Help your competition and help yourself

By Brian Dipert, Technical Editor -- EDN, 12/7/2000

In the late 1990s, two large semiconductor manufacturers, AMD and Micron, seemed to be on a fast train to trouble. AMD's nemesis, Intel, had made it abundantly clear that it had no interest whatsoever in licensing the Pentium III and follow-on µP local buses to its pesky competitor. The then-ubiquitous Socket 7 interface was running out of bandwidth steam and was amenable to neither a separate L2 cache bus nor multiprocessor system configurations. AMD turned to the Digital Equipment-developed Alpha EV6 bus—a capable and proven, but obscure alternative that threatened to strap third-party core-logic chip-set vendors with a steep, long learning cycle. Without matching core-logic support, AMD's Athlon CPU hopes would be stuck in idle.

Micron was in similar straits. Intel intended for the DRAM industry to migrate to Rambus' proprietary, royalty-burdened Direct RDRAM (Rambus-DRAM) memory, beginning with the Pentium III generation and hitting full stride with Pentium 4. These plans countered Micron's historical strength of mass producing low-cost, industry-standard, royalty-free DRAM architectures, such as DDR SDRAM (double-data-rate synchronous DRAM). Intel's ambitions also didn't match up with the project investments and resource allocations that Micron's management had made, threatening to put the then market-leading company behind its competitors.

But Micron had an ace in its pocket. The company has for many years not only bought core-logic chip sets from third parties for use in its PC system designs but is unique among PC (and DRAM) manufacturers in sometimes using chip sets of its own design. It wasn't much of a stretch for Micron to combine its core-logic expertise with its DDR-SDRAM ambitions. And the benchmarking consequences of the resulting chip set, compiled by Bert McComas at Inquest Market Research (www.inqst.com) and others, attest to Micron's success.

Next, the memory maker Micron found an intriguing ally in the µP maker, AMD. AMD certainly knew how to make CPUs, but it didn't have much expertise in core logic. Financial and other details weren't made public, but it's generally known that AMD got quite a bit of help from Micron in developing its various chip sets for Athlon and follow-on CPUs. AMD, unlike Intel, hasn't publicly stated ambitions of becoming a high-volume core-logic supplier. So, although it shipped limited chip-set quantities to start the Athlon snowball rolling, AMD turned around and handed over those designs to other chip-set vendors, such as Acer Labs, ServerWorks, Silicon Integrated Systems, and Via Technologies.

Will these third-party chip-set manufacturers also use the DDR-SDRAM-controller expertise they obtained from AMD in developing chip-sets for Intel CPUs? Highly likely. And will systems manufacturers hook these chip sets up to DDR SDRAMs from companies other than Micron? Most assuredly. But consider the alternative and take a look at the momentum that both AMD's CPUs and DDR SDRAM have established in recent months; you'll probably conclude that the return on AMD and Micron's investment has been pretty good.

This story isn't just about two visionary start-up companies who have beaten back the self-serving vision of a semiconductor giant (helped in no small part by the giant's missteps in executing to that vision). This tale is also a lesson in how you might conduct your own business. The tendency is to see all competitors as, well, competitors and resist any technology-sharing opportunity as a threat to your slice of the market pie. Sometimes, though, by limited cooperation with your competitors, or "co-competing," you make the pie proportionally larger than your slice might shrink. Think about it.

Author info