According to The Wall Street Journal, Intel (INTC -0.36%) is reportedly going to be providing the main processor for the next iteration of Google (GOOG 1.02%) (GOOGL 1.02%) Glass. Now, while there seems to be an implicit assumption that Intel will be supplying an x86-based processor into this device, some have speculated that it may be ARM (ARMH) based, as in the case of the Intel MICA bracelet.
However, I believe that Intel will indeed be supplying an X86-based chip into this device rather than an ARM one. Here's why.
What was in the prior Google Glass?
The first Google Glass contained a Texas Instruments (TXN 0.50%) OMAP 4430 processor, which featured two ARM Cortex A9 CPU cores. In order for the next version of the Google Glass to be an upgrade from the prior one, it would need to sport a faster processor.
Of course, something based on a more modern ARM core, such as the Cortex A12/A15, would be faster, but to my knowledge Intel has no plans to build chips around these newer ARM cores.
Some potential answers
An obvious solution, then, would be to use Intel's very own Silvermont processor core. This is higher performing and more efficient than the 45-nanometer implementation of the Cortex A9 processors found in the OMAP 4430.
However, while it seems that Intel's Silvermont processor core would be more than up to the task to power this next generation of Google Glass, it's not clear based on that core which system-on-chip Google might use. There are a couple of options:
- Custom system-on-chip: Intel may be designing a custom system-on-chip tailored to the Google Glass based on its Silvermont processor core.
- Merrifield/Moorefield/Bay Trail: Intel may repurpose one of its 22-nanometer Silvermont based system-on-chip products (Merrifield, Moorefield, or Bay Trail) for the task.
- Edison: Intel recently introduced a very low power platform known as Edison which uses two low-clocked Silvermont processor cores.
I don't think it would be worth Intel's time to spend the money on a full-blown custom system-on-chip for the Google Glass; the sales volumes probably aren't there to justify such an investment. Merrifield/Moorefield/Bay Trail might be good candidates, but they could be overkill (and relatively expensive) for the Google Glass use case.
Edison, though, which is designed specifically for embedded applications, may be the best bet here. The problem with Edison is that apparently the on-board graphics processor is disabled. However, according to one Intel employee in the forum thread linked previously, Edison's graphics were fused off because "the use cases around what is intended for Edison do not need video." The Google Glass would obviously be a use case that would seem to require video.
At any rate, it's probably X86
While some of the wearable devices that Intel supplies ultimately use ARM chips, I strongly believe that in this case, the Google Glass will be Intel X86 based. Intel can probably use at least one of the many 22-nanometer products in its portfolio for the task. Further, it is strategically important for Intel to push X86 here in order to have developer mindshare for whatever application ecosystem could arise from Google Glass and related devices.