Analysis
The microprocessor world is increasingly exciting as x86 and ARM begin to compete in phones, tablets and other markets. As Intel in particular begins to encroach on smartphones and tablets, ARM and their partners are certainly eyeing notebooks with a degree of eagerness. For any vendor, winning Apple’s business is a big endorsement, and one that gets many customers excited. It would be a huge victory for the ARM ecosystem.
While it is a very interesting thought experiment, the truth is that Apple will not switch from x86 to ARM for their notebooks in next few years. Chris Foresman at Ars Technica is right to be skeptical, and we would go a step further and state that it is not possible. There is no motivation for Apple and a host of significant technical and business challenges. Just to mention a few:
- ARM microproccesors are designed for lower performance and unlikely to match x86 performance in the next few years
- High-end ARM designs may not have a power efficiency advantage over x86
- Emulating x86 on ARM is necessary for compatibility and reduces performance and efficiency even further
- Thunderbolt I/O is probably not available on ARM
- No second source ARM vendors for key components such as the GPU
- Intel and AMD already do an excellent job competing to meet Apple’s needs
It would be very expensive for Apple to design their own microprocessors – perhaps in the range of $200M or more. The entire notion is incredibly risky, given Apple’s lack of experience with end-to-end chip design, especially for high performance microprocessors and graphics. It would be far more sensible and typical of Apple to sit back and encourage other companies to take the risk in the hope of securing Apple’s potential business. Nvidia would be a prime candidate here, but most of the technical and business problems are equally applicable to third parties.
The advantages of an ARM notebook are not particularly impressive. There will be no performance improvement, and power efficiency is likely to be similar to x86. Even if Apple switched all their notebooks to ARM, they would still need x86 for desktops and servers; the software ecosystem would continue to span both instruction sets. There would be some advantages to complete vertical integration for Apple, but it is hard to see any game changing differences. The cost savings are fairly small as well. More importantly, Apple’s computer line is in pretty good shape today, and taking such a big risk for relatively little gain makes no sense.
Strategically speaking, Apple should be focusing their limited chip design resources on the iPhone and iPad. The rise of Android and the potential commoditization of smart phones and tablets is a huge threat to Apple. Right now, Apple is the leading smart phone vendor and pretty much owns the tablet market. In their last quarter, over 50% of their revenues come from the iPhone. For Apple, carefully designing their own chips and maintaining their leadership in phones and tablets would have a tremendous impact. In comparison, producing ARM notebooks that are largely similar to the previous generation would do very little to boost Apple’s sales.
While Apple will not switch from ARM to x86 in the next couple of years, there could be shades of the truth in the rumors. A more plausible scenario is that Apple is planning some sort of hybrid system. There are a number of ways that putting ARM near a computer could be quite interesting and valuable. For example, Apple could include an ARM core in future notebooks for ‘instant-on’ web browsing while the computer is booting – something done in certain Dell and HP models. An extension of that idea would be an update to OS X with an iOS emulator and maybe a co-processor for future hardware. Unifying Apple’s three major platforms (computers, tablets and phones) would make it far easier for developers to target Apple platforms. In some ways, this might be the most sensible, since expanding the iOS installed base would help Apple create a more attractive market for developers compared to Android and could facilitate tying the iPhone to other Mac products.
Long term, ARM could become a viable notebook option. Apple has invested a lot of time and effort in portability (e.g. LLVM, OpenCL, OpenGL), so it is clearly possible. Apple has few attachments to x86 and would not hesitate to shift to a better alternative. However, there are no ARM designs that will meet Apple’s needs for performance and efficiency in the next two or three years. Even if there was, it is hard to see how such a design would be substantially better than AMD’s offerings. Over 5-10 years though, many of the technical and business hurdles may change. The ARM ecosystem is moving forward at a rapid pace, and Apple is watching carefully.
Discuss (95 comments)