About a year ago, Intel (INTC -1.60%) indicated that its first SoFIA products targeted at the low end of the mobile processor market would be built on Taiwan Semiconductor's 28-nanometer manufacturing technology. However, management also claimed that by the end of 2015 or early 2016, SoFIA would be built in-house on Intel's 14-nanometer manufacturing technology.

During a question-and-answer session at the company's 2014 investor meeting, Intel CEO Brian Krzanich made an interesting statement: "2015 is the first year where you'll start to see 14 nanometer, or leading-edge products, in that space. The first ones will be Cherry Trail. [In] 2016 you'll see probably [emphasis mine] SoFIA products."

Normally it would not make sense to read too much into this, but I'm starting to think SoFIA on 14 nanometers is going to arrive a bit later than the "first quarter of 2016" time frame given at the 2013 investor meeting and on the first-quarter 2014 conference call.

Looking at the bigger picture
At a recent investor presentation, Intel mobile group chief Hermann Eul had the following to say with respect to the SoFIA product launch schedules:

And now last but not least, the value on entry part, so Bay Trail entry is currently the workhorse and that will be phasing over the designs toward a SoFIA 3G, 3G-R, and SoFIA LTE. They kick in one quarter after the other. So, we will qualify SoFIA 3G by the end of this year, volume ramp in the first quarter, a quarter later the 3GR, a quarter later the LTE. [...] In 2016, we will be seeing also a 14-nanometer version of the SoFIA LTE.

If I understand what Eul is saying here, SoFIA 3G should ramp up into volume in the first quarter of 2015, SoFIA 3G-R in the second quarter, and SoFIA LTE in the third quarter. Eul also confirmed the 14-nanometer SoFIA LTE 2 for 2016 but did not get any more granular.

Also keep in mind that Intel recently inked a deal with Chinese chip vendor Spreadtrum whereby Spreadtrum is expected to build derivative chips based on Intel's base SoFIA LTE architecture for both companies to distribute. According to Krzanich at the investor meeting, the first fruits of that collaboration will arrive near the end of 2015. Intel has not specified what process geometry those products will be built on, but the timing suggests they will be 28-nanometer parts.

Where does that leave SoFIA LTE 2?
According to Intel, SoFIA LTE 2 will target the same markets that the upcoming SoFIA LTE products target: low cost LTE-capable smartphones. But what would happen to SoFIA LTE and the Spreadtrum derivative(s) scheduled to launch in late 2015 if Intel launched a 14-nanometer SoFIA LTE 2 in late 2015 or even the first quarter of 2016?

For starters, 14 nanometer should be fully ramped up at this point, and the process will be much denser and offer much higher-performing transistors than the 28-nanometer process used in the original SoFIA LTE. Furthermore, Intel CFO Stacy Smith claimed at the 2013 investor meeting that moving SoFIA to the 14-nanometer process would bring "an even further cost benefit," implying that it should be cheaper to build as well. A 14-nanometer SoFIA LTE 2 in early 2016 would utterly destroy the value proposition and useful lives of the 28-nanometer SoFIA LTE chips.

This would be acceptable if Intel were solely responsible for the design and distribution of the 28-nanometer chips, but with the partnerships involved here -- in particular the Spreadtrum relationship which is focused on phones -- it doesn't seem likely that Intel could launch a 14-nanometer SoFIA LTE 2 in early 2016 without a negative impact on this partnership. 

If Intel and its partners actually issue 28-nanometer SoFIA LTE products in the third and fourth quarters of 2015, I would be very surprised if 14-nanometer products in that same market segment follow in the first quarter of 2016. My guess is that SoFIA LTE 2 on 14 nanometer could at best qualify for production by the end of the first quarter of 2016 for steeper ramp-up in the second and third quarters of 2016.