Intel will try to prove its claims that it can be a powerful manufacturer of processors for smartphones and tablets when its Medfield platform emerges in commercial products in H2 this year. In the same time frame, ARM vendors will release their first notebooks that challenge, conceivably, Intel’s most important and profitable business today. Both Intel and ARM are staging aggressive launches and prepare for a fight that will be much more bloody than the historic processor battles between AMD and Intel. Does Intel have what it takes to dent ARM’s segmentation-driven application processor market? Can ARM deliver processors that are compelling enough to face Intel’s prestigious and performance-driven CPUs?
We have been waiting for a confrontation between ARM and Intel since 2007, when Intel announced Silverthorne, the original core of the Atom processor that never lived up to Intel’s hopes to create a multi-billion-dollar market of Mobile Internet Devices (MIDs) that was hoped to revive the vision of the failed UMPC in 2006. Intel as far too optimistic about Silverthorne’s appeal, a processor that was produced for less than $6 per piece and sold for up to $60. It was too expensive and after initial consideration, vendors decided that Intel’s pitch that low power consumption deserved a price premium was unrealistic and went the Diamondville route, a variant of Silverthorne that consumed more power and ended up in almost all netbooks that were sold in 2008 and most of 2009.
When Atom (Silverthorne) was announced in early 2008, Intel explained that the x86 compatibility was the big advantage of the processor and that big x86 applications could now be run on ultra mobile devices. Strangely enough, Intel shot itself in the foot and limited the success of Atom to PC as the company had a super-small and very capable 45 nm CPU, but it chose to combine in with a 130 nm chipset that was too large and too power-hungry for phones. Back then, Intel projected that the successor of the first Atom, Moorestown, would be released by early 2010 and would be small enough for smartphones. In 2009, an Intel executive told me that Intel was “confident” that the company could eventually win the processor business for Apple’s iPhone.
For almost two years, Atom helped Intel to sell a massive number of processors especially into the netbook market, but it largely failed in other areas, including consumer electronics such as smart TVs, which Intel has entirely given up at this time. More than four years after the release of Silverthorne, Intel says it now has Medfield, its first significant smartphone offering following the early 2000s XScale era (which was especially defined by Compaq iPaq handheld devices), ready to compete and proof that x86 has a place in ultra-compact devices as well. By definition, Intel will have to succeed in this market over the long run as its business model is strongly based on selling more processors every year and volume markets that neighbor its traditional PC business will always remain a conclusive field for expansion for the company.
The announcement of Medfield at CES earlier this month showed confidence, but lacked details beyond clock speed and battery performance. Instead, there has been a constant Facebook and Twitter noise level that accompanied the announcement and could be best interpreted as an incredible amount of saber rattling of ARM and Intel representatives. Even if the company remains rather secretive in public, Intel appears to be following a conviction, on a social network level, that the mobile processor is its opportunity to lose because of its performance advantages. ARM, on the other, does its best to tell the press that Intel is not even worth the consideration of a competitive threat as ARM’s omnipresent market segmentation is a wall Intel’ won’t be able to climb.
Market volume
Roughly 464 million smartphones were sold in 2011, according to Digitimes Research, all of which integrated ARM-based processors. PC shipments were 353 million units, according to Gartner. About 190 million of those PCs were notebooks and about 90%, or 171 million, of those were notebooks that sold below $1000 retail price. Intel is seeing its greatest growth of units and profits in this market, but will be facing ARM as a direct rival once Windows 8 for ARM (W8A) is introduced and netbooks could see a renaissance. So, in a generalized view, Intel will be going after a 464 million unit market and compete with ARM and its army of vendors on the promise that large volumes will deliver big profits. At the same time, ARM vendors will challenge Intel in one of its most critical markets on a bet that there will be smaller volumes, but higher profit margins.
The success and failure of Intel will come down to the question whether performance matters to you or not. The sustained success and failure of ARM in smartphones will come down if platform availability has a higher value to vendors than they believe performance it has to consumers. If performance in a smartphone matters and enables unique features and runs highly desirable applications ARM processors cannot run, there will be enough incentive for smartphone makers to build Intel-based phones. If Intel cannot build the perception that Medfield can deliver a unique and absolutely wanted feature, it will have a tough time to push its processor into the market.
Performance versus segmentation, perception and commoditization
If we follow the argumentative chain of both sides, the obvious observation is that Intel ignores ARM’s market position and ARM could be ignoring Intel’s engineering and manufacturing strength. From a public view, ARM’s stance is very reminiscent of AMD’s decision to simply downplay the threat of Core 2 in 2005 and 2006, which strangled the company and resulted in a breakup and has turned AMD into a fabless company. It is dangerous for any Intel competitor to downplay its resources and abilities to come up with killer products. Intel has always been at its best when it was under pressure and there is no reason to believe that this will change this time, even, if Intel shot itself in the foot with virtually every other processor for ultra-compact devices in the past.
Intel very much argues that, ultimately, the performance of the single-core Medfield will blow all ARM rivals out of the water and end ARM’s dominance in the smartphone market. ARM has performance gains on its roadmap as well, but they are much less emphasized than they are at Intel. It almost appears as if ARM simply ignores Intel’s performance claims and states that Intel will not be able to attract many buyers, because the credibility of x86 has, traditionally, been limited to PCs. Phones and tablets are an entirely different game, according to ARM.
In ARM’s opinion, Intel will have to go beyond traditional computing to achieve platformization and enable hardware manufacturers to create a coherent platform that will include not just computing devices, but products such as set-top boxes, TVs, storage devices, etc. If we assume that both Intel and ARM are right, that Intel’s Medfield will be faster than all other ARM processors in phones and if we assume that the platformization of ARM is greater than that of Intel, which product will the consumer choose? Will the consumer, in the end, care whether an Intel or ARM processor is in a phone and will the consumer care whether there is an ARM or Intel processor in a sub-$500 notebook/netbook or tablet?
The answer will largely be answered by the abilities of both parties to influence the perception of consumers as well as the commoditization of either category. In a world of commodities and a commoditized computer and phone, selling technology over a much more general benefit, such as features and support of many user scenarios, is increasingly difficult. In phones, consumers have not, until recently, cared about the processing horsepower in their devices. Samsung, Marvell, Qualcomm and Nvidia are transforming that image by bringing back the importance of clock speed as well as the number of cores. In notebooks, we have learned and relied on the notion that your decision to go with one or the other brand can make or break your computing experience. Intel has built its reputation of building the fastest processors for general computers and it will be tough for Nvidia & Co. to break this perception. On the other hand, we know that Intel has no credibility in mobile phone processors and that Qualcomm and Nvidia processors are the safe choice.
The dilemma of perception
Intel faces an interesting challenge in the smartphone market that could prove to be a huge problem. If performance, in fact, counts, it is surprising that Intel is offering Medfield as a single-core processor. Until 2005, Intel was driving a perception that clock speed was synonymous with performance and a 3 GHz processor was faster than a 2 GHz chip. Since 2003, the company knew that it would hit a wall with clock speeds sooner or later and that the pace of increasing frequencies would be unsustainable. That wall arrived in 2005, when the company had to cancel its 90 nm 4 GHz Pentium 4 CPU. Beginning in 2006, the marketing tone changed and shifted to the number of cores as the best indicator of processor performance. Today, we know that more cores are, in certain and most circumstances, better (and more expensive) than fewer cores. However, today, we are hearing that a single-core Medfield processor is faster than dual-core ARM architectures and faster than upcoming quad-core ARM chips, which is inconsistent with the remainder of the marketing message at best.
Without having seen Intel’s Medfield processor, it is reasonable to assume that Medfield will be, at the very least, competitive, as far as performance is concerned, when it is released. Initial demonstrations also suggest that Intel has applied its talent to running apps that are compiled for ARM much more efficiently on Medfield than they run on ARM. However, the perception, advertising and marketing work against Intel as consumers and sales people who do not read cell phone reviews and investigate benchmark results, will always perceive a dual- or quad-core processor to be faster than a single core processor (and even that assumes that a consumer really cares about processor performance.) Vendors such as Nvidia or Qualcomm will have a significant marketing advantage and credibility that Intel’s marketing strategists will have to overcome. To have only one core in a product that is targeted to kill the competition, is a big miss for Intel, especially when it claims to have the best engineering talent and manufacturing capabilities. Whether two cores are better than one or not, Intel will need at least a dual-core part to compete in the high-end range of smartphones. According to Strategy Analytics, 25% of smartphone processors that’s hipped in Q3 2011 were dual-core units. 57% came from Nvidia and 18% from Qualcomm.
In notebooks, Intel will be facing an onslaught of ARM-based product that will almost exclusively be competing on the basis of cheaper prices initially. The disadvantage Intel has is the novelty effect of ARM that is especially driven by Microsoft and its pitch that it is extending Windows to the world of ARM. Intel’s credibility in computing and its strong brand will be competing with several newcomers that want a share of Intel’s pie and will, inevitably, move upstream if Intel cannot contain their march at the lowest level. Intel will have to do its best to discredit ARM in computers to prevent Nvidia and Qualcomm from becoming perceived household names in traditional computing products. If ARM vendors get their feet on the ground, Intel will have a much bigger problem on its hands than AMD ever was.
Commercialization and platformization
ARM’s claims of platformization have reasonable validity. Smartphones have value that goes beyond the simple feature of making a phone call. We use phones to send data to a game console. We remote control security cameras. We interact with tablets, storage devices and jukeboxes. The technology foundation for the consumer is the software platform, primarily iOS and Android. However, those who build those devices, also consider the hardware platform to deliver the greatest possible compatibility and the greatest possible reduction of development effort. Apple won’t abandon its A5 processor efforts anytime soon. Samsung, which uses ARM architecture in its TVs, is unlikely to consider Intel processors for its phones beyond experimental and alibi products anytime soon. Apple and Samsung shipped a combined 44 million smartphones in Q4 of 2011, and covered nearly 37% of the smartphone market. So far, Intel has only Motorola as a partner, which was an expected move as Motorola is part of Google and Intel has been rather cozy with Google recently.
To make a substantial impact, Intel will have to move beyond the smartphone to compete in platformization against ARM. Success in tablets is a must and Intel will become much more competitive once it expands its brand in products that are computerized, including TVs and in-car entertainment.
The Bottom Line
Any prediction whether ARM or Intel will succeed in smartphones over the long term stands on thin ice as there are too many variables that cannot be reliably calculated at this point and there are plenty of variables we don’t even know yet. What we do know, however, is that both ARM and Intel have significant strength and it is likely that the company that can best react to consumer needs and is innovative enough to drive and change consumer needs while remaining flexible to adopt changes dynamically, will win this game. We also know that Intel is unlikely to win based on performance and manufacturing alone while ARM’s platform value will not be enough over an extended period of time. The winner will be the company that can position itself best between the value proposition between today’s ARM (vendor) and Intel.
You can leave a response, or trackback from your own site.
