Search This Blog

Tuesday, March 30, 2010

Grace Semiconductor Manufacturing: Marketing Trumps Technology?

Dr. Ulrich Schumacher, CEO and president of Grace Semiconductor Manufacturing, provided a standing-room-only audience with detailed insights and plans into the IC foundry’s strategic priorities and technology mix as part of the keynote address at SEMICON China.

I found the presentation fascinating as a description of some massive marketing challenges for semiconductor firms in an era marked by widely variable manufacturing capabilities.

Dr. Schumacher’s presentation, entitled, “China Foundry’s Opportunities
In a Recovering Market” began with an overview of the China chip gap that underlies the long-term expectations for chip manufacturing in the world’s most populated country. While China became the world’s largest IC consumer in 2007, for the foreseeable future China will be able to produce less than 20% of its total demand. The resulting chip gap underlies what some people think will be continued government support for SMIC, Grace and other companies (the US prefers to “support” the low tech industries of banking and agriculture).

Dr. Schumacher gave the audience a generous overview of Grace’s business operations and history. He said the company has already seen revenues rebound to Q3 2008 levels and forecasts all time quarterly revenue levels in Q1 and Q2 2010. The company’s customers are widely distributed by regions (over 50% of revenues from US, followed by China and Taiwan, each contributing 16%), and by technology: logic (44%) memory (16%) and embedded flash (10%).

Like many semiconductor manufacturing companies today, Grace has no intention keeping pace with Moore Law. They aim to excel with a technology portfolio designed to meet a targeted selection of applications in the consumer, communication, computer, and automotive applications. According to Dr. Schumacher, achieving the low cost position in profitable value-added niches still requires consistent technology innovation and capital investment. It also requires, “highly effective technology transfers based on partnerships and services along the value chain".

To explain the Grace strategy, Dr. Shumacher provided detailed market size, growth and technology trends for key application segments, including microcontrollers, SiGe/RF, SmartCards, touch panel controllers, LED drivers, and power management. The target applications drive a Grace technology portfolio that includes geometries of 0.35µm, 0.25µm, 018µm, 0.15 µm, 0.13 µm, and 90nm. Complimenting the mix of geometries are dedicated technology platforms for NOR, EEPROM, embedded Flash, SiGE, and PowerMos optimized for fast time-to-production, yield and cost.

Like many IDMs and fab-lites, Grace has developed a More than Moore technology strategy that is based on complicated market specialization and segmentation. For companies like Grace—and I think the same situation exists for TI, NXP, Infineon, STMicro and others--financial success with this model will be based not on technology or timing massive capital outlays but on very sophisticated marketing. All the key components in the business strategy are essentially classic marketing problems, such as target market identification and sizing, customer needs analysis, competitive analysis, pricing, etc. The business problem has little to do with technology; many companies have the capability to design and manufacture (through foundry or in-house) mid-tier ICs in top computer, industrial and consumer applications, but few can do so profitably for a sustainable period of time.

Increasingly, companies with semiconductor manufacturing assets will have to target specific niche markets to survive. They will have to make very judicious decisions on capital spending and R&D. Maybe these decisions will take the form of Moore-1 or Moore-2, signifying how far off the pace from leading edge Moore’s Law capabilities a manufacturer prefers to operate. Financial health in the mid and low tier IC markets will be driven by margins and capacity decisions that will change quickly overtime. Challenges will always arise as the niche applications reach volumes that allow next node production enabling steep price declines, or during downturns when other foundries will more aggressively compete on price. With a portfolio of production platforms, companies can compete by selectively adapting their niches to keep ahead of competitors and sustaining healthy margins. To do that well, a gutsy commitment to marketing seems a necessity. And, for an industry that shown a high tolerance for technology risk, a business strategy dependent on smart marketing seems like a healthy step forward.

Monday, March 22, 2010


What the LED Industry Can Learn From Semiconductors

I had the great opportunity to talk about SEMI and the LED industry at last week’s SEMICON China (see picture, standing room only). LEDs are the fastest growing semiconductor technology in the world right now (faster than PV). The market is projected to triple to $20 billion by 2013, and that’s just when the mega markets in lighting begin to open up. Currently, lighting consumes about 20% of the world’s electricity and solid state lighting (SSL) can reduce that by 50%. In other words, SSL is the most important energy conservation technology likely to make a big impact on our carbon footprint over the next decade.

The title of my talk was “Moore’s Law and Haitz’s Law and their Importance to Our Energy Future.” Everybody in the chip industry knows Moore’s Law, but Haitz Law is new to many semiconductor folks. Named after Dr. Roland Haitz, a now-retired scientist at Agilent Technologies, Haitz’s Law states that every decade, the cost per lumen falls by a factor of 10 and the amount of light generated per LED package increases by a factor of 20, for a given wavelength (color) of light.

Like Moore’s Law, Haitz's Law is an observation and prediction about the steady improvement over the years for LEDs. It is not based on physical science or natural law that can be observed in nature and proven by experiment, it a mere expectation of a learning curve or process optimization that will take place over time. It is my opinion that for Haitz Law to continue to be realized--like Moore’s Law in semiconductors--effective industry collaboration will be required.

I made my case by comparing semiconductor manufacturing technology of 1975 with LED manufacturing today. Both industries shared the following characteristics:

• Manufacturing was accomplished by highly proprietary processes using customized or highly modified equipment.
• Different wafer sizes and geometries were widespread (no standards)
• Throughput was only 50 wafers per hour
• Yields were low
• And Equipment Productivity was a challenge

In 1975, Moore’s Law had been realized for twenty years by individual companies, each working independently, without the benefit of technology roadmaps and industry standards. But by 1991, technology roadmaps and SEMI International Standards were essential to sustaining Moore’s Law.

The LED industry has been achieving the pace predicted by Haitz Law to the present time without any industry collaboration, such as technology roadmaps and standards. This is about to change. According to many experts, efficiency gains in lumens per watt will be slowing and limited to approximately 2x improvement before reaching their limit. Cost reduction goals according to the Department of Energy to reach large SSL markets, however, will require a 20X improvement. The performance of white LED devices depends on both the correlated color temperature (CCT) of the device and, to a lesser extent, on the color rendering index (CRI). We are beginning to approach what are perceived to be the practical limits of these parameters. Further cost reductions must come from manufacturing improvements in yield, productivity, throughput and other cost reductions.

The conclusion of my talk was that from what we know from the semiconductors and sustaining Moore’s Law, to continue to achieve cost reductions predicted by Haitz Law, highly developed forms of industry collaboration, such as technology roadmaps, industry forums and industry standards will be required.

Currently, no such industry collaboration activities are underway in the industry. Manufacturers of HB-LEDs are often vertically integrated and extremely protective of their intellectual property. Participating in standards activities, roadmaps and organized supplier dialogs are seen as a threat to the highly proprietary processes that enabled today’s market positions. In addition, everybody in the industry is dealing with enormous demand—including reported shortages in MOCVD tools—to meet today’s demand driven by LCD displays. No one has time.

The PV industry was similarly skeptical of organized industry collaboration 2 years ago, yet today discussions on industry roadmaps are beginning to take shape and standards activities are well underway (there are now 22 SEMI International Standards committees, working groups and task forces established with over 400 participants).

We are only at the beginning of a massive SSL market explosion expected to burst in around 4-5 years. To reach the cost reduction targets for commercial and residential lighting markets some of these issues will have to be addressed. And the social benefits and goals are hugely important and go beyond the interests of any one company. Large scale penetration of SSL in the US alone can replace up to 200 coal fired power plants, spewing tons of CO2 emissions.

To begin the process of industry collaboration and break down the walls of secrecy and suspicion, SEMI has organized a HB-LED Steering Committee to search for ways of achieving meaningful, effective collaboration. For more information on the Committee and what SEMI is doing in HB-LEDs, email me.

Monday, March 01, 2010

Winner Take All

I had the great opportunity to see the induction of Mr. Richard Elkus into Silicon Valley Engineering Hall of Fame of last week. He currently serves on the Boards of KLA-Tencor and Lam Research. The award celebrates the accomplishments of engineers in Silicon Valley who have demonstrated outstanding professional achievement and have made significant contributions to the Silicon Valley community. Past nominees included William Hewlett, David Packard, Gordon Moore, Steve Wozniak, Ken Levy and TJ Rodgers. Stan Myers was recognized in 2008 and another former boss, Bob Frankenberg, was selected to the group in 2001.

Previously, Elkus was Co-Founder, Chairman and CEO of Prometrix Corporation, which revolutionized the concept of computerized wafer mapping that is now fundamental to the analysis of wafer development and production in the semiconductor manufacturing process. Prometrix merged with Tencor and later with KLA, becoming KLA-Tencor. He spent his first 10 professional years at Ampex Corporation, where he headed the team that introduced the VCR in September 1970.

In preparation for the evening, I quickly read his book, Winner Take All, which thoughtfully explains how industries develop and how nations, not just companies, compete. Over the past thirty years, the United States has basically surrendered entire industries to Asia. Americans no longer make cameras, TVs, MP3 players, computers, cell phones, or DVD players, and increasingly semiconductors. The conventional wisdom behind this loss of industries is cheap labor costs, but the real reason is American economic policy. While Asian leaders develop and nurture high technology industries because of the spillover benefits to the economy and society, US policy makers disregard and ignore it. “Economic planning” is treated as an anathema by US policy makers, somehow akin to the failed Communist economic planners of half a century ago. Rather than ensure the growth of productive and healthy industries like high tech, US policy makers campaign on “free markets” while rigging favors for political insiders and campaign contributors.

The US has an economic plan, it’s just not rational. We spend trillions on nineteenth century industries like agriculture and banking, and let high-paying jobs in high growth industries leave for Asia. Unfortunately, high tech innovation may just follow high manufacturing overseas, forcing the US try to sustain itself on service industries alone. Unless you’re a banker huddled in well-financed bunker in NYC, this isn’t a policy that’s going to do most folks much good. Many of my Republican friends are quick to claim that jobs leave the US because high taxes and government regulation. I think it’s more complicated than that. Innovative, high technology industries require thoughtful public private partnerships to thrive. There is no free market nirvana where taxes are low and jobs are good, health care is cheap, and opportunity is unbounded. There are only well run and poorly run countries. Richard Elkus knows this well.