MEMS Market Can Expect Steady Growth, Structural Change


Bookmark and Share

MEMS Market Can Expect Steady Growth, Structural Change

By Paula Doe, SEMI Adjacent Markets

High-volume demand for MEMS means steady double-digit growth — and likely also some structural changes in the market. “MEMS is no longer a niche market,” argued Susumu Kaminaga, president of Sumitomo Precision Products Co. Ltd., at the recent MIG MEMS Executive Congress in Scottsdale, Arizona. “It’s now a mature industry with a few dominant suppliers trying to achieve economies of scale. Suppliers are now competing with larger vendors who ignored this market five years ago. We expect to see consolidation, with the winners being those who can best apply CMOS efficiencies to MEMS.” He said Sumitomo’s equipment unit SPTS expects a 200 percent jump in revenue this year.

Market Rebounds Faster than Expected; Analysts Expect Continuing Double-Digit Growth

The MEMS device market has rebounded faster than expected, said market analysts, predicting 12-16 percent annual growth for the next five years. Yole Développement president and CEO Jean Christophe Eloy said this year is on track to see 14 percent revenue growth for packaged MEMS devices, to become an $8 billion business. He expects growth to accelerate going forward, as a spate of new devices now in development hits the market, lower prices spur new applications, and biomedical applications start to take off, to average about 16 percent growth for 2009-2014, to create a $14 billion opportunity by 2014. “The fast recovery was the most surprising thing,” said Eloy. “It was faster than we expected — but the fall in prices was incredible.”

With the faster-than-expected pickup in growth this year, iSuppli MEMS analyst Jérémie Bouchaud has upped his forecast to 19-20 percent revenue growth this year, to a $7 billion market. The rebound has created demand to refill the automotive inventories all through the supply chain, boosting automotive MEMS growth to 27 percent overall, and up to 40 percent at some suppliers, he noted. Picoprojectors, the iPhone 4 and the iPad have also generated strong sales for MEMS, with slate tablets coming from nowhere to become one of the top five consumer electronics products already.

Healthy demand for MEMS in developing countries is also adding to growth, with China now a significant market for MEMS. Demand for automotive sensors in China has in fact surpassed that in Japan, although Japan, though new Japanese plans to mandate electronic stability controls will boost demand there as well. ”One thing that surprised me the most is that the telecommunications market is growing again,” said Bouchaud. “The overcapacity is soaked up and demand for telecommunications infrastructure is growing again, especially in China, which has led to the recovery of the optical MEMS market.” Bouchaud sees the post-recession rebound slowing next year, but figures the sector will maintain a stable 12 percent CAGR 2009-2014, to reach $10B in 2014. “There may be a bit of overheating in the automotive market, so next year automotive growth may be only 2 percent, but it will be a soft landing,” he notes, also suggesting a coming slowdown in the gaming market for gyroscopes. iSuppli was recently acquired by the $1 billion market research conglomerate IHS, now adding semiconductor coverage to its big automotive and energy industry data and general economic forecasting services.

Much of MEMS market growth going forward will come from new devices now coming out of development, and from growing adoption of biomedical applications. Source: Yole Developpement

The two market research firms define MEMS somewhat differently at the margins, so iSuppli typically reports smaller markets than does Yole, primarily because Yole includes the large market for microfluidics made on polymer and glass and on silicon, as well as magnetometers and some other minor devices. Yole also takes a somewhat more optimistic view of future markets for gyroscopes, timing devices, and particularly emerging biomedical applications. But both companies note there’s plenty of room for a difference of opinion in the imprecise art of forecasting future markets for new devices just coming out of development and new applications still in the works. “In 2014, a third of the MEMS market will be for devices that didn’t exist a few years ago,” notes Bouchaud.

Growth in the MEMS market will come from new devices and more inroads into high-value applications. Source: iSuppli

Maturing Market Faces Structural Change

MEMS’ move into new fast-growing, high-volume consumer applications is changing the market structure, favoring those who can get to volumes on 8-inch lines to drive down costs to gain a dominant share of the market. “Motion sensing is becoming more complex, but prices are plunging — it’s a nightmare,” said Eloy. This means a few leading companies dominate each market — STMicroelectronics with 50 percent share of consumer accelerometers, InvenSense and ST dominating consumer gyroscopes — leaving some 50 other inertial sensor suppliers scrambling after their own niches, and consolidation likely in the offing.

And now that MEMS sensors have become a volume market, lots of other players see opportunity to muscle in to capture some of that value as well — from CMOS foundries, to analog and mixed signal chip makers, to wireless telecommunications systems makers, to conventional sensor makers, all increasingly interested in adding MEMS capability, whether buying and integrating MEMS devices or adding their own MEMS technologies. “For the next two years the biggest issues will be consolidation, and mergers and acquisitions,” predicted Eloy, as other semiconductor and system companies look to acquire MEMS capability.

The seven sensors in the iPhone already account for more than $20 of the bill of materials —more value and more silicon real estate than the processor, pointed out Vida Ilderem, VP of Intel Labs and director of the integration platform research lab. And that gets Intel’s attention. She figures the average mobile platform will have 16 sensors by 2015, and will need plenty of processing power, firmware and software to integrate, analyze and act on that data. “Analyzing is the secret sauce,” she said, noting the current paradigm shift from delivering applications to delivering experiences for the consumer, where sensors understand the world and interact with the person, to deliver the right information or action.

MEMS makers aim to provide more of this analyzing to distinguish their components with systems software and firmware, but so do their customers and some other middlemen. CSR Group’s SiRF Technology figures it is well placed to do the sensor integration as the provider of the GPS chipsets. “We expect to be the gateway for integrating sensors,” argued senior marketing director Greg Turetzkey. Samsung plans to do its own application development for its mobile devices. “Samsung requests only the device driver with the MEMS. We try hard to understand the function and then do our own application development,” said Jungkee Lee, principal engineer, director of telecommunication module lab, module solution team, at Samsung.

Manufacturing Issues: Drive to More Integrated Packaging and Processing

The drive to smaller, lower-cost devices, selling integrated functions instead of just separate components, means a transition to new types of tighter integration beyond the typical wirebonding of separate MEMS and ASIC chips in a single package. Typical for MEMS, a diversity of new approaches is developing. Some are building the MEMS device on top of the IC, for the required pixel-level control needed for array devices like microbolometers and micro mirrors, or to eliminate parasitics for better signals with the shorter connections for timing devices, notes Yole. Others are turning to 3D TSVs, with or without interposers, and with some new lower-cost technologies. Still others look to wafer-level thin-film capping, or to bonding the MEMS and ASIC wafers directly together with metal-to-metal bonds that make the electrical connection as well as the hermetic seal. Offering more complex functions at lower cost also increasingly means making different MEMS sensors with a common process so they can be combined on the same wafer or the same die.

Less Overcapacity than It Appears

All those CMOS foundries eyeing high-volume MEMS production, added to all the 8-inch wafer fab capacity already in place for making what are usually very tiny devices for relatively small markets, may look like a great surplus of manufacturing capacity, but it’s not as much of an issue as it may seem. Eloy pointed out that what looks like overcapacity is business as usual for this sector. “MEMS foundries are really contract manufacturers, not foundries,” he noted. “They help their customers with developments, and most of them are profitable even if they have less than 70 percent load. They don’t need 90 percent filling of the fab. Some excess in capacity is just the way it is.” So far the CMOS foundries have captured only a very small share of MEMS production, but the high-volume consumer demand driven by the smart phone may finally be changing that. With the entrance of IC foundries to the MEMS business, the specialty MEMS foundries are migrating to higher-value products, points out Bouchaud.

For more information on MEMS, please contact Deborah Geiger at dgeiger@semi.org. If you would like to get involved in MEMS standards activities, please contact Paul Trio at ptrio@semi.org. To learn more about MEMS, join the Extreme MEMS Online Community.

December 7, 2010