Blog Post

Intel Is Branching Out With Embedded SoCs

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

Intel, the world’s largest chip maker, is spreading its R&D efforts far outside of its server and PC kingdom. The company has just launched a new line of products that will combine four processors onto a single chip, reducing both power consumption and the footprint required by the chips.

Intel rightly points out that in a connected world, devices ranging from ATM machines to mobile phones need more speed (but lower power) to offer next-generation use cases for businesses and consumers (you know, things like your refrigerator texting you when you’re out of milk.) So it’s gearing up for this revolution with a line of system-on-a-chip devices, later versions of which will be based on the Atom processor, which was built for networks and mobile Internet devices. One of these SoCs, code-named Linmore, should be available for mobile phones in late 2009 or 2010.

Intel CEO Paul Otellini told us that Atom would be in mobile phones and in embedded devices, which is smart in case the netbook market doesn’t pan out. But will the embedded market embrace Intel? Will the mobile phone market embrace Intel? Most sources in the industry doubt this will happen given Intel’s behavior as the primary provider of chips in the x86 market.

However, the Intel effort does bring software issues to the fore. One of the reasons Intel argues that its SoCs are better is because developers can use a common operating system that will stretch across multiple platforms, something that’s already proving important in creating a user experience that consumers can embrace. That common software platform is why Nokia bought TrollTech, which means Intel might be able to use the common platform edge to push out other embedded chip guys.

2 Responses to “Intel Is Branching Out With Embedded SoCs”

  1. First off, this “revolution” has been going on for a long time, without Intel’s help.

    So can Intel succeed? Maybe, but with what common OS? The only real common OS that can span all the way from (fairly small) embedded all the way to servers is Linux. And with Linux, x86 compatibility isn’t that important. (Windows is a family of OS’s- WinCE isn’t the same OS as Vista; Linux does it with one kernel).

    I don’t see Windows taking over embedded, either. So Linux is #1, then Windows, VxWorks, etc.

    Many (most?) of these embedded devices will not have a display — the interface will be through the browser; home routers are an example (and use ARM). And both the browser is cross-platform and most server tools (e.g. web servers) are cross platform.

    I’m also skeptical about the “common user interface” – different tasks should be best with different user interfaces. Playing music is different from configuring a router.

    Finally, I took a quick look at the first Intel SoC. It does look like a nice chip for high end embedded, but it’s nothing remarkable for the industry — others have been doing this for years. It does have some interesting features for Intel, such as CAN bus support (which leads some to speculate that Intel is going after the automotive market). At first glance, it looks like competition for Marvell and such, and I’m not sure Intel can take them on. Also, “7 years availability” doesn’t sound great to me – for embedded, that’s not a long time, and the embedded CPU/MPU/MCU/DSP companies typically provide a much longer lifetime (>10 years). Heck, Freescale is still selling the 68000.