It's time to stick a fork in the myth that there still are thousands and thousands of ASIC design starts every year that have revenue-generating potential.
iSuppli Corp. in the second quarter of 2002 reported that the number of new ASIC design starts was not the 6,000 to 8,000 that commonly had been reported, but rather a much, much lower number.
A previous iSuppli survey of the top ASIC suppliers, which account for the majority of the segment's revenues, revealed that the actual number of new ASIC design wins in 2001 actually was in the region of just 2,000 -- only a fraction of the number that is the accepted wisdom. iSuppli recently concluded a more extensive and detailed survey on the number of new ASIC opportunities that confirms our previous findings.
This latest research collected data from ASIC vendors comprising more than 80% of all ASIC revenue, and because of that high response rate and comprehensive coverage, iSuppli considers the results of the survey to be conclusive. The respondents, including eight of the top 10 and 15 of the top 20 suppliers based on 2001 market share, plus a few select smaller providers, reported the total number of new ASIC designs in 2002 was slightly less than 1,000.
The grand total for the industry comes out at roughly 1,450 or so for the year.
This is a fraction of the number that is frequently reported. iSuppli counts only those ASIC designs that have at least a possibility of generating revenue, so we are ignoring ASICs created for university research, for example. These designs, while important to the designers, are for the most part inconsequential to those participating in the industry.
Significant results of iSuppli's survey include:
--The 0.18-micron process node now is the leading manufacturing technology both in terms of product revenue, at 35%, and design activity, at 34%. However, the 0.13-micron geometry is likely to overtake the 0.18-micron process in 2003.
--Design activity foreshadows increasing ASIC potential in the consumer and industrial/medical segments, with data processing and storage likely to lag.
--Gate-array ASICs claim a higher percentage of design wins, at 18 percent, than they do ASIC dollars, at just 11%. With gate arrays yielding much lower revenue per design than cell-based ASICs, this indicates that gate arrays are likely to continue their declining share of the ASIC pie.
--Roughly 75% of designs were reported to work on the first pass, with a surprising one out of 10 ASICs requiring two passes or more before working, if ever.
While iSuppli will resist the temptation to speculate on the reasons why other reported numbers for ASIC designs are so grossly out of touch with reality, we do believe these inflated and misleading figures represent a danger to industry participants. A market with a small and decreasing number of new designs makes the competitive landscape much more intense for ASIC suppliers. While the average amount of revenue associated with each design is increasing, it is not sufficient to offset the reduced number of designs. Because of this trend, the ASIC market will be depressed in the 2005 time horizon and beyond.
ASIC vendors will need to pursue the remaining ASIC sockets more and more vigorously, and must be more selective about where they allocate sales and marketing resources. Suppliers to the ASIC design and supply chain that depend upon design starts for revenue -- such as intellectual property (IP) and Electronic Design Automation (EDA) vendors"need to do the same -- and prepare for lean times in general.
False expectations of a fairy-tale market of many thousands of new ASIC opportunities puts those suppliers in a precarious position indeed.
*Jordan Selburn is a principal analyst with iSuppli Corp.'s Market Intelligence Group. Contact Selburn at firstname.lastname@example.org