While the processor design may have been a factor in Apple’s success in the early stages, I would argue that this is a diminishing factor going forward.
I do agree that Apple is very good at optimizing user environments and applications to run on the company’s devices for maximum performance and power efficiency, but this is really a software expertise.
This expertise could be applied to processors from other vendors, maybe not as easily, but it could be done. In addition, as we have seen with both smartphones and tablets, the user interfaces, applications, content, and services are really the key differentiators, not the processor.
While you may get slightly different experiences from a smartphone with an Apple A5, Nvidia Tegra, Qualcomm Snapdragon, TI OMAP, or Samsung Exynos processor, most of the differences are driven by the features, services, and software optimization. And to many users, the difference between processor, device, and/or software generations is much more noticeable than between the various processors of the same generation.
In fact, the most noticeable difference between devices, the battery life, has more to do with the user settings, displays, and antennas than the processor. So, while it might make sense for smartphone OEMs that already have the investment and expertise in designing processors, it would seem to make little or no sense for other OEMs to start designing processors.
I am addressing this issue because of recent market rumors. In mid-2011 rumors that LG was considering designing processors for the company’s mobile devices, and just recently rumors that Huawei was designing processors for the company’s smartphones.
While there is no confirmation that the rumors are true or false, I have no doubt that these and other smartphone and tablet OEMs are pondering the potential of designing processors for future devices. I would, however, argue against pursuing these ambitions.
Not only would the costs and risks be high, the potential for a positive return would be low. Every semiconductor entity would like to think that they can sell their products to multiple customers, but history tells us otherwise. Nokia stopped designing processors years ago when other solutions became available, Motorola Semiconductor (now Freescale) failed to land a significant design win outside of the Motorola handsets, and Samsung Electronics is facing similar challenges.
While I indicated that this question has and will continue to arise as new electronic segments emerge, some things have changed.
The first is the issue of manufacturing. With the vast majority of new processors being manufactured through semiconductor foundries, companies must no longer make capital investments for fabs. However, the cost of getting a product to manufacturing does continue to rise dramatically.
There is also the issue of reuse. We are now in a period where almost all processors or logic control units are considered system-on-chips (SoCs) because they are very complex solutions that combine what were once discrete functions like processing cores, arithmetic units, graphics processing units (GPUs), video processing units, audio processors, memory, I/O, and wireless interfaces on a single piece of silicon.
As a result, these processors are a combination of multiple functional blocks combined through a high-speed network. Each functional block may have a different lifecycle depending on advancement in the technology or industry standards, which means that not all blocks will change from one processor design to the next even when targeting different applications.
A good example of this reuse is tablets. Instead of designing new processors for tablets, tablets began using processors designed for smartphones. While the two may diverge in the future, they are likely to share common characteristics. This is not to say that innovation in processor design is dead, because new electronic products are likely to drive innovation in processor design, and the question of make vs. buy will arise again.