THE COURSE / M13 THE INDUSTRY / M13.4
Research and the roadmap
After this lesson you can:
- Say how much the industry spends on research, and who pays for which kind.
- Explain why competitors share research in consortia, and what SEMATECH and imec are.
- Say what a roadmap is for, and how long an idea takes to reach a fab.
Loading the lesson…
Sources
- A. Varas, R. Varadarajan, J. Goodrich and F. Yinug, "Strengthening the Global Semiconductor Supply Chain in an Uncertain Era" (Boston Consulting Group and the Semiconductor Industry Association, April 2021). An industry association's study, with its consultants. The value chain in seven activities, 2019: design 53% of industry R&D, 13% of capex and 50% of value added; wafer fabrication 13%, 64% and 24%; assembly, packaging and test 3%, 13% and 6%; EDA and core IP 3%, under 1% and 4%; equipment 9%, 3% and 11%; materials 1%, 6% and 5%; pre-competitive research 15 to 20% of R&D. R&D 22% and capex 26% of chip revenues, about $90 billion and $110 billion in 2019. Segment economics 2016–2019 (gross margin, R&D, capex, operating cash flow, % of revenue): IDMs 52, 14, 20, 17; fabless 50, 20, 4, 20; foundries 40, 9, 34, 15; OSATs 17, 4, 16, 2. IDMs about 70% of 2019 sales; fabless under 10% of sales in 2000, almost 30% in 2019; foundries 35% of capacity, 50% without memory, 78% at advanced and trailing 300 mm nodes. Top 5 foundries' capex about $75 billion 2015–2019; top 5 fabless firms' R&D $68 billion; the top 3 in each activity hold 50 to 90%. A leading fab $5 billion (analog) to $20 billion (logic and memory); 2 to 4 years to build, 3,000 to 6,000 staff; one EUV machine $150 million; 10 to 15 years from paper to volume, EUV almost four decades. About 75% of capacity in East Asia and mainland China; below 10 nm, Taiwan 92% and South Korea 8%. More than 50 points where one region holds 65% or more. Incentives worth 25 to 30% of a new fab's 10-year cost of ownership in Taiwan, South Korea and Singapore, 10 to 15% in the US and Europe. Disruptions: Sumitomo epoxy 1993, the 1999 Taiwan earthquake, 2011 Japan, Japan's 2019 controls on hydrogen fluoride, fluorinated polyimide and photoresist to Korea, the 2020 and 2021 substrate fires.
- K. Flamm and Q. Wang, "SEMATECH Revisited: Assessing Consortium Impacts on Semiconductor Industry R&D," in Securing the Future: Regional and National Programs to Support the Semiconductor Industry (National Research Council, 2003). In a perceived crisis, 14 US semiconductor makers formed SEMATECH over 1987–1988 with US government support; Japan's VLSI projects were seen as having advanced Japanese producers. The Defense Department paid half of a consortium budgeted at $200 million a year, for its first eight years; government funding ended by mutual agreement in fiscal 1996. Its focus "zigged and zagged", with a growing emphasis on improving the equipment and materials US chipmakers used. Members committed to common projects; the data suggest SEMATECH reduced members' R&D somewhat, which Irwin and Klenow read as cutting duplicated R&D. Its effect on cooperating suppliers' R&D could not be established.
- IEEE, "International Roadmap for Devices and Systems" (read 28 October 2026). The ITRS was produced annually from 1998 to 2015 by industry experts from Europe, Japan, Korea, Taiwan and the US, as the main reference into the future for university, consortia and industry researchers. The IRDS began in May 2016 under IEEE sponsorship as its successor, to simplify academic, manufacturing, supply and research coordination.