The purpose of this paper is to identify the organizational constraints on science innovations in the midst of the increasing complexity of technology and markets and to search for measures to overcome them. For this purpose, we scrutinize the rise and fall of Japanese chipmakers in their commodity DRAM business during the last three decades, during which time all of them have been deeply wounded. We take up this business case mainly because the Japanese semiconductor industry seems to be a forerunner of various science-based industries facing rapid globalization and could provide instructive examples for them in an age of speed-to-market. We think that the rise and fall of Japanese chipmakers in their commodity DRAM business has been deeply influenced by three kinds of ever-growing complexities: the growing market-complexity triggered by the collapse of commodity DRAM prices in 1996, the growing (manufacturing) system-complexity boosted by the advent of 200mm fabrication plants (fabs) in the early 1990s, and the growing process-complexity in fabrication technologies necessitated by 64Mb commodity DRAMs. We explain how and why, compared with U.S. and Korean competitors, Japanese chipmakers could not respond to these growing complexities in a systematic and well-organized manner.