Mentor's Rhines explodes IC market “myths”
At Semicon West, Mentor CEO says that the semiconductor industry recovery will be faster than that of the overall economy.
by Rick Nelson, Editor-in-Chief -- EDN, July 16, 2009
San Francisco, CA. Untethered from the economy at large, the semiconductor industry is poised to bounce back from recession as manufacturers move to new technology nodes, drive down unit prices, explode unit volumes, and open up new application areas. That’s the conclusion suggested by Walden C. Rhines, chairman and CEO of Mentor Graphics, in his Wednesday keynote speech at Semicon West.
“Despite the tough times you are to be commended for coming out,” Rhines told his audience, adding, “It’s really hard to believe that in recession like this one the semiconductor market can fare better than over all economy,” which embraces such troubled markets as those for automobiles, chemicals, and steel. But in his address, “World Semiconductor Dynamics: Myth vs. Reality,” he described why he believes the semiconductor market is indeed poised to fare better than other markets.
He did not sugarcoat the current condition of the semiconductor market, noting an “awesome decline” in semiconductor revenues to 2006 levels, adding that the Q4 2008 to Q1 2009 unit volume decline represented the largest in history.
He then took on what he called the myth that a decline in the overall economy drags the semiconductor market with it. He presented a graph comparing GDP with semiconductor revenues; the graph showed superficial correlation, but, Rhines said, if you do a regression fit, you’ll find the regression coefficient R2 is 0.0864, “so essentially they are not correlated at all. The semiconductor market follows a boom and bust cycle of its own. We have a phenomenon that can cause the semiconductor market to be healthier than the overall economy,” and vice-versa.
Rhines then took on the “myth” that manufacturers are slow in adopting new technology. “You see quotes that Moore’s law has collapsed” and that customers resist technology-node transitions, he said, presenting curves based on VLSI Research data showing that 65-nm and 45-nm adoption rates track those of earlier nodes, adding that if anything wafer starts at 45 nm are outpacing 65-nm starts at the same point in the 65-nm technology’s ramp-up. He cited other factors—such as reticle sales and fab capacity—that similarly indicate that manufacturers are adopting new technology nodes. Fab capacity, he said, is most plentiful at older technology nodes, concluding that “the adoption of leading-edge technology is at the same rate as in the past despite the recession.”
Rhines then addressed what he called the myth that “We are in an industry that is consolidating; as you mature, you consolidate.” To counter that myth, he pointed out that today’s top supplier, Intel, has about the same market share that Texas Instruments, the top supplier in 1972, had at that time. Further, he added, the top five suppliers today have less combined market share than the top five did 35 years ago; a similar characteristic holds for the top ten then and now. What changes, he said, is the makeup of the top players, with, for example, fabless chipmaker Qualcomm joining the top ten in 2008. He noted, “Nearly 500 Chinese companies who claim to be semiconductor producers intend to be competitors,” suggesting continuing evolution in an industry with lots of players. Part of the churn, he said, is due to companies unable to master transitions: from germanium to silicon, for example, or from transistors to ICs, bipolar to MOS, or from microprocessors to SOCs.
Rhines next looked at a key factor that differentiates the semiconductor industry from other industries: “If you want your costs to go down, you must grow unit volume.” Automobile production and steel tonnage production doesn’t grow much, he said, making it hard to continuously take advantage of learning-curve efficiency improvements. He said that the semiconductor industry, however, enjoys a circle of innovation that starts with unit volume growth, which drives increased R&D leading to 35% per year transistor cost reduction, opening the door to new applications and companies. He cited the digital camera as an example, which are just now entering a high growth phase, in which falling prices ignite volume sales. That in turn drives image-sensor revenues, enabling their vendors to achieve lower cost to enable new applications in medical and security equipment that needed low-cost image sensors.
He also cited opportunities in the PC area. He said that desktop PCs represent a mature business, but emerging netbook markets provide opportunities for makers of low-cost microprocessors and SOCs. Also, he said, high-end cell phones are gaining market share. The market for phones containing less than $40 worth of semiconductors are holding up, those whose silicon content ranges from $40 to $80 are declining, and those having semiconductor content greater than $80 are growing.
Every era has its drivers, Rhines said, citing the military in 1970s, desktop computing in the 1980s, communications in the 1990s, converged appliances now, and systems of systems in the next decade. But moving forward, he questioned, won’t rising design costs limit opportunities for new applications? No, he answered definitively: “Turn to numbers—the number of electronic engineers doesn’t increase much—perhaps 3% per year—compare with the number of transistors produced. The transistors produced per electronic engineer have increased by four orders of magnitude in 25 years.” He did note that embedded software design represents a challenge and noted that chip designers will be doing more system-level engineering.
In summary, he said that the semiconductor recovery will be faster than that of the overall economy, new technology adoption is at the same rate as in the past, there is no evidence of overall consolidation in the semiconductor industry. He added that despite the decrease in the design coat per transistor, design costs will increase because of the increasing role of embedded software. He concluded by saying, “Unit growth and innovation will drive 35% per year per transistor cost reduction needed to enable the next wave of wave of applications.”
-
An extremely well written summary. Kudos to Rick Nelson for accurate, readable reporting.
Wally Rhines - 2009-20-7 11:02:00 PDT -
Admittedly, key-noting Semicon this year is a tough assignment. But is that really his "good news"? --that "the semiconductor industry follows a boom and bust cycle of its own?". With semiconductor technology now reduced more and more to a lower end commodity status, and the world no longer awash in extra cash to fund the "booms" of an undisciplined industry, just what kind of ride is he really predicting?
Thomas Mintner - 2009-16-7 13:20:00 PDT
