The Apple/TSMC Partnership Bears Surprising Fruit - In China

Summary

TSMC's announcement of a new chip production facility in China is a surprise, given that the project was on hold as of October.

I believe the decision to go ahead with the plant stems from the need to expand production as a result of its supplier relationship with Apple.

The plant is indirect confirmation that TSMC is elbowing Samsung aside to become a sole source for Apple's custom systems on chip.

Taiwan Semiconductor Manufacturing Company (NYSE:TSM) announced on December 7 its intention to build a new wafer fab on mainland China. Investing in the plant represents a statement of confidence in TSMC's future growth, a confidence that probably stems from a strengthening supplier relationship with Apple (NASDAQ:AAPL).

(click to enlarge)

Source: TSMC

Offloading Production

The announcement of the plant is very surprising considering that it was on hold as recently as TSMC's Q3 earnings report in October. At the earnings teleconference, CFO Laura Ho had this to say:

My last comment is regarding China investment. In the last several months we have been actively evaluating the potential investment of a 12-inch fab in China. The consideration includes the positive development of a semiconductor market in China and a large pool of engineering talent. On the other hand the manufacturing costs will be higher in China than in Taiwan due to lack of economy of scale and the many other reasons. So as of now we are still evaluating the potential investments.

The plant announcement is all the more surprising given that TSMC pointed to a slowdown in the Chinese smartphone market from the beginning of the year. This slowdown was somewhat offset by the ramp of 16 nm FinFET in July and the resulting revenue from Apple. Nevertheless, TSMC guided to a 9% revenue decline for Q4.

TSMC expects that the Nanjing, China plant will produce about 20,000 300 mm (12 inch) wafers per month. The added production of 240,000 wafers per year will increase TSMC's current 9 million wafer per year production capacity by about 2.6%. The added capacity is roughly equivalent to producing about 40 million Apple A9 systems on chip (SOCs) every quarter. This is a very large chunk of what Apple consumes on average.

What tipped the scale in favor of the plant? Going forward with the plant indicates to me that TSMC and Apple have reached a long-term supplier agreement. This gave TSMC the financial visibility to plan for an expansion that won't be completed for two and a half years. It's estimated that TSMC currently provides about 37% of A9 production, while Samsung (OTC:SSNLF) provides the rest.

The percentage will probably increase going into next year until Samsung is phased out. Apple is motivated to phase out Samsung in order to not be dependent on a foundry that is also owned by Apple's most significant rival in the smartphone market. Apple would necessarily have to reveal proprietary design information to the foundry. In such cases, the supplier would "firewall" its customer activities from the rest of the company. But I doubt that Apple was completely satisfied with whatever assurances of confidentiality it received from the Samsung foundry.

We're already starting to see rumors about the upcoming iPhone 7 and its A10 SOC. According to one report, TSMC is expected to fully take over production for the A10. I expect TSMC to replace Samsung as soon as it's able to ramp up production of its 16 nm FinFET process to meet the needs of Apple, sometime in 2016.

Right now, virtually every fabless semiconductor company in the world wants 14-16 nm FinFET technology, including AMD (NASDAQ:AMD), Qualcomm (NASDAQ:QCOM), and Huawei. There are only two foundries in production, Samsung and TSMC, so there's only so much production capacity to go around.

But I doubt that any Apple production will be ever be done at the plant. The purpose of the plant will be to service Chinese fabless semiconductor makers, according to TSMC's announcement. But it will also serve to offload production capacity in Taiwan, where Apple's A9 SOCs are being made for the iPhone 6s/6s Plus.

Racing to 10 nm

While TSMC's Nanjing plant services its Chinese semiconductor customers on the 16 nm node, TSMC will have been in production for about a year on the next 10 nm node at its facilities in Taiwan. An article in the NY Times also pointed out that the process node at Nanjing would lag what was in production in Taiwan:

In a nod to concerns about advanced chip protection technology being leaked to Chinese competitors, Taiwan Semiconductor said in a separate statement that by the time the facility in China is up and running, an advanced operation in Taiwan would be churning out chips a generation more sophisticated.

Clearly, early 10 nm production will be reserved for Apple, as the early 16 nm production was.

Keeping Apple's SOC production in Taiwan helps to further assure Apple that its proprietary information will be protected while also assuring Apple of first priority on the most advanced production nodes in the future.

TSMC's Taiwan facilities also provide Apple with a safe place to invest its offshore cash. As I pointed out in my article on TSMC's Q3 results, TSMC's lower capex spending for the year strongly suggests that Apple is furnishing funds for capital equipment for TSMC's production facilities. TSMC lowered its 2015 capex estimate to $8 billion from $10-11 billion.

I believe that TSMC was able to lower its capex spending due to an influx of capital from Apple in the form of funds for Apple owned equipment. Apple does this frequently with suppliers. As of the end of the 2015 fiscal year in September, Apple reported in its SEC 10-K that it had $7.3 billion in off-balance sheet obligations for product tooling and manufacturing process equipment, among other things.

With help from Apple's considerable financial resources, TSMC will probably be the first ARM foundry to bring its 10 nm FinFET process into production in early 2017, as TSMC indicated in the Q3 teleconference. Intel has indicated that it will ramp 10 nm production in the second half of 2017, so TSMC will probably beat Intel, and may even beat Samsung as well.

Unlike what happened with Samsung's 14 nm FinFET process, in which Samsung was able to employ it first in its own SOCs for the Galaxy S6, Apple won't have to wait for access to TSMC's 10 nm process. Apple will be assured priority by virtue of its investment in plant and equipment.

Investor Takeaway

Use of the 10 nm process will make Apple's processors among the most energy efficient available in any mobile device. As I pointed out in my Theory of Apple's Product Success, the growth of mobile devices has been driven by their greater energy efficiency. The most efficient devices enjoy a competitive advantage compared to less efficient devices.

Consumers don't necessarily realize that they're preferring energy efficient devices, but they do consciously want the by-products of that efficiency: long battery life while maintaining high performance. Apple's iPhone 6s has been demonstrated to provide high performance for games, video, and web surfing for extended periods of time. This is because Apple doesn't throttle the processors by reducing clock rate in order to conserve battery power.

Reaching 10 nm will be a significant milestone for Apple and TSMC. It will demonstrate the power of the paradigm shift that Apple pioneered in designing its custom SOCs. It will provide Apple with a significant competitive advantage compared to SOCs fabricated on older nodes, and further fuel the growth of Apple and TSMC. I recommend both TSMC and Apple as buys.

Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.

Source: The Apple/TSMC Partnership Bears Surprising Fruit - In China