Glossaries 2

Definitions 3

Table of contents 8

Scope of the report 10

Report methodology 11

About the authors 13

Companies cited in this report 15

What we got right, what we got wrong 16

Who should be interested by this report? 17

3-page summary 23

Executive summary 27

Context 70

• Semiconductors in the time of COVID-19
• End-markets

Market forecasts 101

• Forecasts by end-markets
• Forecasts by device

Market trends 148

• New products
• Market trends by device
• Evolution of sensing?

Supply chain 182

• Player’s ranking
• Market shares
• Ecosystem dynamics

Technology trends 220

• Manufacturing trends
• Device technology trends
• New types of MEMS devices?
• From sensors to system

Outlooks 290

Appendix 298

• Manufacturing process flows
• Reverse Costing®: Structure, Process and Cost Analyses

A $2.1B MARKET OPPORTUNITY BY 2026 FOR FAST-GROWING APPLICATIONS, IN A $18B MEMS WORLD

Last year when COVID-19 erupted, the effects on various industries were felt immediately, which in turn had an after-effect on the MEMS business. But the crisis did not impact all end-markets the same way. The MEMS market relies heavily on consumer applications, which are 62% of the total market, and the automotive industry, which is 16% of the total market. We therefore expected that last year the MEMS market would suffer due to COVID-19’s negative effects on end-system shipments of smartphones and cars. But this was not the case. MEMS sensor sales recovered during H2 2020. Manufacturers resumed refilling their inventories after signs of stabilization and recovery in key end markets. In fact, the strong demand for consumer MEMS has completely offset the automotive industry slow-down. Therefore, the MEMS market was worth almost $12.1B in 2020, up almost 2% from the year earlier.

After a weak 2019 and 2020, we expect the market to grow by 11% in 2021, reaching $13.4B. After that, high-single digit growth will increase annual MEMS revenue to $18.2B by 2026. This promised growth could be delivered by new emerging technologies and opportunities, with a few notable ones such as:

• Microphones and inertial MEMS in wearables and particularly True Wireless Stereo (TWS) earbuds. This is driven by the need for better sound capture using microphones and voice activity detection (VAD), noise reduction with accelerometers for voice detection and bone conduction. 3D Audio is being pushed forward by Apple and will probably jump to Android devices, causing a surge in demand for Inertial Measurement Units (IMUs).
• Gas sensors and environmental combos, integrated in wearables and other devices. These help monitor indoor and outdoor air quality around users, as they become more concerned about the air they breathe, and especially due to COVID-19.
• Optical MEMS for Light Detection and Ranging (LiDAR) and Augmented and Virtual Reality (AR/VR). Revenue in this sector might still be weak in five years’ time. However, opportunities lie beyond this horizon, as the Advanced Driver Assistance System/Autonomous Vehicle (ADAS/AV) and AR/VR markets further develop.
• Piezoelectric Micromachined Ultrasonic Transducer (PMUT) devices, which are used in ultrasonic fingerprinting. PMUTs are also used as a replacement for physical buttons and haptics in smartphones and cars. Capacitive MUT (CMUT) devices are also very promising for the consumerization of low-cost ultrasound imaging, close to the Point-of-Care (PoC).
• MEMS microspeakers, which should show their advantages in TWS in-ear designs first, replacing older electrodynamic or balanced armature speakers.
• MEMS-based sensor-shift optical image stabilization (OIS). This which could replace the flexible Printed Circuit Board (PCB)-based sensor shift used for the first time in the iPhone 12 Pro’s camera module. It may also enter other camera modules in other handsets or consumer devices.

1. PLAYERS’ REVENUE GROWTH AND RANKING DYNAMICS

The pandemic, global lockdowns and the US-China trade war heavily impacted the supply chain of the semiconductor industry. A proper development strategy has therefore become the key to success. Some players profited from the crisis and some didn’t, leading to massive changes in our global MEMS ranking. Bosch, Broadcom, Qorvo, STMicroelectronics, Texas Instruments, Goermicro (Goertek), HP, Knowles, TDK and Infineon now comprise the top-10 with at least $6.5B combined revenues, more than half of the total market. However, the companies that increased their revenue were not necessarily the ones holding the biggest market shares.

• Companies related to COVID-19 prevention technologies, such as thermal imaging and sensing or pressure sensors, grew strongly last year. For example, Guide IR, FLIR Systems, Lynred, which sells microbolometers, and Melexis, which sells thermopiles, profited due to the increase demand for elevated body temperature measurement application while Amphenol saw its pressure MEMS grow due to demand for respiratory end-systems like ventilators and Continuous Positive Air Pressure (CPAP) machines.
• Broadcom and Qorvo grew strongly last year, providing highly reliable RF MEMS filters for 5G deployment, which was stronger than expected. SiTime is continuing to push strongly for replacement of traditional quartz timing solutions with its MEMS-based timing products.
• AAC and Goermicro (Goertek) profited from a fruitful MEMS microphone demand environment. For the first time in 15 years, the MEMS microphone leader Knowles has been replaced by Goermicro.
• Companies with business both in consumer and automotive like Bosch, STMicroelectronics and TDK, were able to offset potential negative effects from automotive due to the impressive consumer MEMS business and the pick-up in demand during H2 2020. Infineon has entered the top 10.

2. 

3. GOING FROM SENSORS TO SYSTEMS: MEMS PLAYERS CLIMBING THE VALUE CHAIN

4. MEMS market challenges are evolving. In the past, focus was on product shrinking, price decrease and volume increase. Today, the requirements are changing with use cases being more important while different sensors must fuse with software. Power consumption must fall at the system level even though significant data processing is needed, especially for always-on applications like audio/voice human-to-machine interfaces (HMI).
5. Normally, all the data coming from the sensors are processed. The trend is to process them nearer to the sensors. This allows low latency, more safety and more privacy, since no cloud computing is implicated. However, component value is tied to the data. The higher the position on the signal chain, the higher the value will be. We see a growing movement toward a mix of competencies across the supply chain, from front-end manufacturing to packaging, the modules and the integration of systems. The move to a system level approach and the integration of different components like MEMS, application specific integrated circuits (ASICs), antennas and power sources that use different materials and processes in the same housing, is creating needs for more sophisticated System-in-Package (SiP) technologies, with the goal to increase system level performance and decrease overall power consumption.
6. We have been seeing various movements from MEMS manufacturers going to a system level approach, either by internal development or by mergers and acquisitions. For example, Knowles started selling a stand-alone Digital Signal Processor (DSP) for the Google Pixel phone in 2019. STMicroelectronics has an IMU with Artificial Intelligence/Machine Learning (AI/ML) libraries in the embedded Microcontroller Unit (MCU). Bosch released an accelerometer with edge AI on an MCU and a gas sensor with AI for olfactometry and odor detection. TDK also released a new family of smart sensors. STMicroelectronics has acquired Cartesiam for its edge AI capabilities, to further enhance its existing inertial products that include AI/ML. Finally, mCube has acquired Kinduct for its cloud analytics capabilities, offering actionable insights to customers.
7. Finally, we are starting to witness the first signs that our devices could emulate and sense their environment with human-like capabilities, using the five best-known senses in a “phy-gital” meta-world. The final step for MEMS would be to evolve from simple deterministic data-collection sensors to more empathic data-interpretation machines. Devices could discern and predict user moods, feelings and emotions, responding and reacting to needs and intentions, much like some humans are empathic.
8. 

AAC, AKM, Alps Electric, Amphenol, Ams, Analog Devices, Apple, Arioso, Asia Pacific Microsystems, Aspinity,  Atomica (ex IMT), Audiopixels, Boehringer Ingelheim Microparts, Bosch, Broadcom, Butterfly Network, Canon, Cartesiam, Colibrys, Collins Aerospace (EX UTC), Cirrus Logic, CSEM, Denso, DRS, Epcos, Epson, First Sensor Technology, FLIR Systems, Formfactor, Fraunhofer IPMS, Fujifilm Dimatix, Gettop, Goermicro (Goertek), Google, Hanking Electronics, Hewlett Packard, Honeywell, Imec,  Infineon Technologies, Knowles Electronics,  Lynred, Maxim, Melexis, Memscap, Memsensing, Memsic, Micralyne, Murata, Nxp, Omron, On Semi, Panasonic, Qorvo, Qualcomm, Raytheon, RF360, Rohm, Samsung, Sensata, Sensirion, Si Time, Silex Microsystems, Silicon Sensing Systems, Sintef, SMI, Sony, STmicroelectronics, Syntiant,  Taiyo Yuden, TDK, TE Connectivity, Teledyne Dalsa, Texas Instruments, TowerSemi, VIS (Vanguard International Semiconductor), VTT, TSMC, UMC, Usound, X-Fab, Xmems, and more.