Intel had a solid year of shipping 32nm processors from their Westmere family and followed the company’s “tick-tock” strategy to spread the engineering risk out among generations.
Intel has new competition on the low-end from ARM and ARM’s new best friend Microsoft and on the top from GPUs in supercomputers and server clusters. To offset some of that impact the company has moved into the low end low power SoC market and invested in software by buying Wind River and Mcafee.
Here’s what I see for Intel in the near future.
Processors and process. No company ships more processors than Intel – period. But even though Intel ships hundreds of millions of x86 processor into the PC and affiliated markets, it sees the billion mobile phones being shipped each year and it looks like green fields to them. The company has invested mightily in developing low power x86 solutions, and has even developed a SoC menu that can allegedly be built at TSMC, although there haven’t been any takers for it yet. The Atom family developed in Israel despite the initial skepticism from management in Santa Clara has proven to be a breakthrough product with long legs – I expect to see continuing promotions in and from the Israeli group. The Atom architecture is being positioned for anything that’s not a PC, and PCs selling for less than $500.
And no company has better process technology than Intel today – although there are contenders, not the least being the IBM-GlobalFoundries partnership. The company has been pounding out 32nm parts for over a year now. Skipping the next logical node, 28nm, Intel will be in production in 2012 with 22nm parts. Ironically some of the first parts will be FPGAs for a fabless semi customer. However, I don’t think Intel is going into the merchant fab business.
Get low. Intel is adopting their basic low power (in watts and FLOPS) Atom processor into three market segments: midsize mobile devices like tablets and netbooks, small mobile devices like smartphones, and embedded systems. The company will meet competition from ARM on the midsize mobile machines, ARM and maybe MIPS on the small mobile devices, and ARM, MIPs, SH, and others in the embedded space (which includes STBs for example). Intel has a huge list of embedded devices it is targeting with its Tunnel Creek version of the Atom processor. All of the new versions of Atom have embedded graphics using the Imagination Technologies GPU/video IP, which will put them on par with almost all the ARM-based OpenGL ES and Open VG GPUs.
Get big. Having recovered from the embarrassment of the Larabee misstep, Intel has dusted itself off and brought out a multi-processor development system called Knights Ferry. Intel has also been steadily investing in parallel programming tools and techniques to exploit multi- (x86) processors. The development of x86 parallel processing software by Intel is primarily for the HPC segment where the company is being challenged by FPGAs and even more so by GPUs. Slow to embrace Open CL, I expect to see big parallel processor software development and tools announcements from Intel throughout 2011. If Intel is to gain any traction with its many-core x86 designs it’s got to blunt the momentum of Nvidia’s CUDA ecosphere and cachet.
Hold the middle. Intel has dominated the mid to low end PC market in notebooks and desktops with their IGPs, capturing 63% and 46% respectfully. With the introduction of the first core i5 chip code named Clarkdale the company put the IGP in with the CPU, creating a new category Embedded Processor Graphics (EPG). With Sandy Bridge (which we’ve been calling Sandra) the company integrated the CPU and IGP in one chip. This natural evolution is also designed to blunt AMD’s Fusion. It’s an interesting mix with Intel having better CPU performance and AMD having better GPU performance in their integrated parts. To offset AMD’s GPU advantage Intel is being forced to play nice with Nvidia and embrace its Optimus technology. As a result rumors of a settlement waft through the webosphere daily. As you might remember, Intel and Nvidia have been known to be quite prickly with each other.
Games, ray tracing, video, and workstations. Intel wants to be all things to all people (all the time). It has a footprint in every major segment and a varied level of success. Intel has valiantly listed the games that will run well on Clarkdale only to be scoffed at by Nvidia and AMD. When Sandy Bridge is made public, it too will have a games list, but sadly only DirectX 10 in a 2011 DirectX 11 world. Intel will side-step that issue and if they get pushed on it, they will point out that more people are and will continue to play DirectX 10 gamers than 11. Ray-tracing has long been a fascination of Intel’s as it looks like a tremendous cycle sucker. However, except for styling, CAD, and photorealism there really aren’t many applications for ray-tracing. And Nvidia is well on its way to convincing the world that you need GPUs to do ray-tracing. Intel may have a trump card however, with Imagination Technologies acquisition of Caustic Graphics Intel will have a GPU-based ray-tracing capability (if they choose to deploy it). Video in Sandy Bridge is good and maybe the ‘Intel can’t do video’ crap will finally stop. As for workstations, they own it. Intel’s challenge is figuring out how make the workstation market expand – they’ve got FLOPS to sell.
MIDs. Remember MIDS? Mobile Internet Devices – something between your phone and your PC? Introduced in 2007 at IDF Beijing, it was a category that never was realized – or was it? I think tablets, especially the less than 10-inch tablets are the long lost tribe of MIDs. I came here to praise MIDs not to bury them….
Summary. Intel does so many things right it’s hard to fault the company. And yet the stock market isn’t impressed. Barely keeping pace with the DJIA, INTC isn’t an exciting stock. However, for ten straight years Intel has been named one of top companies’ in corporate governance by the Dow Jones Sustainability Indexes (DJSI). Intel needs to show some traction in the mobile space with tablets and phones to get the stock market to pay attention.