Intel is showing off the technology choices that it hopes will make the next generation of mobile devices the heirs of the PC architecture, not the cellphone – notably its latest ultra-mobile PC platform and its WiMAX Centrino roadmap. Creating the mobile future around PC roots is an ambition that is central to the chip giant’s growth strategy, since if ubiquitous wireless is mainly driven by devices based on PC-style designs, this will be a natural market for Intel chips to dominate. But with powerful interests determined to keep Intel out of dominating future wireless, is the chip giant relying too heavily on its old allies in the PC industry, or does it badly need to get new friends, particularly Nokia, on board?
The success of Wi-Fi drove a generation of laptop replacement, and faster standards will drive another, but this is not enough for a company with the scale of Intel. WiMAX needs to do a lot more than just encourage the affluent of the world to upgrade their notebooks and purchase one with Intel chips inside. To be the key engine for ubiquitous wireless it needs to go into a huge range of devices for use in every conceivable application, territory and income group. Most of these will look nothing like the notebook or even its currently emerging offspring, the Ultra Mobile PC (UMPC). These will be important devices and Intel, which demonstrated its latest ultra-mobile platform at its developer forum in Beijng last week, is packing them with features that should enable it to trump the more conventional smartphone designs in many markets – adopting Linux instead of Windows to support new interfaces and appeal to Asian buyers; choosing the Siano chipset for mobile television; promising Wi-Fi and WiMAX for mobile broadband (but not a 3G option).
But Intel’s aim of making the UMPC and its successors the default client device for the new generation of mobility carries huge risk, since it is up against powerful challengers with their own need to control the new mobile platform and their own approaches to the architecture – approaches which would currently favor non-Intel chips, for technical or political reasons. Most important of these challengers is, of course, Nokia, with its ally Texas Instruments in tow. Yet, with Intel putting distance between itself and its familiar, Nokia’s enemy Microsoft, and with a shared interest in sidelining Qualcomm, the two companies have more common interests than ever before.
There have been moments when an Intel-Nokia axis appeared to be on the cards, particularly when the two giants joined forces to lend support to the then-obscure 802.16 standard in 2003, and when they announced a deal to create HSPA laptop cards. But both displays of harmony, and others besides, have foundered on the basic conflicts of interest between these two, derived from their very different heritages, and their need to drive the next mobile generation, not just ride the wave. After a show of interest in WiMAX, Nokia retreated to the sidelines until it was clear that the technology would become a market force, and so lost some of the initiative in creating the 802.16 challenge, despite its success in the Sprint Nextel deal. Its fit of conservatism, clinging to the Ericsson line that HSPA and LTE must be the basis of 4G, also cost it the opportunity to create a new balance of power by allying with Intel. Then Nokia broke out of its 3G straitjacket and embraced a strategy that relies on supporting multiple architectures and broadening its customer base beyond the major cellcos to the enterprise and emerging consumer groups with very different expectations of the mobile device. This has led to innovations such as the non-cellular Wi-Fi/Linux Internet Tablet and the renewed enthusiasm for WiMAX, showing Nokia accepting the open IP world and positioning to profit from it.
Moving from the tightly controlled, mobile-specific architecture and commercial structures of the cellphone, to the resource-hungry, uncontrolled platform of IP and off-the-shelf components, has been a challenge for Nokia, and one it has not yet fully addressed. But it has made impressive progress in extending its roots and embracing new concepts that are more natural to the PC industry – indeed, far more impressive than the efforts of the PC makers themselves to support mobility. Dell may have hired former Motorola handset chief Ron Garriques, sparking expectations of an assault on the mobile market, but along with the other PC giants, the company has so far done nothing more interesting than shrinking existing notebook designs and adding Linux options to the standard Wintel.
Intel, too, is falling victim to a blind clinging to the old patterns. In the earlier years of the decade, then-CTO Pat Gelsinger articulated one of the most inventive and convincing visions of the mobile future and how Intel might drive that, but in the past year, the company has been far too conservative. It has done a superb job of mobilizing its marketing machine to make WiMAX an industry force, but has yet to show real signs of profiting from that effort itself to any great degree. Unlike Nokia and even bugbear Qualcomm, Intel has not yet resigned itself to the fact that the road to 4G will not be single-technology. The simplicity of the Wintel=PC/xCDMA=3G equation will never return, and that makes the new mobile world a commercial and political minefield, where successful companies will have to support many networks, including those over which they have limited influence.
Nokia and Motorola have recognized this, taking Texas Instruments with them; in its purchase of TeleCIS we believe Qualcomm has too (a eal it cut this week). Intel has been much praised in the past week for distancing itself from Wintel by making a full-fledged commitment to Linux for the UMPC, but this is not radical, it is overdue. The advantages of the cosy Microsoft alliance have made Intel too cautious about Linux for too long, allowing Nokia to push its Internet Tablet (effectively a UMPC under another name) into the market ahead of any major Intel-based equivalents. Now it looks defensive in its support for the open operating system, and is following a general shift of the PC industry towards multiple OS support, not driving it.
Far more serious in the mobile world is the determination, now it has spent so much energy and money on making WiMAX into a real force, to try to ensure that 802.16 is the lead network in the road to 4G. There may be high level talks of rapprochement with LTE and converging standards, but that is mainly geared to achieving a level playing field for WiMAX, by making it one of the official 3GPP and 4G options alongside UMTS/LTE, CDMA EV-DO iterations, TD-SCDMA and so on. Intel believes it needs to sideline the LTE platform and community, because it seems to have given up hope of being a leading chip vendor in this market. Hence the breakdown of its deal with Nokia over HSPA laptop cards – regardless of what consumers may want, Intel could not see embedded HSPA in the devices it really does control as anything but a threat to the vision of universal notebook WiMAX/Wi-Fi. And by selling off its cellphone architecture and chips, XScale, to Marvell last year, Intel made a clear statement that it had given up hope of being a major cellphone chip supplier.
This is likely to prove to be a major miscalculation, driven by shareholder panic at some poor quarters and a failure of the confidence needed to see its 3G-plus efforts through. Keeping XScale would not have made Intel a 3G giant – Texas Instruments, Qualcomm and the others would see to that. But it would have given Intel the R&D platform and market experience to hone its mobile efforts and create a unified system in the next generation that could support multiple radios and apply WiMAX to a mobile-optimized design, something Centrino and other notebook-oriented silicon cannot claim to be. Such a development program would have given Intel a chance in the broader mobile device market, beyond notebooks and smartphones, particularly with the potential to create chips for new applications, especially in mobile consumer electronics, where the playing field remains quite open and Texas Instruments has not yet asserted dominance.
A fully mobilized, multi-radio architecture – even one that favored WiMAX – would have taken several generations to fine tune to the extent that it could challenge the mobile chip specialists, but Intel has the resources, expertise and determination to have achieved that in time for the pre-4G generation. And it would have made it far easier to work with Nokia, by offering the Finnish behemoth a new and differentiated chip option that also opened some useful doors in the enterprise device market that Nokia yearns to lead.
Such an alliance would hurt Microsoft and Dell, but with their relatively poor progress in the mobile world so far, that might not be of so much concern to Intel by 2008 and beyond. It would also hurt Texas Instruments, which would certainly be to Intel’s gain, and indirectly, Qualcomm. Nokia, again by contrast, is casting off some of its old loyalties, adopting a more multi-faceted chip sourcing policy and reducing its tight symbiosis with TI, but there seems little hope for Intel that it will be part of the new-look Finnish ecosystem. And with Nokia determined to use its handset power to take a lead in WiMAX devices as these develop, this could not only spoil any remaining chances Intel has in the LTE space, but could harm it even in its own WiMAX heartland.
For instance, the future roadmap for its successful Internet Tablet – a Linux/Wi-Fi device that will incorporate WiMAX at a relatively early stage for Sprint – looks very similar to the UMPC but steers entirely clear of the Wintel axis and of Centrino. This will provide a powerful threat to Intel-based devices in classic UMPC target markets such as the enterprise. But Nokia also has a multitude of other mobile assets, from smartphones to music/video players to R&D projects focused on super-small dongles for low income bases. Not all these will succeed as individual products, but their concepts can be incorporated into a flexible and broad ranging set of future mobile devices, many of which will not be naturally adapted to Intel silicon now that the giant has pulled in its mobile horns to focus on Centrino and its spin-offs alone. Even in the short term, when most pre-4G efforts will focus for the first generation on notebooks, Intel is not assured of making WiMAX win. Sprint Nextel selected Nokia as one of its suppliers partly on the basis of the Internet Tablet, yet that has no Intel chip inside.
Intel HAS to succeed in mobile devices – smartphones, media players, low cost widgets for low income communities and so on – or it will become stuck in laptop replacements, a growing market but not large enough to carry the future of such a giant. Adopting WiMAX and Linux are good tactics, but remain heavily focused on the traditional customer base, some of which, Dell included, are struggling. A company of Intel’s scale, to be credible in the new mobile generation, needs Nokia as a customer and an ally. Far more radical moves than putting Linux on the UMPC will be needed to create that potentially world-changing axis.
The new Intel products:
At the Developer Forum in Beijing last week, Intel showed off the next generation of Centrino, codenamed Montevina, which supports WiMAX. It will follow the imminent refresh, Santa Rosa, some time in 2008, and will include a new 45 nanometer dual-core processor called Penryn; a chipset called Cantiga that will support a 1,067MHz bus and 800MHz DDR3 memory; and a WiMAX card, codenamed Shiloh.
The company is currently gearing up to ship Santa Rosa, the upcoming Centrino Pro update, which was once supposed to support HSPA as a result of the then-aborted Nokia deal.
Santa Rosa includes a 65 nanometer, Core 2 Duo processor called Merom; a new chipset with an 800MHz front-side bus (Crestline); and an 802.11n wireless card (Kedron). The refresh is expected to ship next month. Santa Rosa also supports Intel Turbo Memory (codenamed Robson), a flash memory that stores sufficient data so the CPU does not need to use the hard drive to boot up or access many applications. Intel came out and categorically said Santa Rosa and Montevina, the next refresh, would not support HSPA or successors, just WiMAX and Wi-Fi.
Intel also debuted its latest – and first really commercial – UMPC architecture, McCaslin, which is based on one or two of its Steeley processors, running at 600MHz or 800MHz and with integrated graphics processors, and designed specifically for this class of device. Initial UMPCs will run Windows Vista but Linux versions will appear in 2008, when Intel will announce McCaslin’s successor, Menlow, based on a 45nm low power processor called Silverthorne and a new chipset, called Poulsbo – this will reduce the current two-chip chipset of McCaslin to a single-chip design more suited to small devices.
Internet Tablet or UMPC?
Nokia’s Internet Tablet is conceptually the same as the Intel UMPC, though it has got to market more quickly and taken an earlier and braver stand on Linux. Intel itself predicts a global market of 180 million units a year by 2010 for mobile internet devices of this kind. Its new UMPC, like Nokia’s product, looks slimmer and more truly mobile than first generation attempts such as its initial UMPC, launched last year, or Microsoft’s Origami (though no doubt Microsoft is gearing up to update that design and make a serious attempt on the UMPC space). The new generation of mobile internet devices typically have smaller screens than the originals, simplified Linux-based user interfaces rather than Windows, and ‘instant-on’ performance.
Intel’s new UMPC, in its Linux form, will run Red Flag ‘MIDinux’, the result of a joint project that will yield results this May. MIDinux will use a full Mozilla browser and, like Nokia with the tablet, has largely excluded embedded frameworks, with the exception of J2ME. Instead it uses compact versions of desktop software packages, and Red Flag is aiming to use the Intel muscle to create a major development community around optimizing applications for MIDinux. Open source packages currently supported in the OS distribution include Mozilla browser core with Flash, RealPlayer, PDF, and Java plugins; Gnome VFS (virtual file system); BlueZ Bluetooth stack; Telepathy for instant messaging and presence; DBUS for inter-process communication; Busybox and J2ME Java. Like Nokia, Intel/Red Flag will also use the Matchbox window manager along with a proprietary Hildon user interface and application framework.
Intel is also working on a ‘master user interface’ to live on top of this – which of course is where the real challenge to Nokia and to Microsoft could come, since the winner in the new device war will be the company that can finally make the mobile internet useful and navigable. Nokia has been expending huge efforts on this, as have Microsoft, Qualcomm, Apple and others, in the bid to create a user interface that is actually suitable for the mobile internet. It is too early to tell whether Intel has a chance of leading this effort, but it will be a critical success factor.