Memory Lithography and Product Strategy. Consistent Development Cadence with Probability Management.
G. Dan Hutcheson
The Chip Insider®
Memory Lithography and Product Strategy: Micron recently announced that it was jumping back into EUV with plans to install a tool into its DRAM fab in Taiwan. Samsung has been a proponent of EUV, but both Micron and SK Hynix have been reluctant to adopt the technology. This was most notable when Micron canceled its EUV tool order - let the Logic boys do the learning. Last year, TechInsights’ Senior Technical Fellow Dr. Jeongdong Choe answered the question brilliantly, “If you really believed 3D-DRAM was… this roadblock to EUV adoption has been cleared.”
Consistent Development Cadence Results from Solid Probability Management. It is a general feature of technology development that there is always a probability of failure along any single path to success. Most of the time, these failures are not fatal, just setbacks – like losing a hand at the poker table. What makes the development of technology so fraught with failure is its inherent reliance on the ‘we’ll think of something’ strategy. Failure to think of something results in failure to bring a new product to market. Investing heavily in R&D is the classic business model for addressing the ‘failure to think of something’ mode. Enough money in and eventually something monetizable should come out, right? But not always. Hence the failure of so many technology giants over the decades due to not directing the money into the right projects. IBM broke this mold, reinventing many times over its 100+ year history. Instead of Research AND Development, IBM pioneered a research, then, or as Theodore von Karman put it, “Scientists describe what is. Engineers create what never was.” In the early days of the semiconductor industry, R&D was mostly a silicon sandbox for PhDs to play in. This was especially true at Fairchild Semiconductor which would eventually drop out, in no small part, due to this. As Andy Grove emphatically told me years later, “NOTHING got out of R&D into manufacturing at Fairchild.” Intel would go from a node-missing basket-case in the 80s to the giant the world would come to see it as in the 90s. Key to that would be the development of the … Using this probability management model, Intel was able to extend its lead throughout the 2000s with technology it had been working on for 10 or more years.
Maxims: Technology Adoption Probability: The likeliness of technology adoption is proportional to the value of the problem it solves less the pain of implementation. https://www.chiphistory.org/935-technology-adoption-probability Customers are not visionaries and developing markets are not big. https://www.chiphistory.org/936-customers-are-not-visionaries-and-developing-markets-are-not-big
“Execution is Everything” — John Doerr
Subscribers can view the full article in the TechInsights Platform.