Yertle the Turtle was king of all he could see. Two years ago, all Intel could see was PC processors, and it was king of that market. But one day, Craig Barrett piled up all of Intel's money and sat on top of the stack.
From that lofty perch, he could see markets that Intel had never seen before. Wireless handsets. Network infrastructure. And Barrett declared, "I must be king of all I can see."
Sorry-that was Yertle (courtesy of Dr. Seuss). Barrett actually said, "Our goal is to be a leader in any market we play in." He clarified that a leader is one of the top two vendors in a market.
At the Intel Developer Forum, the company unveiled XScale, its weapon for these markets. The next-generation StrongARM processor covers an amazing range of 10 milliwatts to 1,000 Mips.
I must admit I was skeptical about Intel's ability to extend the StrongARM line after buying the technology from Digital.
Upon hearing of the purchase, Digital's StrongARM design team left en masse, eventually incarnating as Alchemy Semiconductor. Intel assigned Strong-ARM to its Chandler, Ariz., design team, which had theretofore produced i960 chips that were, shall we say, underwhelming in performance per watt. But the team, under engineering manager Jay Heeb, delivered an impressive product-even better than Alchemy's newest chip.
XScale's biggest target is the cellular handset market, which should top 400 million units this year and could double by 2003. Because third-generation (3G) phones need a much faster CPU than the ARM7 cores used in most 2G phones, they represent a discontinuity that will let new vendors into the market.
XScale delivers 3G performance within the cost and power budgets of today's phones. The biggest barrier to its success is that the Big Three handset vendors typically design their own integrated processors based on ARM cores. Intel instead will offer an integrated XScale chip customized for cell phones but not for any particular vendor.
I expect many cell phone vendors-enough to achieve Barrett's leadership goal-will choose to reduce internal development costs by accepting Intel's solution, which will also include flash memory as well as the DSP being co-developed with Analog Devices. With a 20 percent to 30 percent share of the handset market, Intel could be shipping more XScale processors than PC processors by 2003.
Linley Gwennap is the Founder and Principal Analyst of the Linley Group (www.linleygroup.com), a Technology Analysis firm in Mountain View, Calif.