Intel (INTC +7.06%) has tried for years to gain significant share in the mobile device market. Although the company was able to muscle its way into the smaller and less profitable tablet market, it has yet to make meaningful inroads into the smartphone market beyond a handful of "point designs."
In order for Intel to be able to capture share in the smartphone market, I believe that the company needs to show potential customers that it can consistently churn out products with leadership performance, power consumption, and feature integration in a timely fashion.
One of the key drivers of product competitiveness is the manufacturing technology upon which its chips are built. Although there is far more to the competitiveness of a semiconductor chip than the underlying manufacturing technology, there's no denying that all else equal, better manufacturing technology means faster, more efficient processors.
Intel has long claimed that it has a significant lead in achieving cost-effective, high-volume production of advanced manufacturing technologies relative to its competition. However, with the company's recent public disclosure of a delay in the launch of its PC processors based on its 10-nanometer manufacturing technology, the notion that Intel has a manufacturing "lead" when it comes to mobile devices is called into question.
When should we expect a 10-nanometer Atom processor?
Intel generally releases two major variants of its chip manufacturing processes. The first variants out of the chute are its CPU-oriented processes, and the second variants are its system-on-chip oriented processes which are suitable for, according to Intel in a technical paper, "a wide range of system-on-chip products including tablets, smart phones, ASIC, embedded, Internet-of-Thing [sic], baseband and RF products."
Historically, there has been a significant delay between the launches of the first products based on Intel's high-performance CPU processes and its more general system-on-chip processes. At the 22-nanometer generation the delta was approximately a year and a half, and at the 14-nanometer generation, the delta shrank to roughly six months.
If we assume that Intel launches its flagship Cannonlake CPUs on its 10-nanometer CPU-oriented process in the third quarter of 2017, then applying a two-quarter delay (as we saw with 14-nanometer) would mean that we won't see the first 10-nanometer Atom parts from Intel until sometime in early 2018.
What does this mean for Intel's competitive positioning in mobile?
Both major semiconductor logic foundries have signaled that they plan to begin volume production on their respective 10-nanometer manufacturing processes by the end of 2016.
It takes time to go from production start to products on the shelves, but I'd be quite surprised if the iPhone 7s, which is expected to launch in the fall of 2017, didn't feature a processor built on a foundry 10-nanometer manufacturing process.
If the foundries are ramping 10-nanometer mobile chips for broad availability in the second half of 2017, and if Intel doesn't launch its first 10-nanometer Atom parts roughly when it launches its 10-nanometer PC-oriented Cannonlake parts, then Intel will actually be behind the foundries (and their top fabless customers) in terms of mobile-oriented manufacturing technology for a time.
On the bright side, Intel has minimal share in the smartphone market today, so it's not as though Intel stands to lose significant share if it is "beaten" to a manufacturing node in mobile chips. That said, if Intel ever wants to be a competitor in the mobile market, manufacturing node parity with the top foundries/fabless customers is -- in my view -- a necessary but insufficient condition to succeed.
I look forward to hearing what Intel management has to say about the competitiveness of its mobile roadmap at its investor meeting this November and hope that any such discussion includes a view of the competitive environment with respect to chip manufacturing technology.
