Quitting DRAM volume production — and, ten years on, semiconductors (1998)

How long do you hold a business you cannot win

At the centre of this decision is the question of how quickly to let go of a business you have concluded you cannot win. Shinozuka Katsumasa reached that conclusion about DRAM in July 1997 and cut off volume investment in the following year’s rebuilding plan. The judgment itself was fast. But because he chose a halfway form — keeping the plants and the workforce, continuing to build current products — another ten years passed before the business was separated completely. Those ten years can be read as evidence of how hard it is for a company carrying employment and production equipment to finish, while still in the trough, a retreat decided in the trough.

What came after the exit was not a simple success either. The year to March 2009, in which the semiconductors went, coincided with the global recession: a loss of $481.1M (¥45bn) and sales falling from the ¥700bn range to the ¥500bn range. Even so, no longer owning a business in which tens of billions of yen swing with supply and demand was the precondition for putting ATMs and contract manufacturing back at the centre. Yamamoto Masaaki spoke in 1969 of the low margins inherent in public-interest equipment; Shinozuka spoke in 1998 of a gap in investment capacity — in Oki’s history the question of how to measure its own size keeps returning. The semiconductor exit was the answer for which it paid the highest tuition.