We discussed the company's plans for a Xeon refresh earlier this week, so today we'll primarily discuss what's coming in 2010 for desktops and the mobile (Atom) segments. In desktops, Intel expects Clarkdale to steadily push into the mass market at nearly all of the price points currently occupied by the aging line of Core 2 Duo processors (Core 2 Quad chips having already been mostly pre-empted by the Core i5 launch last year). Intel stated that it intends to ramp 32nm production more quickly than it ramped at 45nm, but that the speed of the shift will depend largely on consumer demand. 2009 was unusual in that it was consumers, not corporations who drove IT spending. The former is historically more fickle than the latter; IT companies of all sorts will be watching trends and fashion more closely in 2010 than they might otherwise. One final note on the 32nm shift before we turn to Atom. As process nodes have grown smaller, the cost and difficulty of moving to each successive generation has grown significantly.
As a result, a number of companies have either pushed back their leading-edge process deployments or, like TSMC, have struggled to meet the needs of their customers. Intel's 32nm ramp, in contrast, appears to be going quite well; sales of 32nm products had a 'significant' effect on ASPs (which in turn helped drive the company's record gross margin).
Intel's Paul Otellini demonstrating the Atom-powered LG GW900.
Atom's popularity shows no sign of waning, but the point of sale for the diminutive systems may be shifting towards cell phone companies or other carriers, with carrier-based models of subscription and payment. The number of total netbooks distributed by carriers as opposed to Dell or Asus grew markedly in Q4 2009; Intel expects this to continue to grow throughout the ear. Atom accounted for $1.4 billion in total sales in 2009; sales the CPU manufacturer believes are completely additive to its core businesses. Mobile sales, including low-end notebooks, have not been negatively impacted by Atom's popularity.
While there's no reason to doubt Intel's candor on the impact of Atom on its core mobile business, the company's statements are a tad disingenuous. While the new Pine Trail platform should improve battery life, the company's new LM10 Express chipset continues to lack features like digital video output or a video core with built-in support for hardware-accelerated video decoding. (Pine Trail allows for the use of third-party decoding chips, but Intel elected not to provide such support itself). This isn't to knock Pine Trail as such—if the new SoC meets your needs and improves battery life, that's great—but it's impossible to compare the feature sets of the new NM10 Express and NVIDIA's ION without realizing that Intel is keeping certain features away from Atom deliberately.
All things considered, 2010 looks like it'll be a pretty quiet year for Intel, at least as far as competition is concerned. AMD will most likely compete against Clarkdale's CPU+GPU combination with cheap Athlon II quad-cores and its own excellent integrated GPUs, but Sunnyvale won't be in a position to even attempt to challenge Intel at the high end of the market until the Bulldozer and Bobcat architectures debut in 2011. Atom will continue to dominate the netbook space, with AMD and VIA possibly nibbling on the edges, and by the end of the year we'll supposedly see at least a few mobile phones based on 45nm Moorestown, though Intel probably won't make a concerted play for the space until the 32nm Medfield platform is available sometime in 2011.