By as late as April 2014, Intel (INTC 2.97%) had been telling investors that its Broxton system-on-chip was slated for mid-2015 "production readiness." This part was expected to provide best-in-class performance and completely eliminate the need for "contra-revenue" dollars by having a very tightly optimized platform bill-of-materials cost.
However, by late 2014, it became clear that Intel had delayed the project into 2016, and indeed the company announced at its investor meeting this year that it would be a 2016 part. What I have spent a great deal of time thinking about, though, is why Intel decided to delay this project into 2016 rather than stick to its original "mid-2015" production schedule.
Could it have something to do with SoFIA MID?
On Intel's April 2014 earnings call, CFO Stacy Smith said the following:
And then we also have new products coming into the marketplace, like SoFIA, that's targeted at the low end, and then in 2015 you'll see Broxton, which is an SOC more for the mid-range to high-range of the market coming into our product portfolio.
Note that Smith referred to Broxton as a part for both the high end as well as the midrange of the market. Broxton, for those of you not familiar with the chip, is a stand-alone applications processor -- it doesn't integrate the cellular baseband processor.
Now, if we look at the roadmap presented at this year's investor meeting:
Source: Intel
Notice something interesting? Broxton is no longer targeted toward the "performance and mainstream" as it was in 2013 -- it's strictly a "performance" part. Notice also that a new chip just popped up on the roadmap known as SoFIA MID. This part is, as the name suggests, aimed at the "midrange" or the mainstream of the mobile market. It's a 14-nanometer part with a quad-core Atom processor and integrated LTE.
What does this part have to do with the Broxton delay?
As of last year, Intel had indicated that Broxton would be aimed at the mid-range and high end, while a 14-nanometer SoFIA (which we now know, per the investor meeting, is called SoFIA LTE 2) was intended for the "value" segment of the market. We now have three 14-nanometer mobile product families: SoFIA LTE 2 for the low-cost market, SoFIA MID for the midrange, and Broxton for the high end.
My guess is that Intel realized that Broxton aimed at the midrange wouldn't be sufficient and that a higher level of integration would be needed to be successful in that segment of the market. Furthermore since the midrange of the mobile chip market is likely larger than the high end, and since more of the midrange market is likely open to merchant chip vendors like Intel, my guess is that Intel pulled resources off of the Broxton project to work on SoFIA MID.
What if it's performance/feature problems?
Another explanation that makes sense is that the original variant of Broxton simply wasn't competitive enough. Intel executives used some very strong words to describe its performance. Here are some quotes from November 2013 from members of the Intel executive team:
- "[Broxton is] going to have a stunning level of performance, particularly in graphics." --Intel CFO Stacy Smith
- "Broxton; that is the high-end product. It will mark leadership performance for hero devices." --Intel Mobile & Communications GM Hermann Eul
If the first explanation (shifting resources to SoFIA MID) doesn't hold true, I wouldn't be surprised if Broxton were delayed simply because Intel underdesigned the part and the chip requires additional features/performance. These would require more time to implement, thus pushing out the schedule.
