Qualcomm and Intel go head-to-head in mobile internet devices
Killer phones or kill-a-phone?
The convergence of PC and cellphone architectures has brought Intel and Qualcomm increasingly into conflict, and at the Consumer Electronics Show in Las Vegas last week, both showed prototypes of their platforms for next generation mobile devices.
Both silicon giants, along with device makers like Nokia, see the current notebooks and smartphones evolving into what Intel calls an ultramobile PC (UMPC), Nokia a multimedia computer or internet tablet, and Qualcomm a pocketable computer.
And both need desperately to take pole position in chips for these devices as they emerge from this year, as by the time UMPCs achieve mass market status – probably from 2011 – their traditional revenue streams will be stagnating.
Qualcomm’s COO Sanjay Jha showed off a prototype ‘pocketable computer’, saying “if you can't carry it in your pocket, you can't carry it with you”.
Predictably enough, the device has a slide-out qwerty keyboard and supports HSPA, GPS, Wi-Fi and Bluetooth. It has an 800 pixel screen designed to support viewing of regular, rather than specifically mobile, websites.
The initial operating system is Windows Mobile – not too surprising, given Qualcomm’s low profile but long term closeness to Microsoft, and the fact the CDMA giant has not really shown its hand yet on Mobile Linux, unlike Intel and Nokia, which are increasingly including the open source OS in their strategies. Also, with the immature and fragmented state of Mobile Linux, it is currently easier to demonstrate an impressive UMPC platform using Windows Mobile.
The key chip for these devices will be Snapdragon, arguably Qualcomm’s most critical current product, as it will underpin many of the company’s important moves to diversify its business into areas such as UMPCs and consumer electronics. Indeed, Jha fuelled the rumor mill by naming Amazon’s Kindle e-book reader as an example of a non-phone device that could use Snapdragon.
The pocketable PC prototype shown in Las Vegas, codenamed Anchorage, sported a 1GHz processor claiming peak power of 0.5W, talking up Qualcomm’s claims that Snapdragon will steal a march on rivals in the all-important power/performance ratio, an area where Intel is also making bold claims for its forthcoming 'Menlow' architecture.
Samsung and HTC are the first publicly announced OEMs for the design underpinning Anchorage, and expect commercial devices late this year, aiming to leap in before companies like Apple, which is rumored to be readying a UMPC-style product for late 2008.
Jha said that, in the medium term, he would target volumes of about 200m for Anchorage or its successors – about the total current base of high end smartphones, as compared to 1.3bn cellphones overall.
Contest with Intel
Neither Qualcomm nor Intel was shy about recognizing the other as the key enemy. Despite the distance opening up between Texas Instruments and Nokia, it remains highly likely that the Finnish giant will rely heavily on its old familiar for future mobile PC developments, and unless this policy changes radically – potentially because Nokia and Intel come to a détente that, arguably, both need to wrongfoot Qualcomm – TI’s own R&D program will remain strongly influenced by Nokia.
Jha and Intel CEO Paul Otellini made similar points in setting out their respective competitive positions, emphasizing the benefits of their particular heritages. Jha, generally the friendly face of Qualcomm, was diplomatic. "We come at this problem from an understanding of wireless," he said. "Intel comes at it from an understanding of computing. We both bring different things to the table. Time will tell how our vision works versus Intel's direction."
He even praised Nokia’s major device in this category, the N800, saying (as Nokia is reticent to do) that it had outsold expectations.
Otellini made the opposite point: “It's a lot easier to add communications to a small computer than add computing to a small phone,'' he said in his keynote, where he introduced Menlow. “It's a major element in our growth strategy.”
While nobody could doubt Qualcomm’s mobile credentials, Otellini had the difficult job of instilling confidence in Intel’s, given that its last $5bn adventure in the cellphone market, in the form of its XScale processor family, ended in disappointing sales and the offloading of the unit to Marvell in June 2006 for $600m.
Intel re-enters mobile
Otellini was keen to imply that XScale had been targeting the cellphone in its current form, which is an area where Intel had no experience, but that Menlow is focused on the future form of the mobile device, and this is one that Intel aims to define in its own image, playing to its PC strengths. "What we are focused on now is where we think phones are going, not where they are today," Otellini said in an interview at CES. "In the past, we aimed at building yet another chipset for phones."
Five mobile chips, codenamed 'Silverthorne', which will underpin the Menlow mobile internet platform as well as future notebooks, were unveiled. All were built using the company’s latest 45nm process, for improved power/performance ratio. Commercial devices based on Menlow are expected late this year, and Apple is rumored as an early adopter, while Lenovo and Toshiba were among the supporters at CES.
Intel now has three low-power mobile internet device reference platforms - 'McCaslin', Menlow and 'Moorestown'. Menlow consists of a Silverthorne processor, a support chip called 'Poulsbo' for I/O and graphics, and a communications module that can be support Wi-Fi or WiMax (or, in future, other systems). Moorestown combines the functionality of at least the first two chips into one for even lower power consumption.
The big question is when (or if) Intel will incorporate 3G into its devices. Currently, it is attempting the ambitious feat of trying to claim the high end mobile market entirely for Wi-Fi and WiMax, technologies over which it has a large measure of control, sidelining 3G (and Qualcomm and TI) into lower end voice-oriented phone markets.
This is almost certain to prove a failure, since even Intel must recognize that next generation mobility will be about multiple networks. The real future must lie in chipsets that intelligently move between these, and both Qualcomm and Intel itself have important software defined radio developments. Intel’s dilemma is whether to play directly in 3G, where it has failed before, or to win strong market share for Wi-Fi/WiMax devices and then focus on dominating the whole range of networks once SDR and new networks like LTE evolve.
This more gradual approach has strong commercial logic, but is high risk in terms of ceding market share at a critical point in the mobile market’s evolution to TI and Qualcomm. This remains a tricky issue, and Intel’s last effort to bridge the divide between cellular and all- IP networks, its venture with Nokia to put HSPA in notebooks, collapsed under the weight of its conflicts of interest.
Otellini has a tough job this year, with many of Intel’s recent growth drivers, notably Centrino, facing maturity, slower growth rates and falling margins. Analysts estimate Intel's revenue growth will slow to 6.1 per cent in 2009, according to Bloomberg, down from an average of 13 per cent between 2003 and 2006, the heyday of Centrino wireless laptops. As well as UMPCs, WiMax is of course a high profile, and high risk, element in the growth building strategy.
Even this is a field where Intel may soon have to contend with Qualcomm. Although the latter has been fierce in its opposition to WiMax in the mobile market, if it sees 802.16e gaining momentum it will undoubtedly want to play, as it reluctantly did in Wi-Fi, despite the technologies being at odds with its own roadmaps and IPR practices.
Jha said, in one of the most definite statements yet that Qualcomm has not closed the door on WiMax: "We're working on WiMax as well but our priority is clearly 3G.” Meanwhile, Intel will support Mobile WiMax in its upcoming Montevina notebook platform, due next quarter, which is the latest iteration of Centrino.
It is also likely to deliver WiMax to UMPCs during 2008 via Menlow. Intel’s WiMax radio is called Baxter Peak and will be used by Nokia in its initial WiMax products, due later this year. Montevina can optionally use 'Echo Peak', a mini-PCI card that integrates WiMax and Wi-Fi on one chip.
Copyright © 2008, Wireless Watch
Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.