EETimes.com – Viewpoint: Is semiconductor industry consolidation inevitable?

EETimes.com – Viewpoint: Is semiconductor industry consolidation inevitable?.

Inevitability of semiconductor industry consolidation seems to be widely accepted. Hitachi and Mitsubishi combine to form Renesas; next Renesas and NEC combine; AMD acquires ATI; etc. Articles discussing the consolidation that has already occurred seem to appear everywhere, with dire forecasts for the future.

Yet the data show exactly the opposite. According to my chief market researcher, Merlyn Brunken, the semiconductor industry has been slowly “deconsolidating” since the 1960s. Consider the market share of the No. 1 semiconductor supplier, Intel. Intel’s market share is the same today, about 13 percent, as the No. 1 supplier was 35 years ago when it was Texas Instruments. In between, NEC was No. 1. The names have changed but the market shares haven’t.

What about the combined market shares of the top five semiconductor suppliers? That percentage has been slowly declining since the 1960s and is about 33 percent in the most recently reported data, less than the 35% reported in 1972. And the combined market share of the top ten? Also decreasing, albeit very slowly, from 48 percent in 1972 to 46 percent in 2008.


Click on image to enlarge.

We might speculate that consolidation has been focused on companies in the manufacturing-intensive part of the semiconductor industry, like the DRAM industry, especially during the past decade. The data show, however, that the market shares of the top one, three, five and ten DRAM suppliers have all decreased since the turn of the century. Did they consolidate earlier? Not in the last 25 years. In fact, the market shares of the top one, three and five DRAM suppliers are exactly the same today as in 1983 (the earliest year for which I have DRAM data) and the top ten have modestly less market share than in 1983.


Click on image to enlarge.

So what’s going on? Why isn’t the semiconductor industry consolidating like most maturing industries? The answer seems to lie in the way the semiconductor industry regularly reinvents itself. Turnover of names of the top ten semiconductor companies is high, with more than 50 percent disappearing from the top ten since the 1950s.


Click on image to enlarge.

Maintaining, or growing, market share in the semiconductor industry requires achieving a leading position in whatever new technology is driving growth. Few companies have been able to do that. To be in the top ten in the 1950s meant you were a leader in germanium or silicon transistors. In the 1960s, sustained growth required leadership in bipolar integrated circuits, eliminating 40 percent of the original top ten in a single decade and 80% of the top ten by the 1970s. In that decade, continued growth was driven by MOS memory, bringing three Japanese companies, NEC, Hitachi and Toshiba, into the top ten. In the 1980s and 90s, leaders in microprocessors moved up the ranking, particularly Intel and Motorola. This changed in the 1990s, as SoC leaders like STMicroelectronics and TI gained momentum. In the most recent decade, fabless companies like Qualcomm, and the primary foundry, TSMC, moved into the top ten.

Looking back across the past six decades, TI is the only company that has remained in the top ten throughout semiconductor history. When consolidation does occur, it is mostly among those companies that fail to adapt to the next level of emerging technology.

What about the future of this turnover phenomenon? Will market shares of leading semiconductor companies continue to decrease? That, of course, remains uncertain. But the semiconductor industry does have an attribute that makes it different from other major industries. That characteristic is the phenomenal growth in unit volume. Nearly 15 percent more chips and 50 percent more transistors are shipped each year than the prior year. Compare that 50 percent compound average growth rate to a mere 0.1 percent for automobiles over the last ten years, 1 percent for crude oil, and 9.3 percent for computers.

Growing cumulative unit volume at such an aggressive rate drives costs down the learning curve, reducing the cost per transistor 35 percent per year over the last ten years. And that enables totally new applications to move into the mass consumer market on a regular basis. For instance, a 2001 MP3 audio player cost about the same as today’s portable video player which has ten times the memory capacity with the same size and power as the 2001 version. If unit volume growth continues (and it shows every sign of doing so), new applications will emerge and semiconductor pervasion will continue to increase. The result? There will be new leaders, as those emerging applications become the growth drivers of the decades ahead.


Click on image to enlarge.

One of the more amazing aspects of the increasing pervasion of semiconductors into new applications is the significant growth in revenue of existing applications as the cost per unit decreases. Consider the digital camera. Most of the semiconductor content of a digital camera consists of non-volatile FLASH memory and the image sensor. In the early 1990s, solid state image sensors sold for $20-25. Image sensors were a negligible portion of the semiconductor total available market (TAM) until the current decade. During the 1990s the price per sensor fell dramatically from the $20-25 range to about $5. At this price point, unit volume soared, making image sensors more than 3 percent of the semiconductor TAM in the last few years. At the same time, NAND Flash memory prices fell nearly 65 percent per year, propelled by the growing unit demand. The result was a substantial net growth in the market for digital cameras and the semiconductors required to make them.


Click on image to enlarge.

This model is repeated again and again in the semiconductor industry. Unit volume growth drives reduction in the cost per unit of semiconductors, fueling market growth and new applications. The revenue generated by this growth provides funds for development of new technologies and further cost reductions, enabling additional applications. In the case of the digital camera, the cost reductions are only the beginning. Low cost solid state imaging and storage have enabled a host of high-volume applications of digital photography such as security, medical imaging, automotive applications and many more areas. The total semiconductor TAM grows because of—not in spite of—the 35 percent per year reduction in the unit price of transistors.


Click on image to enlarge.

When we have an application that is extremely high volume, the cost reduction activity drives changes that enable increased semiconductor consumption globally. The handheld wireless phenomenon is a good example. New subscriptions for wireless communications in India alone in the fourth quarter of 2009 were more than 50 million. That volume has driven down costs for application specific chip designs so that a $15 cell phone is now a reality. That’s a along way from the $3,995 price tag in 1983 for the early commercial handheld cellular phones (Motorola’s DynaTAC 8000X).

What does all this say about our future as an industry? As long as transistor unit volume continues to grow at such high rates, we will continue to enable new applications that grow the total market. On the surface, the makeup of the semiconductor market looks very stable, with the only change in the last 15 years being the five point gain of wireless communications at the expense of computing. But underneath, there have been dramatic changes. The market for desktop personal computers has matured while the market for notebooks and netbooks is growing rapidly. Consumer electronics has remained a relatively constant percent of the semiconductor TAM in the last fifteen years but the growth of video games, flat screen TVs, MP3 players, digital cameras, etc. has dramatically altered the content of that segment.


Click on image to enlarge.

Click on image to enlarge.

For the future, new applications will continue to change the character of the semiconductor market as it grows. New companies will enter the semiconductor industry with new architectures, new device structures and new packaging techniques to satisfy emerging needs.

What about consolidation of the semiconductor industry in the years ahead? It’s possible that our industry will mature and broad consolidation will occur. But if history has anything to teach us, it is that new applications will create new challenges. Companies that innovate to meet those challenges will move ahead. And semiconductor companies that continually reinvent themselves will successfully compete with new comers for the top ten positions.

Walden Rhines is chairman and CEO of EDA vendor Mentor Graphics Corp.

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: