The Mystery of the Apple Inc. A9 Chip Continues

Ahead of the launch of the Apple (AAPL -0.57%) iPhone 6 and 6 Plus, there was plenty of speculation as to which contract manufacturer would be handling the manufacturing of the A8 chip found inside of the devices. However, following an analysis from Chipworks, it's pretty clear that Taiwan Semiconductor (TSM -4.86%) is the manufacturer.

The A8 is now old news, though, as investor attention now seems focused on which company will handle the manufacturing of Apple's next generation system-on-chip, presumably dubbed the Apple A9.

I've seen the following possibilities published:

• Samsung (NASDAQOTH: SSNLF) will build these chips on its 14-nanometer technology. 
• Taiwan Semiconductor, or TSMC, will build these chips on its 16-nanometer technology. 
• Both Samsung and Taiwan Semiconductor will build the chips on their respective 14- and 16-nanometer technologies.
• Taiwan Semiconductor will build it on the same 20-nanometer technology that is currently being used to manufacture the A8 processor inside of the iPhone 6 and 6 Plus.

While Samsung has apparently confirmed that it's going to supply 14-nanometer FinFET chips to Apple, it's worth taking a look at the final possibility raised here: that Apple will be using TSMC's 20-nanometer process again.

There's a precedent for that: Intel's tick-tock
Interestingly enough, Intel (INTC -1.79%), the world's highest volume logic chip manufacturer, tends to transition to new manufacturing technologies every two years. In fact, the methodology it has adopted internally is a strategy known as "tick-tock."

Tick-tock essentially stipulates that, in any given year, Intel is either transitioning an existing design to a new manufacturing technology ("tick"), or is releasing an entirely new chip design on proven technology ("tock").

Over the years, as chips have become about much more than CPUs, the distinction at a chip level between a "tick" and a "tock" has become unclear. For example, with Intel's 14-nanometer Broadwell chip (a "tick"), the company included a completely redesigned graphics processor, and made a number of enhancements to the CPU core.

Nevertheless, the precedent is clear: Intel generally uses a given manufacturing technology -- although Intel has indicated that it fine-tunes a given process technology over time -- for two CPU generations.

Will Apple do the same at 20 nanometers?
Apple's adoption of new technology nodes has been impressive. Below is a table of each Apple chip generation, and the manufacturing technology that said generation was based on:

Source: Wikipedia.

It looks as though Apple spent two years on the Samsung 45nm node, but then marched pretty quickly through the 32nm, 28nm, and 20nm nodes. While I might be inclined to expect that Apple will use the 20nm node for the A9 under normal circumstances, I don't think it makes sense here for one crucial reason.

20 nanometer has always been viewed as a stopgap
On TSMC's Oct. 2013 earnings call, then-CEO Morris Chang stated that TSMC views the 20-nanometer and the 16-nanometer nodes as "one node." The well-telegraphed idea here is that, in going from 28-nanometer to 20-nanometer, TSMC would get a significant scaling benefit, but it wouldn't see a huge increase in transistor performance/power consumption. Then, it would graft a new type of transistor onto the back end of line from the 20-nanometer node, known as a "FinFET," to create the 16-nanometer node.

TSMC wouldn't see a big chip area reduction in moving from 20-nanometer to 16-nanometer -- SRAM cell size decreases from 0.081 square micrometers to 0.07 square micrometers -- but the FinFET transistor promises to bring substantial improvements in performance and power consumption (TSMC claims 40% more performance and 60% power saving for its 16 FinFET+ over 20 SoC, though I suspect that "and" should be an "or").

Foolish bottom line
Given how much better the 16 FinFET+ technology seems to be over the 20 SoC technology currently in use by Apple, I expect that Apple will want to transition to it, and/or the Samsung equivalent, as quickly as possible to get a meaningful performance boost for its next generation iPhone.

The ball is, it seems, in Samsung's and Taiwan Semiconductor's respective courts. If one, or both, can deliver in time for the new iPhone, then 2015 iPhones will likely offer some pretty stunning performance. If not, then I suspect that Apple will still find a way to improve performance with a new architecture on the 20-nanometer process -- assuming Apple has a team working on such a "backup plan" -- but the gains over the A8 would likely be far less meaningful.