Chipmaking giant Intel has been saying for years (and years … and years) that it will be a major force in the smartphone and tablet markets, but things haven’t worked out that way. Although the power and sophistication of Intel’s CPU’s are largely unmatched, the mobile world is almost entirely driven by processors based on power-efficient designs from the UK’s ARM Holdings.
However, now that Intel says it’s finally ready to compete with ARM processors in mobile devices, it seems to be eschewing the high-margin, mobile-savvy markets like North America and Europe in favor of emerging economies like India and China. Is Intel running scared, concerned that smartphones based on Intel chips don’t yet stack up to ARM-based devices? Is it focusing on markets where consumers are perhaps less likely to notice the difference? Or is it a case of Intel following the Gretzky rule of business: skating to where the puck is going to be, rather than where it has been?
With great power comes great power consumption … and great price
Intel’s dominance of the PC processor market has been built on continuously pushing forward both processor design and manufacturing processes – the famous “tick-tock” iteration of CPUs that has driven the company for the better part of the last two decades. Intel has managed to stay steadily ahead of the competition. It’s current Ivy Bridge processor line is built using a 22 nanometer process, and the next “tick” due in 2014 should shrink processors down to an astonishingly small 14nm process. Smaller parts mean being able to put more transistors and cores and busses into chips and operate them with less power. Progress.
However, all that investment in design and manufacturing technology comes with high costs: chip fabrication facilites cost billions apiece, and that means Intel’s latest-and-greatest chips are expensive. As king of the CPU mountain, Intel charges up to $1,000 for its top-of-the-line PC chips, and up to double that for server chips. In contrast, a couple months ago, IHS iSuppli estimated Apple pays a measly $17.50 for its custom high-performance A6 processors in the iPhone 5. Quite a price gap.
Intel’s challenge, then, is to find a way to make its processors competitive with ARM designs in terms of both power consumption and price. Intel could reduce cost and power consumption by making its chips simpler, but that would quickly make them incompatible with existing x86 software and toss aside (literally) decades of development work on the desktop, in servers, and in embedded systems. Intel isn’t going to do that and tick off its long-time partners. So its mobile chips are at many levels as complex as its PC chips, and they’re reducing their power consumption by migrating techniques from high-end processor designs. This approach dovetails nicely with Intel’s Atom line, originally intended as low-power, low-cost processors aimed at netbooks and set-top boxes. For mobile, Intel shifted these efforts to back to the larger, most standard 32nm manufacturing process because, after all, Intel still has those multi-billion-dollar 32nm facilities. High-end ARM-based designs like Apple’s A6 chip are also built using a 32nm process, but a key differentiator here might be experience: Samsung (which makes Apple’s processors) is pretty new to 32nm manufacturing, where Intel has been at it for roughly four years.
Medfield and Clover Trail
Intel’s mobile offerings to date are the “Medfield” platform aimed at smartphones and the “Clover Trail” platform aimed at tablets. Intel promised it would be able to compete with ARM-based designs in both processing power and battery consumption, as well as deliver processors that were fully compatible with x86 technology. That’s why Microsoft’s Surface Pro devices – which can run traditional Windows desktop software – are being built on the Clover Trail platform.
The Medfield line has two other chips on the horizon: the high-end Atom Z2580 and the low-end Z2000. The Z2850 will increase raw clock speed to 2GHz without increasing battery consumption, but (perhaps most importantly for high-end phones) will add 4G LTE capability: anyone wanting to build a smartphone on the Z2850 won’t need to invest in separate LTE hardware. The Z2000 will be for entry-level devices: no LTE, and a clockspeed around 1GHz.
Clover Trail: Intel’s Clover Trail chips are right on the horizon: they’ve missed the Windows 8 launch and the end-of-year holiday buying season, a fact that can’t make Microsoft and partners looking to launch Windows 8 tablets very happy. Unlike Medfield, they’re dual-core chips, and they also feature new low-power states when idle or asleep that should give devices longer battery life and standby times. We’ll see if those work out: reports of the Surface Pro’s battery life are not promising.
What’s next? For tablets, Intel’s next platform is dubbed “Bay View,” with the first processors codenamed Valley View. They’ll feature anywhere from one to four processor cores based on a new architectural design running at up to 2.4GHz. They’ll also sport four new Intel Gen7 graphics engines derived from its mainstream Ivy Bridge processor line – that’s one more graphics core than Apple’s got in its current ARM-based A6X. Valley View will also be built using Intel’s 22nm process currently used for Ivy Bridge. Remember, smaller always means less power consumption. However, these chips are probably a year away – and Apple and other ARM tablet makers won’t be sitting on their hands in the meantime.
Can Intel earn money on mobile chips?
If Microsoft’s Surface Pro is any indicator, Intel may not have fully resolved power consumption issues in Clover Trail to make tablets competitive with ARM chips, but those single-core Medfield processors do seem to be able to give ARM designs a run for their money in terms of performance and power consumption. The question then becomes “how much money?”
ARM-based chips are cheap, with most ARM-based chips from Qualcomm and Samsung priced between $15 and $30 each. Intel has shipped some inexpensive CPUs in its day, but so far as I can determine the cost of ARM processors is below anything Intel has shipped in recent years. It’s clear that Intel’s margins on mobile chips will be considerably smaller than on its mainline PC and server chips, and (therefore) Intel will have to sell a lot of mobile chips if it wants to make up the revenue it’s losing from the shrinking PC market.
However, Intel might be able to persuade some device makers to get on board. Folks who want to use ARM processors not only have to pay companies like Qualcomm, Samsung, or Taiwan Semiconductor to make them (Texas Instruments has bowed out of the ARM business), but also have to pay a license fee to ARM for the designs. That gives Intel a little wiggle room on price, since Intel has full rights to the design and owns its own chip fabrication facilities. Where using ARM chips is complicated, Intel can offer one-stop shopping.
Even so, Intel is going to make a lot less money on every mobile chip it sells compared to PC and server processors. To make ends meet, Intel has to bet not only on the extraordinary growth of the mobile market (mobile devices outshipped PCs back in 2010, and the installed base of mobile devices should eclipse PCs next year) but also on consumers replacing their mobile devices more often than they replace PCs. That could work out: where consumers may replace PCs only ever three to five years, they are likely to replace smartphones and tablets more frequently – and they’re also likely to own more than one mobile device. In theory, Intel could be in a situation where it can sell a typical consumer three or four mobile processors in the same period of time it might sell them just one PC processor.
So why are Intel’s smartphones overseas?
If Intel is so keen to get into the mobile devices – and so much of the company’s future seems to ride on mobile – then why are Intel-powered smartphones mostly launching in emerging overseas markets? If Intel’s Medfield chips really do compare favorably with ARM-based phones, why aren’t smartphone makers looking to launch devices in the most lucrative smartphone markets like Europe and North America, where consumers have proven they will camp out overnight (in the cold and rain!) to be among the first to own a hot new phone? Are Intel and phone makers embarrassed about Intel-powered phones? Are they worried the devices won’t stand up to the scrutiny of savvy consumers, so they’re launching the devices in markets were customers perhaps aren’t as experienced or discerning?
The answer comes down to who Intel has been able to bring on board as partners … and LTE.
Partners: Intel’s approach to smartphones has not been to build and market its own phones, anymore than the company has built and marketed its own PCs. Instead, it has been creating reference smartphone designs running Android that it can pass along to phone manufacturers, who in turn customize, augment, and tweak as they like rather than market under their own brands.
So far, Intel has been able to bring a number of partners on board, but they’ve mostly been manufacturers willing to run with phones very close to Intel’s reference designs. Major partners include Lenovo (now the world’s top PC manufacturer) and ZTE in China, Lava in India (Intel just announced two more smartphone partners in India – where mobile Internet use has already outpaced traditional PCs), and Russia’a Megafon.
Intel has not been eschewing developed markets entirely: Motorola (now a unit of Google) has launched the Razr i in Europe and Latin America as well as a Medfield-based smartphone through China Mobile, and Orange UK launched the San Diego, a barely-touched version of Intel’s reference phone design.
LTE: What you’ll notice all these Intel-powered smartphones lack is LTE. While LTE is the immediate future of most mobile broadband efforts around the world, right now the United States is in the ironic position of having the broadest LTE deployment on the planet. That means U.S. carriers really aren’t all that interested in new phones without LTE (and highlights why the Google Nexus 4 disappointed so many), and many major phone manufacturers aren’t yet interested in making Intel smartphones for the North American market. Without LTE, it’s a tough sell. The next high-end chip in the Medfield line, the Atom Z2580, will have built-in LTE capability. It should be ready in the first half of 2013, and should enable phone makers to start targeting the American smartphone market seriously … if they choose.
Where will the puck be?
It will be interesting to watch how aggressively device makers target the U.S. market with Intel-powered smartphones. The United States may represent one of the most lucrative smartphone markets on the planet, but it’s also one of the most mature and it’s still dominated by the Apple iPhone. It takes more than a processor to compete successfully with Apple, which has proven the value of emphasizing design and usability. That means, in the United States market, Intel-powered smartphones are likely to come across as just-another-Android device: they might be better or worse than others, favored by some and disliked by others, but will almost certainly land in an already-crowded field that already has some standout players (like Samsung’s Galaxy line).
Overseas, however, Intel might have a better chance of establishing its devices as high-end or aspirational brands – in part because iPhones are often out of reach of many consumers in developing markets, but also because the Android offerings are often less complex and emerging middle class consumers are only now beginning a serious migration away from feature phones.
There are lots of potential customers in emerging markets: right now, only about 11 percent of India’s population has any sort of mobile device: that represents a potential market of hundreds of millions of people. And that’s still small time compared to China: China Mobile alone has more than 700 million customers: that’s more than double the total population of the United States. In the long term, establishing a firm position in those markets might be worth more to Intel than fighting for a share of the United States’ smartphone market.