The Texas Instruments DRAM venture, and selling out to Micron (2000)

Borrowed technology goes stale unless you keep borrowing

As a way of touching leading-edge manufacturing technology without owning a single patent of its own, being folded wholesale into TI’s production system was, at the time, a rational choice. An 80% yield was the proof of it, and president Kametaka Sokichi had grounds for saying he had no intention of running a losing business. But borrowed technology goes stale unless you keep borrowing. From 0.28 microns to 0.18, and then to 300-millimetre wafers, a staircase of investment invisible at entry piled up over ten years.

The exit was decided at a moment when the venture had climbed to ¥5.1 billion of recurring profit, with accumulated losses in sight of being erased within two years. That it was decided not on near-term profitability but on whether the next step up could be climbed is what marks this judgement. President Mizukoshi Koshi made “synergy” the yardstick because he could find no reason to keep pouring steel’s earnings into a business to which neither steel’s technology nor its people transferred. What survived that same yardstick was power generation, which could use the operating experience of a steelworks directly. The technology borrowed in semiconductors left with the shares; the technology used in power stayed. What separated success from failure in diversification was the difference between technology whose value vanishes when you stop investing and technology that remains in your hands when you do.