For all the talk about Intel struggling to take a bigger chunk of chip sales to the embedded market from ARM-based devices, the general manager of Intel's embedded unit says the company has grown its embedded business significantly in the past few years by taking market share—a trend he predicts will become more pronounced in the years to come.
Intel set a record in 2010 with $43.6 billion in sales. But 92 percent of that revenue came from chips for PCs and servers. Combined, Intel's Embedded and Communications, Digital Home and Ultra-Mobility groups accounted for only 4 percent of the company's 2010 sales, or about $1.74 billion.
But according to Ton Steenman, general manager of Intel's Embedded and Communications group and vice president of the Intel Architecture group, over the past 10 years, Intel's embedded business has grown at a compound annual growth rate of 18 percent—more than the overall market and faster than Intel's overall growth.
"We clearly have gained market share," Steenman said during a recent press briefing in San Francisco.
What's more, Steenman said that after growing revenue in embedded 35 percent in 2010 compared to 2009, the unit expects sales to grow about 25 percent per year over the next three to five years.
Why? Steenman said seeds planted in the form of design wins over the past few years are starting to bear fruit. Lead times on embedded products are longer than those in the PC and consumer electronics markets—upwards of 18 months to several years. But once the products finally hit the market, Steenman said, they hang around for five to 10 years, if not more. Thus, design wins scored by Intel late last decade are paying dividends now and should continue to do so for the foreseeable future, he added.
"It's really about building an annuity business," Steenman said. "When that business builds up, you've got that long tail of revenue. That's where we've seen a lot of our growth."
Over the past couple months, Steenman—who took over Intel's embedded group late last year and has been with the company since 1982—has been talking up Intel's work with the Addias shoe company to build a "virtual footwear" retail kiosk. He's also been touting Intel's work with a company called Respondesign to build a connected exercise gym in recent appearances, including at International Data Corp.'s Smart Technology World last month.
In these cases and others, Steenman said, Intel engineers worked with the customers to develop a concept from the ground up. He acknowledged that Intel must pick its spots in such collaborations, which involve engineers and other resources. He said the company allocates resources to such products not necessarily because of the volume of chips that that particular customer may buy, but instead chooses them because they may serve as examples for other companies that may develop similar products.
"It really helps us to make it clear how our technology can solve some of these problems," Steenman said.
Sure there are all these things, but how many will use Intel parts?
Why use Intel parts in a multicore tablet, multimedia player or such?
Intel has tried to play in embedded space and has pulled out many times. It seems they are committed to only selling high-margin parts. Most embedded parts are not high margin.
Intel have a problem: Their investors expect growth to sustain high stock prices. Intel pretty much own PCs and servers, so they need to break into other markets to continue to grow. Unfortunately for them embedded is a low-margin game and you have to work a hell of a lot harder to get the same returns.
This is something Intel just does not know how to do, which is why they pulled out of embedded before.
Intel's "It's really about building an annuity business.." comment shows that they are used to being top of the pile and getting business come to them. That attitude is just not going to work in the embedded market.
No. They normally sell off or shut down the whole division.
What happened to 8080, 8051, 80251, USB parts, flash, XScale,...
Not to mention the companies they have bought, screwed up, then sold off like Dialogic.
Like many of these companies, Intel takes many of their strategic cues from Wall St. When it is fashionable to diversify, they buy into a new market. When it is fashionable to "focus on core competence" they sell off/kill parts of their business. Wehn it is fashionable to offshore they send fabs and development offshore...
I have personally seen Intel drop/sell-off the embedded communications market too many times (3-4 times) to trust them again unless they offered a very significant advantage and no disadvantages especially cost. Let's not forget the Network Processor "sell-off" along with the other embedded processor lines.
They ALWAYS say they are in it for the long run and then burn you 3 years latter.
10 to 20 years is a long time in this industry. Chips from that long ago are hard to process today because the fabs keep moving forward with new processes and the old ones are discontinued. It is important to partition your system in a way that standard products can fill most of the sockets. Full custom can cause these kinds of issues. They may redesign it in the new process, but designers always like to make "improvements" when they do that.
They still seem to be thinking of "embedded" only in terms of "embedded PC"
That is really a rather thin slice of the market.
The big volumes are at the microcontroller end of the market: even though Intel has disowned the 8051, it remains a very widespread architecture.
And whatever happened to the 8096, 80196, etc?
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.