September 28, 2012 - Want another snapshot of who’s leading the pack among pure-play foundries, and who’s falling off the curve? Look more closely at <45nm offerings, as framed by a recent analysis from IC Insights.
Overall, total pure-play IC foundries will register about 30% of their sales in <45nm process technologies, up from 22% in 2011, IC Insights calculates. Older (>0.18μ) technologies will account for 13%, down from 14% in 2011 and 15% in 2010.
GlobalFoundries, which earlier this year passed UMC to become the No.2 foundry, has rapidly narrowed the technology gap enjoyed by TSMC over its foundry rivals. GF actually has had a significantly higher percentage of its sales coming from 45nm and below process technologies than TSMC (55% vs. 26% in 2011, 65% vs. 37% in 2012), thanks to its MPU-centric focus. TSMC’s still far ahead in terms of actual dollars ($3.8B vs. $1.9B in 2011, $6.2B vs. $2.8B in 2012).
In contrast, look at the next two "Big 4" pureplay foundries: UMC’s <45nm technology generated 6% of sales ($243M) in 2011 and 11% ($427M) in 2012, while SMIC has gotten less than 1% in both years ($1M and $6M). IC Insights thus connects the dot from <45nm sales, through the metric datapoint of revenue/wafer, and into profit margins. (To be fair, UMC is skipping 45nm and offering a 40nm process.)
That lead for TSMC and GlobalFoundries is widening more now that 28nm devices are starting to emerge. "Although many of the pure-play foundries other than the Big 4 focus on specialized processes and technology, the process technology gap between the other significant pure-play IC foundries and the leading-edge producers is enormous," IC Insights notes. Beyond UMC and SMIC who are falling behind, of the next 14 pureplay foundries only four (TowerJazz, Grace/HHNEC, Dongbu, and Xinxin) are expected to have even limited <90nm capabilities this year. Their total overall sales will amount to $4.6B, only about 15% of the total pureplay IC foundry market.
What’s the takeaway from all this? Successful (i.e., profitable) foundries will be the ones "at the leading edge of the process technology roadmap," IC Insights says. (Note that GlobalFoundries is already talking about its 14nm FinFET technology coming in 2014, barely a year after its new 20nm process.) And for all chip companies, those who have the money have the ability to invest in R&D as required to keep up with more complex IC designs and new process technologies.
Major pure-play foundry comparisons. (via IC Insights)