5 reasons why Apple should build its own iPhone SoC (and 5 why it won't last)

  • Opinion – News that Apple in fact will build its own iPhone processor sparked lots of controversy this week. Even within our own staff, we got caught up in heated discussions that did not exactly meet on common ground, but were carried out with same passion our readers obviously enjoy to discuss Apple news. So we decided to take our argument into the public and ask you to chime in and let us know who has the more reasonable view – our in-house Apple enthusiast Christian Zibreg or Managing Editor Wolfgang Gruener, who admits to admire Apple's achievements but be critical of the company's attitude.

    Christian: Apple-designed SoCs make sense!
    1. Cutting-edge gadgets with killer hardware. Today's iPhone uses off the shelf parts accessible to others (such as PowerVR MBX graphics cores also found in Nokia’s N95). A deal with ARM and Imagination will give it access to next-gen mobile graphics and processor core designs and exclusive rights to use them in a custom SoC. Exclusivity and broad access to the intellectual property enable Apple to design a SoC with one-of-a-kind features that other vendors will not be able to copy and keep them at a distance.

    2. Cheaper gadgets. Apple may put custom processor and graphic cores, memory, sound, and interfacing components on a single die and outsource the manufacturing to Samsung. Cost-wise, this strategy beats buying separate components anytime, especially with high volume SoC manufacturing at an output of 100 million units annually. Design and volume impacts the bill of materials, which should also positively impact the price of the entire product.

    3. Secret new products will remain secret. Apple's reliance on commoditized suppliers makes rumors boring. We knew pretty much everything there was about the latest iPods weeks before Jobs took the stage to unveil them. This will change. Apple's in-house team of P.A. Semi SoC engineers is working under the same roof with software and hardware guys - the curtain will remain pulled over mysterious new products until the last moment. The best Steve Jobs' "one more thing" moments are yet to come.

    4. More battery life. Apple gadgets and long-lasting batteries don't go together. But power efficiency is P.A. Semi's field of expertise: Its 2 GHz dual-core processor consumed 4-5 times less power than other designs in its class at the time. Moreover, SoC designs tend to consume less power than the separate components it replaces. The combination of the two can turn Apple's Achilles’ heel into a key advantage. Expect future iPhones, iPods and Mac notebooks to have longer battery life than comparable competing products.

    5. Apple to become the mobile gaming leader. Only Apple has exclusive rights to adapt Imagination's next-gen PowerVR SGX (graphics) and VXD (video) core designs in its own SoC. SGX delivers hardware-accelerated, shader-based 3D graphics and OpenGL ES 2.0 compatibility. VXD plays HD video on a mobile device or external display, with power consumption comparable to the existing audio playback chips. Future iPhone will have killer graphics, no question about it. In fact, future iPhone will render Nintendo DS and Sony PSP obsolete and turn Apple into a key player in mobile gaming.

    Read on the next page: The iPhone won’t last 5 years with Apple’s own SoC!


    Wolfgang: The iPhone won’t last 5 years with Apple’s own SoC!

    1. Losing focus. Let’s be realistic here. Apple builds great iPods and iPhones, but it isn’t a semiconductor company. Apple’s success is based on its talent to the innovative use of existing chip technologies, put a pretty package around them and integrate those products with its software. It is what Apple does best and a decision to build chips is an unnecessary distraction from its core business, which Apple may regret a few years down the road.
    2. Can’t react to market shifts quickly enough. The micro-processor market is highly competitive and any established company in this market isn’t aiming for second place, especially Intel. Transmeta felt it, AMD felt it and ARM may have to rethink its business approach as well. If market trends shift, for example to new chip design requirements that impact performance and power consumption, it is integrated R&D and manufacturing that counts and a few hundred chip designers and the reliance of a whole chain of companies can become a tremendous disadvantage for Apple. It happened with the Power architecture and it is just as likely to happen with a new chip again.

    3. Quality in limbo. Here is what I don’t get: Apple acquires the design talent of 100-or-so chip designers who created a great processor design before. It will base its designs apparently on ARM blueprints and manufacture the chips with Samsung. But how will Apple manage the validation of these chip designs? AMD has several hundred validation engineers, Intel a few thousand. I just don’t see that Apple has enough horsepower to reach the level quality of these chips across design and manufacturing when compared to AMD or Intel. And just ask those two guys (or Nvidia) how errors in your own chips can impact financials. When an error happens, Apple will be on the hook - and not its suppliers.

    4. It’s the software, stupid! Software is one of Apple’s great advantages today, especially form a developer’s view. There is a great ecosystem, lots of willing buyers and a very efficient distribution platform. But there is no doubt that it is a platform limited to a phone that still makes up only about 1-2% of the market and if we believe the most optimistic estimates about 5% next year. The problem with cellphone software today is that developers need to write dozens of versions of their software to reach all or most platforms. If Intel’s 32 nm Moorestown SoC is as good as Intel claims, it could also bring a big advantage for developers, especially if made available in Android phones. Moorestown runs on the x86 software stack, which means that every software that was developed for a PC will run on a Moorestown cellphone as well.  AMD is expected to go the same way. From the developer perspective, x86 can turn into a much more attractive platform than the iPhone Mac OS X modification – if Intel and AMD can attract a significant competing platform such as Android.

    5. Need for speed. I just don’t believe that an optimized ARM design will be able to beat a ground-up design that may come from Intel and AMD – or a design such as Nvidia’s Tegra, which has a big focus on graphics. Designing competitive chips is serious business and the industry is a constantly moving target. Resources, and experience come into play here again and that is exactly what Apple lacks. If the first iPhone SoC isn’t close to its rivals, Apple will have a huge problem. On the other side, if it can pull a miracle chip out its hat, it could extend its iPhone business dramatically, but spark a new performance race and switch to someone else in the end. We have seen that before and it may happen again.

    Christian’s rebuttal

    What you call "loosing focus" is Apple adding one more variable, actually a crucial one, to the equation. Apple is not a newcomer to the SoC field, they designed SoCs for PowerPC-based Macs in the past. Innovation IS Apple's focus and Apple now chooses to innovate the chipset. Changes in hardware platform won't affect iPhone developers because of OS X's platform - and processor-agnostic design. I agree that hardware changes will come down to the software - another area of Apple's expertise. Accelerometer is perfect example how Apple trickles down hardware features to the software level in innovative ways.

    You are underestimating Apple's vertical integration. Apple will soon become the only company in the world that makes its own unique SoC design, hardware, operating system and tailored applications. Finally, Intel. To paraphrase Jobs, "I wouldn't lose too much sleep over that." Apple is not going after Intel's market and Intel is not pre-determined choice for Apple gadgets. Intel better works harder to bring the already overdue Moorestown to the market because others (ARM, Samsung, Imagination, etc.), not Intel, are reaping the benefits of MIDs' sudden popularity.

    If Intel delivers a killer SoC platform for MIDs down the road, Apple will switch in no time. I have no doubt in my mind that Intel will give Apple preferential treatment once again and provide the company with a "special" version of its design, like it did with the processor for the iMac. When your key customer is most respected consumer electronics company in the world, with magnitude of orders measured in tens of millions units, it's a no-brainer.

    Wolfgang’s thought: Let’s see.

    I am not convinced that Apple can compete in the semiconductor space just because it has done so (unsuccessfully) in the past and purchased a chip designer that may have some limited success in vertical markets, but lacked access to cutting edge manufacturing. Can Apple combine all the puzzle pieces? I don’t know, but it is clear that it won’t be easy.

    In my mind, if Apple in fact designs its own iPhone processor it will either be a failure or a success. The smartphone market is competitive enough that even good-enough processors could be considered as a disadvantage in the not-too-distant future. In a best case scenario, Apple may surprise other manufacturers initially, but as said: This is not a second place industry. Especially Intel will go after this market vigorously and Apple simply does not have the experience and resources to match what Intel (who will go after the smartphone market with its 2009/2010 Moorestown chip) may be able to come up with and may be forced to switch sooner or later again.

    What are your thoughts? Chime into the discussion by writing a comment below.