGAs. Retail on a Stratix starts at $200 and goes up from there, WAY up, topping out at well over $30,000 PER DEVICE. Certainly this will have a larger effect on Intel’s bottom line, without the worry of price squeezing. Intel itself only averages a sale price of $107 per device. So yes Apple having Intel fab their A7 is a win for Apple, but it is merely business as usual for Intel.