Mobile Internet World: Intel faces tough battle in mobile devices

But the history of the WiMAX and Wi-Fi markets to date have revealed the gaps in Intel's understanding of the mobile device market--in terms of its supply chains, routes to market, complexity of interoperability, and the tradition of subsidies. While Intel and Google expect the mobile world to move to open access and the PC/Wi-Fi model, the carriers are finding a rich seam of business from applying subsidies to laptops and tying those supposedly free-ranging devices to their networks. The biggest breakthrough in bringing 3G to laptops without carrier obligation has come not from the PC community but from Qualcomm, with its Gobi product. And worst of all for Intel, even if its x86-centric view of what the mobile internet device will look like comes true in some or even many markets, it may still be pipped at the post in terms of providing the chips.

Intel has made great progress in recent years in reducing the power consumption of x86 and making it appropriate for miniature, mobile devices, as well as finally offloading all its own activities based on the main rival, ARM, to Marvell in 2007. But while the Intel machine has cleverly presented this as a sign that it no longer needed ARM because of the progress made by its key architecture, it is worth remembering that the sale of the ARM-based cell phone processor business, XScale, to Marvell was widely perceived as an admission of defeat, not victory, at the time. So in a head-to-head debate at the Munich summit--despite the glitz of the prototype devices that had been showcased in a rousing keynote by Pankav Kedia, director of the Ultra Mobility Group's global ecosystem at Intel--Texas Instruments' Bryce Johnstone, head of strategic marketing, still gained many murmurs of approval from the audience when he pointed out that the multi-billion unit installed base of ARM in the mobile world (a good slice of that down to TI of course) was unlikely to go away any time soon.

The issue is not just one of politics, though vendors have used support for ARM to make Intel's task harder--ARM is a company that is dependent on a wide base of licensees, rather than being a political powerhouse in its own right, and so is ideal as a mascot for its larger customers in the battle to stop Intel taking a major chunk of a mobile market that, in the conventional cell phone era, has largely excluded it. For users and carriers, the outcome of the vendor manoeuvrings is important because none of the grand plans for open web services, advanced multimedia content, social networking on a huge scale, or next generation user interfaces, can possibly materialize on the mobile device unless the chipmakers continue to optimize their architectures for high performance video and other rich content; for high speed and sometimes ad hoc connections; and for very low power.

Intel has made strides in these directions, and the mobile internet devices (the PC-derived category that the chip giant hopes to make dominant in the next generation of mobility) shown off by Kedia certainly proved that these gadgets can be as slim and attractive as an iPhone (certainly a long way from the first wave of Wintel "smartphones"). But it has an uphill battle against powerful entrenched interests with generations of specifically mobile expertise, and is widely expected to lose out to ARM in the first push for the mobile internet market. A recent report from Strategy Analytics came to this conclusion, and expects Intel and x86 to make little headway in the sector until its Moorestown product, due late next year, gets fully established. The report forecasts that the MID market will take off gradually because of high prices but will achieve annual sales of 69 million units by 2014, when it will be worth over $17 billion. This year, the unit shipments will be about 1 million, with ARM-based devices dominant. "Intel's Moorestown system-on-a-chip will be critical to the company's MID strategy," said Peter King, director of the analyst firm's connected home unit. "But until this arrives in 2009 or 2010, ARM-based vendors will use this window of opportunity to establish market leadership positions." King suggests that the proven advantages of the ARM ecosystem in mobile devices will eventually outweigh those of the Intel platform and that ARM devices will comprise the majority of MID sales though 2014.

The triumph of ARM does not necessarily bode well for TI, of course, since most of its arch-rivals also use the architecture, and Qualcomm, in particular, has made great progress in adding value to the basic platform to optimize it for advanced mobile broadband (Qualcomm is one of the few chipmakers to have an architectural license for the ARM processors, which means it can make changes. Intel also had one of these, inherited when it bought Digital Equipment's StrongARM unit, which was then incorporated into XScale). TI has its own challenges in its cell phone and DSP heartlands, and last year lost its cell phone market lead to Qualcomm for the first time. It is increasingly looking to the analog and embedded markets to sustain growth, rather than its core businesses. In five years' time, the company said recently, it expected the analog and embedded categories to bring in 75 percent of its revenue. Embedded chips now bring in 10 percent of TI's revenue, and analog chips account for 40 percent. "Growing at 20 percent a year over the next five years, you have an analog business that's 60 percent of TI," CEO Rich Templeton said during a recent analyst meeting. TI has 13 percent of the global analog market, making it the leader, but with room to grow. He said embedded chips, including those for wireless infrastructure, would account for 15 percent of revenue in five years at the current growth rate of about 22 percent.