Shocked by the fact that the semiconductor industry is DECONSOLIDATING, rather than consolidating? In my April 5, 2010 article in EE Times, I showed the data for the largest one, top five and top 10 semiconductor companies. Surprisingly, the combined market shares of all of these categories have been flat, or decreasing, for more than 35 years.
As a result of the article, many people asked me if this phenomenon was limited to just the top 10 largest semiconductor companies. The answer is “no”, at least for recent years. Figure 1 shows the market shares of the top 1, top 5, top 10 and top fifty semiconductor companies during the last decade. The combined market share of the fifty largest semiconductor companies has DECREASED by about nine percentage points (900 basis points) during the last decade.
Figure 1. Deconsolidation of the semiconductor industry
Source: IC Insights
Why does this result seem so surprising to most people? It appears to be related to our general awareness of the consolidation of semiconductor MANUFACTURING. Foundries have been taking over a significant share of silicon wafer manufacturing for all but the largest IDMs, most memory companies, and some analog production.
This consolidation has been apparent to everyone but hasn’t progressed as far as you might think. Figure 2 shows that semiconductor revenue based upon foundry wafers grew to about 20% of total semiconductor sales by the year 2000 and then remained flat for most of the next decade. The 20% share of revenue assumes that fabless semiconductor companies sell their foundry-produced products at about a 2X multiple of their foundry cost. This 20% share has recently begun to grow again. Exceptionally large capital investments by foundries in 2010 and 2011 will inevitably increase the share of semiconductor wafers sourced from foundries.
Figure 2. silicon foundry revenue as a percent of total semiconductor revenue
Source: Semico Research
Thus far in 2011, about 33% of semiconductor capital equipment spending has come from foundries. That percentage isn’t likely to be sustained indefinitely but the effects of the last two years are still likely to cause excess capacity sometime in 2012 for foundries producing leading edge technologies at 28nm and below. Wafer prices will undoubtedly experience downward pressure, as will component prices of fabless producers. Lower prices will accelerate design activity at the advanced nodes and new applications will emerge, leading to the next wave of growth in late 2013 or early 2014.
What about the relative market share of foundry-sourced products? While the overcapacity will drive down wafer and component prices, the growth of new applications to fill that capacity will eventually lead to growth of the total semiconductor market and foundries will be one of the beneficiaries.
Despite the growing concentration of manufacturing capacity among fewer companies, the market shares of the companies who sell semiconductors continue to decline. As discussed in the April 5, 2010 article
, this trend has continued for the last 35 years because new technologies and applications of semiconductors continue to emerge and the leaders in each new category are frequently newcomers to the semiconductor industry. Many semiconductor industry analysts argue that the trend must reverse itself because the cost of designing and marketing complex system chips is too high for new, or smaller, companies. And yet, deconsolidation continues. Will it ever change? Probably. But only when new technologies and applications cease to emerge. That might be a very long time in the future.
--Walden C. Rhines is chairman and CEO at Mentor Graphics.