After three relatively flat years in 2007 to 2009, final figures indicate total MEMS sales jumped some 25% in 2010, to $8.6 billion, thanks largely to the surprisingly fast recovery of the automotive industry and its inventory restocking, as well as the boom in smart phones and tablets with motion sensing. But stronger than anticipated sales of optical MEMS for both telecommunications and consumer applications, and of inertial and RF devices for industrial applications also contributed to the outsize growth. Though growth will now settle down to more sustainable post-recovery rates, we now expect continued stronger demand across these multiple sectors to drive 14% compound annual growth for the MEMS industry for the next five years, to reach a $19.5 billion market by 2016. All these figures are updated and increased from our forecast last fall, based on our current soundings of the industry.
What will the industry look like in five to six years out? We expect it to be dominated of course by a $9 billion consumer device market, but the industrial market will outpace the automotive market to become a $2 billion opportunity, and biomedical applications will see the strongest growth of all, becoming a $5 billion market. On a unit basis, however, MEMS will become even more overwhelmingly a consumer business, with the predominantly low cost consumer devices increasing to more than 80% of all packaged devices in 2016, driving unit growth by 24% CAGR to 15.8 billion units.
Volume consumer markets, integrated systems mean new business models
High volume consumer markets favor large companies who can ramp high volume production to drive down costs, likely gradually changing the structure of the highly fragmented MEMS business to fewer, larger suppliers.
The maturing business does also means there’s now room for companies to make a decent business by specializing in just one part of the value chain, like fabbing the die, or doing just the packaging and sales, or even selling just the motion processing software. Infineon, for example, is selling undiced MEMS microphone wafers to customers who dice and integrate it with the ASIC, and then sell to their own established customer base.
But growing demand for integrating multiple different components into a single system is likely to mean companies will want to fab more of their own devices. While the strong sales of 3-axis gyros and electronic compasses for smart mobile gear so far have been mainly for separate packaged devices, there’s growing demand to integrate the two parts, and first products are now appearing from the likes of STMicroelectronics. Besides the usual reductions of size and cost, putting the two devices in the same package means they can cross calibrate each other, with each making up for the weaknesses of the other sensor to increase the accuracy of both. While most accelerometer companies will start by buying the compass from another supplier and packaging them together, the trend will be for the same company to make both parts, just as the accelerometer suppliers are moving to make their own gyroscopes in house to control the whole family of products. The same trend will move across other products as well, as the large suppliers aim to increase volumes by controlling more of the component portfolio, gradually displacing the merchant suppliers.
Blockbuster automotive growth will level off for now
The stronger than anticipated recovery of the automotive market, particularly in developing markets, and the restocking of depleted inventories from the recession, drove 15% in the automotive market in 2010, to reach the $1.5 billion level. Demand for tire pressure measurement systems continues to grow, and safety systems with accelerometers are becoming compulsory worldwide, but there are few new applications to drive demand going forward. So we expect 2011 to see a leveling off, with perhaps $100 million in new business, for ~6%-7% growth. But new applications will drive a new round of growth come 2013-2014, as microbolometers and other emerging products start to reach significant volumes.
While the industry is mainly using 1 axis gyros now, 2 axis products are in validation, and the industry is debating whether to move to multi-axis combo sensors, using fewer but more sophisticated sensors to feed signals for multiple systems into one central control unit, to reduce the cost of wiring. But there seems little industry agreement on what best to combine with what.
Cycles of innovation will continue to drive growth
Our growth forecast is based on firm belief that MEMS innovation will continue to drive new growth markets, and on the innovations we see now in development. From the beginning, the development of the MEMS industry has always followed the same pattern. In the highly fragmented MEMS market, every two to three years a device reaches levels of price and performance that meet a real market need, and it then booms into a big business. First the MEMS accelerometer took over the airbag business, then the MEMS gyroscope created the electronic stability control market, then the inkjet head enabled the consumer printer business. Now it’s the booming demand for the 3-axis gyro to bring motion sensing to handsets that is driving much consumer market growth.
These new MEMS products are always linked to the invention of a new function, and creative new innovation is very hard to predict, though we spend a lot of time on this issue. Some of the really crazy things really can happen. Early on, looking at the microphone business, we concluded that MEMS could not match the quality of the electret microphone. But then Knowles Electronics–who was trying to make a sensor for its hearing aids—came up with a thin MEMS microphone that wasn’t sensitive enough for hearing aids, but that could be soldered for easy assembly into handsets, and it turned out to be about the only successful MEMS device ever introduced directly into the consumer market.
What will be the next new MEMS function to jump to volume by 2013-2014? We think there are multiple good prospects. MEMS oscillators provide a new function and demand is really growing. MEMS speakers are another example of a completely new function, though are still some ways from commercialization. Another likely prospect is the pico projector—it will happen at some point. The MEMS RF switch remains one of the likely competing solutions to handset customers’ need for tunable filters and antennas, despite the time it has taken to get into high yielding volume production. Also close to the takeoff point is the microbolometer, where the existing military and industrial business can help drive the volumes and costs for wider adoption in the automotive and consumer markets.
While the accumulated experience, developed technology and established infrastructure of the maturing MEMS sector is now starting to mean it’s taking less and less time and money to develop a new MEMS device, introducing any new technology to high volume, low cost consumer markets is extremely tough, especially for startup companies, which have driven much of the MEMS sector’s recent innovation. Companies aiming at the consumer market will need a completely new function, or a really disruptive technology.