Texas Instruments once touted its strategic partnerships with communications companies like Nokia, Ericsson and 3Com as the linchpin to its dominance of the analog and mixed-signal IC business. This year, TI is quietly grumbling about low margins in the custom business and loudly proclaim- ing its desire to generate more revenue from standard catalog parts-albeit those for specialized niches, such as cell-phone power management and DSP data conversion.

National Semiconductor, under president Brian Halla, promoted the massive mixed-signal system-on-a-chip and advertised, "What can we build for you today?" More recently, National has touted the specsmanship of its analog building blocks. Its Cyrix processor unit, once deemed the key intellectual property in its integration strategy, has been sold. Still, that hasn't stopped the company from challenging ADI in the Taiwan-dominated custom scanner-chip market. More recently, National has defined a strategy for capturing the cellular basestation market with high-speed data converters and gain amps. ("What's a nice multimarket supplier like National doing in a focused market like cellular basestations?" joshed marketing director Mark Levi.)

If you find this confusing, watch how Lucent's Microelectronics Group defines itself in the analog world. The company is building a product group under TI veteran Mark Granahan, and as TI is now doing, will try to define specialized building blocks that take advantage of its talents and skills in communications ICs.

The truth of the matter comes from National's Levi: Standard parts or not, sometimes you see a target of opportunity so big that it makes sense to ramp a whole business around it.

A Silicon Valley acquaintance gave me a piece of advice when he got his law degree and switched his own business from market research studies to patent law. "Once you hang out a shingle," he told me, "your business is what your customers want it to be."