Will the loss of one of its star generals signal Intel’s retreat in the battle for the mobile market?
Intel recently bid adieu to one of its brightest executives, Senior Vice President of the Ultra Mobility Group, Anand Chandrasekher. As Mr. Chandrasekher leaves “to pursue other interests”, there is a lot of speculation to this being linked to Intel’s lack of an effective front in the battle for the smartphone and mobile devices market. Regardless of the many rumors behind Mr. Chandrasekher’s exit, Intel Architecture Group’s Executive Vice President David Perlmutter, has assured us that Intel will continue “to make the investments needed to ensure that the best user experience on smartphones and handhelds runs on Intel architecture, and to ship a phone this year.” It will be up to Intel’s interim co-managers, IAG Vice Presidents Mike Bell and Dave Whalen, to execute on that promise. But why has this been such an uphill battle and will Intel step up the attack or ultimately retreat?
To understand why Intel is waging a seemingly unwinnable and costly battle, we need to first understand Intel’s motivation. Its current market base is manufacturing processors for computers; which it does exceptionally well with over 80% of the central processing unit (CPU) market. In terms of revenues, this market represents over a third of the overall processor market. So why should Intel care about handsets? In the Comprehensive Processor World Report, 2010 Edition, I predict desktop CPU revenues are expected to stagnate over the next five years with a continual average growth rate (CAGR) of 1% from 2009 to 2014, while processor revenues for portable computing and media devices including notebooks, netbooks, smartbooks, tablets, eBook readers and smartphones are each expected to grow at a CAGR near 20% or more. By 2014, mobile handsets alone are expected to comprise over 18% of the total processor market. This consumer paradigm shift to mobility represents a battle from which Intel can ill afford to retreat.
Intel has a legacy of supplying some of the highest performing processor solutions available, and they are based on x86 architecture. Its latest foray into the mobile market, the Atom processor (codenamed Moorestown), was a drastic reduction in energy consumption over even its highly efficient Core2 processors for notebooks. It also represents a major shift in architecture toward the system-on-chip (SoC) solutions common to mobile devices instead of the CPU and associated northbridge and southbridge chipsets common to computers. Despite these improvements, it still met with heavy resistance from mobile device manufacturers. Even though it was leaked that LG would produce the first smartphone armed with a Moorestown, Intel has yet to see any top tier smartphone releases with an Intel processor. So what does Intel need to do to make good on Perlmutter’s assurances?
Intel still has the following hurdles to overcome before any serious battle can be waged. First, consumers demand battery life; no matter what the media functions may be, the device is still, first and foremost, a phone. If Intel is asking a smartphone manufacturer to take a leap of faith by designing on x86, thus risking damaging a relationship with ARM, the designer of over 90% of the application processors in handsets, it had better be more than barely adequate at energy efficiency. Second, the OS has to allow the Intel processor to shine. Consumer response is strategically important and Intel must find a way to generate a uniquely positive user experience. Because mobile operating systems have evolved targeting compatibility with a much lower range of processor performance, they are not optimized for a high performance processor and any new OS will also have to enable easy transition for apps programmers to provide that killer app. Finally, Intel must enter with enough impact to get noticed. As Moorestown demonstrated, it is counterproductive to have a chip on the market that is largely ignored. Intel needs an ally to release a captivating new smartphone on an Intel core simultaneously with any new marketing campaign in the smartphone arena.
To overcome the first barrier, Intel should be jumping into the battle with both barrels blazing with a 22 nm cutting edge design. Intel’s 32 nm manufacturing used in the design of the next generation mobile processor, Medfield, is promised to have competitive standby time and longest active power time, and if it truly is the fastest processor and includes advanced graphics as promised by Anand Chandrasekher before his exit, Intel will have overcome the first hurdle, but if not (which is more likely) one has to wonder why it did not jump to the 22 nm process? The answer is likely that the uncertain mobile SoC market lost the war in the board room to the strong income generating notebook CPU market for the limited 22 nm fabrication. In the long run, this may have cost them greatly.
Next Intel’s work on an operating system that will allow x86 to shine is moving very sluggishly. Intel has been working with Nokia to develop MeeGo, a Linux based OS, but Nokia continues to delay the release of a handset with the MeeGo operating system. Nokia has entered into an agreement with Microsoft and the latest word is that it has dropped the anticipated N9 which was to debut MeeGo. However, Nokia CTO Rich Green assures that the N950 scheduled for release this year, will feature MeeGo. It isn’t clear if there is any truth to rumors of Nokia including an x86, but it is more likely that the N950 will include an ARM-based Scorpion from Qualcomm. If MeeGo fails to take the field, Intel may still have some cavalry coming over the hill from Wind River, its recently acquired embedded OS provider.
The hunt for an ally seems to be the biggest hurdle of all. LG’s announcement last year of the Moorestown armed GW990 never materialized, and this failure creates a hurdle of its own. Clearly there is a lot of reluctance for a tier one smartphone vendor to risk its relationships with their entrenched ARM-core based suppliers such as TI, Qualcomm, Samsung, and many others. Despite Intel’s assurances that it will be in a smartphone by the end of the year, it is more likely that we will see a stronger front in other portable media devices. This should be the more valid beachhead with the netbook, smartbook, eBook reader and tablet processors predicted to demonstrate a 32% CAGR from 2009 to 2014. There is opportunity to incorporate higher performance features, and the power and size constraints are more tolerable. With some tablet design wins to compliment Intel’s dominant netbook market share, there will be much more opportunity in a growing market and an easier sell to a smartphone manufacturer.
While it will be a pleasant surprise to see an Intel processor debut in a smartphone this year, if it does not, it would be neither wise nor reasonable for Intel to surrender this market. A comprehensive view of the processor market clearly shows a trend toward convergence on mobility. Intel must recognize the need to establish a front in the battle for mobile convergence. This is one case were we can be thankful that the war will probably wage on.