Read Yole Développement (Yole)’s interview with InvenSense CEO Behrooz Abdi, and dive deep into the future of sensing.
Soon after InvenSense released its latest financial results, Yole Développement had an opportunity to interview Behrooz Abdi, InvenSense CEO. Under Mr. Abdi’s leadership, this leading global provider of MEMS sensor platforms targeting mobile, wearables, smart home, industrial, and automotive products is ranked #8 in our latest TOP 30 MEMS players ranking, with 2015 revenue of $438M.
As highlighted in Yole’s latest report, Status of the MEMS Industry 2016, the inertial MEMS business represents $3B of the $13B global MEMS market, and many opportunities are still emerging for established as well as new applications (mobile, IoT, VR/AR, drones, etc.).
With MEMS competition fiercer than ever, hardware and software innovation, use-cases, and diversification becoming essential to MEMS players’ survival, we asked Mr. Abdi for insights into this very challenging, very promising business.
Yole Développement (YD): Can you give us a quick overview of InvenSense in terms of its background, technology, portfolio, and latest evolutions?
Behrooz Abdi (BA): InvenSense was founded in 2003 and built on the premise of assembling a technology platform for MEMS that would be scalable across very different products. The first product we started with was the inertial sensor: the gyro and the integration of gyro and accelerometer. Over the last few years we have extended the technology to other products such as microphones and environmental sensors. Also, in the background, we’re doing some research on pressure sensors and we’ve been pretty public about it, as well as the fingerprint sensor, which is a team-up device.
I realized that what I really wanted to do was to build a sensor system on chips. It’s a pocket size system because we think it’s an ecosystem… it requires an ecosystem to be put together. It’s a new market that will grow for the next few years if not decades, and we think that the internet of things really needs an internet of sensors. This means that every device will have some kind of sensor or multiple sensors. Really, every device around us will be “multi-sensory”, and there are many, many examples.
One thing we don’t do is RF MEMS. We focus on sensors, and to do that we also deal not only with the sensor itself, but micro-processing, integration, and micro-controlling, as well as software (a lot of software) to make it easy for people to adopt sensors and design them. So that’s how we’ve been involved in the last 2 - 3 years, and we’re starting to see the results in a number of different use-cases. When I came to the company, the question was, “What’s the use for? What’s the best use-case?” For example, for gyroscope it’s image stabilization and navigation, so we’re doing a lot of stuff around it and creating devices that are as fast as possible.
As you know, the technology itself is built and bonded with proper technology and it’s very elegant, and we bond MEMS and CMOS together. And by doing that the technology not only improves and gives us an advantage, but it also really transitions the business to a fabless model and that’s a significant advantage for us in our market. We’re not depending on packaging as much as our competitors because everything is within the chip. So we push all the technology, all the innovation into the chip. That’s InvenSense’s background.
YD: As you mentioned, the goal is to have a fabless business model that will help you to create a flexible link to the adaptation of volumes and the needs of your customers. Also, our understanding of what you have built is that you will be able to include in your products not only the sensor but sensors with adaptive algorithm software functions that can help your customers integrate these kinds of complex devices much faster compared to a normal IC. In this sense, InvenSense has really been a pioneer in pushing software to the sensor, to your customer, and you have strongly leveraged your product’s added value and the sensor, thus enabling your customers to move into volume production faster.
BA: Exactly. My first mission is to increase sensor adoption in devices, and number two is to compete with my competitors. Because if I can grow the market, the competition will also benefit to the point where they won’t have to resort to price erosion as much as they normally do. And for example, with EIS (electronic image stabilization), we want to enable gyro adoption in mid-tier phones. As you might know, a lot of phones have adopted gyro for various different use-cases, including OIS and activity recognition, navigation, things like that. However, mid-tier phones and smartphones haven’t really adopted gyros. So the thing we started working on a few years ago was to really develop software for EIS. Then, the value proposition to the customer is that with very little incremental cost (because if you’re already using accelerometer, gyro is not that high of a cost), by adding it you can offer a camcorder function, a stable camcorder in a mid-tier phone. So that was one of the things that really prompted us to drive EIS and really develop the software, and it enables somewhere around 500M additional units of volume, making it valuable to us and our competitors.
Courtesy of InvenSense
One very positive thing happening right now is the augmented reality game Pokémon Go, which has really exploded in the US. Pokémon Go basically fuses augmented reality and uses camera, gyro, display, and navigation, so you are using your GPS inertial navigation to look for these characters around you. You go out and look for them, and then as you find them you point at them and the display camera turns on and you can see the character superimposed on top of a real scene, and then you capture them.
Now, the significance of this is that you really can’t do inertial navigation without gyro. For example, when the blogs talked about it, some mentioned that if your phone doesn't have a gyro or doesn’t have a good gyro, you need to make sure you walk in a straight line and don’t move around, because there’s no gyro. Ideally, the phone uses the magnetometer and gyroscope to really superimpose the character and tap into the camera.
Some things that come along, i.e. Pokémon Go, really drive faster adoption because they’re visible to the consumer. So that really is the premise of InvenSense: “Let’s make it easy, let’s create software and solutions that really are visible to the end-consumer”. And the two things we decided to start on with the inertial navigation were image stabilization and navigation, because both of these - if you can bring a better mapping experience to the consumer, then they will demand a certain product from OEMs that allows your OEM to differentiate. That’s the kind of the thing we focus on. For example, we announced earlier this year that our navigation software was adopted in some of the Huawei phones, and now adoption has increased because it brings a better experience inside tunnels, urban canyons, places where you don’t get a GPS signal. That’s really the essence of InvenSense: “How do we drive that?” Because the company’s vision and mission is to become the “sensor solution company”, which means we want to deliver sensor solutions to a wide range of markets.
YD: In a previous interview with you, we drew a parallel with the image sensor field, the image sensor and camera phone manufacturers being able to increase the price year after year because they were delivering value to the customer, because it was clear that it was a better picture, or a better movie…the ability to have really interesting functions integrated with the camera module, to use effects on the pictures they were taking, and so on. But the ability of the consumer to say “Well, I have a very good gyro on my phone, or I have a very bad one”...no one was able to make such an evaluation of the quality of the gyro they were using. So there was limited value linked to gyro quality. But now, with Pokémon Go and a few other applications, you’re able to demonstrate the quality of your initial measurement unit. It’s a big change in the industry, no?
BA: Exactly, we think the same thing. That’s why, for example, a lot of people ask, “If the 6-axis gyro is a commodity, why was it so price-competitive?” and I tell them, “People didn’t know what to do with it.” But now, with image stabilization for example, we have a product that we just announced, the all-in-one 6-axis, that actually brings OIS into the main board using the main 6-axis, and the reason we’ve done that is even if you could potentially eliminate two axes in the OIS module, we think there will be a lot of different camera phones, with dual camera, single camera, front OIS, back OIS, etc. We think there will be many different architectures, and if you want to really get a better picture it’s OIS, and if not you can still have OIS but it might not be as good as a dedicated two-axis. So for the customer it isn’t necessarily about axis or architecture; we talk to them in the language of a photographer. We say, “We can give you so many f/stops of aperture control”. So when we talk to these OEMs, their engineers, we talk about a camera phone that really enables better photography. Or when we talk about EIS for a camcorder, because that’s for frame to frame…so it really drives the company to discuss every opinion on how to reach the end-market and something that the consumer will recognize. So that’s what we’ve done, and then we look at these technologies and the reason why we picked inertial navigation, AR, and imaging is that they are multimarket and address mobile phones.
One of your questions is, “Is it time to go beyond mobile?” and the answer is “Definitely”. We absolutely need mobile to grow because it’s really hard to compete with mobile volumes, but also we need other markets for stability, and we think that technology like image stabilization is something we’re already seeing, for example being applied to drones, and in the future even applied to automotive via various cameras in the car. The same thing with navigation, which we’ve done with the phone; a solution with our 6-axis inertial sensor is getting lots of interest in the automotive market today, because navigation will become more interesting for location, for positioning. So we think these are technologies that can cut across multiple markets and you need both: mobile and other markets.
YD: Okay, we see the emergence of AR and VR. The first products on the market are just arriving, so it’s difficult to know whether they will have a real impact, but do you have a specific project and/or product for VR goggles, or VR systems like the different ones supported by Microsoft, Google, Facebook, and so on?
BA: Yes, absolutely. The way I look at virtual reality is that I think it will start with gaming, 3D gaming, virtual reality gaming headsets…the unit volume that people are forecasting is 30 - 50M in the next four years. You’ll have the headset, a controller, and other accessories - swords for example, or wheels. The other thing is that in VR today, at a minimum you must have arms and legs, so how do you detect these things in the virtual world? There are a couple of ways: either with a sensor where you can see your hands and fingers, or something that goes on your wrist or fingers, like gloves or other things with sensors that actually the reinforce the impression of being in a virtual world. So the bottom line is, it’s not just a virtual reality headset, but a bunch of accessories that are also very interesting. So even though the VR market itself will be big, all of these accessories could be well north of a 100M units with hardware and software content, and so we’ve been focusing on both. We’re talking with our designers now, and we have designs for inertial sensors and microphones, and we think we can get a lot more content (sensors and software) to these headsets, so that’s definitely a focus market for us.
YD: Shifting more to your company’s revenue and the revenue evolution, the last quarter saw a decrease of your company’s sales, and we really have the feeling that mobile phone sales have decreased, and also that there’s been a strong increase (at least in percentage) in IoT, drones, and other segments you’re targeting. Can you provide us with more data points linked to this?
BA: We think that the cell phone market will still be pretty interesting because if you look at the total cell phones, it’s well over 1.3B - 1.5B units, and today InvenSense is in roughly half of them. Our market share is 50%, which means there’s a lot of opportunities within the smartphone (market) to grow because not all smartphones have adopted gyro, for example. So the smartphone market remains very interesting for us, but unfortunately this year everybody is suffering from the maturation of the smartphone’s top end, so the first thing that we ask is how can we drive adoption of the inertial sensor? Beyond that, we know that the smartphone market has certain dynamics: 1) it is very competitive, and 2) it fluctuates. So it’s not predictable at all. What we need is a broad-based market of IoT - anything that moves or listens or requires a certain sound or inertial input. This includes drones, which have been a great market for us. Robotics and robots are also very interesting and growing, just like VR headsets. Gaming is also what put us on the map.
Beyond that, we are still looking to smartphones and other devices…a lot of them, like smart home devices such as Amazon Echo. You know, that’s one way you’re going to communicate with devices around you, and that will see plenty of adoption. And then as we look further out, automotive is a very sensor-rich environment. Automotive safety, both inside and outside the cabin, requires a large volume of smart sensors, especially as we move towards ADAS and autonomous driving. We think that performance and reliability should be very, very high with inertial sensors, but we think that a lot of other sensors will enter. So the percentage of the IoT segment has grown, and we are going to go there in the next few years, but at the same time it takes longer. There are many companies that are purely on IoT and they have good margins, platforms, and lots of customers…the challenge that they have is growth. For us, growth will come from mobile phones and IoT’s continuous build-up. As I said in the beginning, we want to be more broad-based in the market with our sensors, but really the solution part of it is very important; creating software that is very close to the sensors we’re developing. So we really focus on use-cases because that’s what matters, that’s what customers want to know. “What does this sensor do for me?”
YD: What you’re saying is that the 1st quarter of 2016 decreased compared to the last quarter. Is that just life in the mobile phone industry, with better business returning in the coming quarters?
BA: Yes, we think that this year was tough for everybody in this market. We see everyone being challenged. We also think that a bright future is ahead of us, with new applications requiring lots of sensors, so this is the kind of thing we can focus on as we move forward. You know, “How do we go to other markets?” and, “How do we create a smarter, richer world of sensors?”
YD: One year ago you announced development of a fingerprint sensor with ultrasonic technology, and in your financial communication you’re including it in your serviceable market evaluation for 2020. What’s the development status? Are you able to tell us the status of this fingerprint sensor? Is it something that is close to reaching the market?
BA: We started researching this ultrasonic sensor three years ago. We looked at it because we thought it was really a perfect match for our technology and platform, if you look at the value of CMOS and MEMS together. If you do it, you don’t want to have a lot of interconnects between the CMOS and the MEMS. So we thought this is really a good match for our technology. Secondly, we think ultrasonic has many different applications: object or gesture recognition, collision avoidance in drones and things like that…so really, fingerprint sensing is one of them. It has, for example, interesting applications in the medical field. And obviously today, ultrasound is used pervasively, so we think ultrasonic has a great future in terms of market applicability. To get more practical, we need to get the revenue, and that’s the reason why we selected ultrasonic fingerprint as the first market, because we think that cellphones and IoT fingerprint-authentication are now becoming mainstream and will be here for a while. Especially if you can make it work under thick glass to avoid dust, or in special buttons, things like that. It’s really great…and then long-term, the ability to potentially go beyond the outer layer of the skin, into the dermis layer, to determine your position, or a mobile payment in terms of being hack-proof. So these are the things that prompted us to focus on the fingerprint sensor, and that’s how our internal technology development was shared with OEMs. We have a lot of excited OEMs waiting for this technology, but we’re selectively working with a couple of them to productize this in the next year. So the goal is to get something out - a product sample, and then have the OEMs with a phone in the market by next calendar year, or maybe early in 2018.
YD: Ok, so within 1 year, 1.5 years, it’s something that will be on the market?
YD: And you’ll deliver both the device (the fingerprint sensor) and the integrated software in order to deliver more value for your customer?
BA: To improve our time-to-market, we’re working with a lot of third-party software companies, so the amount of software that we’re doing is not as pervasive. Because the thing is that this is a more mature ecosystem we’re working with.
YD: Regarding financial communication: we do not think you have disclosed the share between inertial sensor and microphones. Is this something you are disclosing in terms of information?
BA: We haven’t disclosed it, but in terms of total market, in terms of inertial sensor for consumer mobile, we believe we’re well over 60% (Note: Yole’s estimate is 55% for 6A-IMUs), so that’s what we’ve shared. In microphones our share is very small, really meaningless at this point.
We’ve been focused on higher performance and on IoT. We really want a solid fabless support chain ecosystem, and so we’ve done that. The market takes time, so what I said on the call last week was that it’s been growing by double-digits every quarter, in terms of percentage. But in terms of percentage of our revenue, it’s still in the single-digits. My goal is to cross the double-digit percentage of revenue. There’s growth, but we still need a few things to make that a reality. There’s definitely something going on in IoT, but we haven’t really done much with the mobile microphone, though we think that in terms of mobile phones and differentiating the image sensor and designing a better camera, there are definitely performance requirements that apply to us at least. The next level in innovation will be in audio, with audio beam-forming, and this will require a higher SNR ratio and better microphone performance, and that’s where we think we’ll have our opportunity. We’re not going after the mainstream microphone market.
So to sum up, in terms of share, that’s what we disclosed.
YD: We have seen very strong interest from the industry and consumer electronics manufacturers (mobile phones and others) concerning gas sensing, particle detection, and a few other new sensors that could be quite rapidly integrated in mobile phones. Is InvenSense also looking at this?
BA: Yes, it’s another area we’ve started to research. There hasn’t been much progress; we’re still under wraps. But we think our technology definitely applies. Gas sensing will take a little while to develop further. With microphones you sense sound. With inertial sensors, you sense motion. With gas…the problem is, what gas do you sense? And that’s one thing that in the sense of market development people are still experimenting with. Some people might choose particle detection, some people might choose hazardous gasses, other people might want to pursue other segments. I think the market and revenue will be a little too fragmented for a while, but we’ll be ready for it when it comes in, because we definitely have been doing some things with the technology that we think are real game-changers. We think that the market will be longer-term, but the short answer is “yes”.
YD: Are there other device applications you’re looking at in terms of long/medium-term?
BA: We’ve definitely been looking at the pressure sensor, and we’ve tried to drive the attach rate. This means that it has been in development for a while and what we’ve been looking at is, is this something that will be used in drones, or eventually in mobile phones? I always look at pressure sensors and environmental sensors as an inertial sensor. I think it’s more relevant as an inertial sensor; nobody understands or cares about the pressure sensor, but they really care about the location. So we’ve been working on that and we think we’re pretty close in terms of having the right performance, but then again the market has been slow to adopt pressure sensors, so that’s an area we’re working on. Beyond that there are a lot of other sensors we think MEMS will apply to, but as a small company we can only choose certain things, so we’ve been holding off on these because we think the inertial sensor itself has huge, huge potential in mobile with adoption and new use-cases (IoT followed by automotive), so we try to stay focused where we can.
There is one thing we have done in the past: we have looked at licensing the technology to other people who want to develop their own MEMS technology. One of the things that I see in the MEMS industry, the challenge with it is that you have the same challenges of semiconductors, but worse. If you want to have a semiconductor startup nowadays, you have to develop the chip. If you’re on Moore’s law, it gets very, very expensive. So as a small company, it gets very hard to scale. MEMS is the same thing, except that you also have to develop your own technology and operations, and you must convince the fabs to work with you, and then you have to develop the chip, and the software, and the use-cases. And that’s why we don’t see a lot of small companies scale. We know we won’t have every sensor, so we’re going to license the technology and create an ecosystem of innovation, and it’s taking us longer than we wanted. We’ve done it with some small companies, we haven’t done it as much, but we now think we have the tools because we have the base technology that companies can use to develop their own sensors, and we also have our Firefly platform which is a smart sensor, and we have tools designed like Sensor Studio. We have the relationship with TSMC and GlobalFoundries, and we have the market. We have IoT and mobile markets. So we think we can create a very synergistic ecosystem and help startups really focus on what they’re good at, which is building a sensor. That’s something we’ve been working on and we’ve got some really good traction with the companies we’ve been talking to. That’s something you’ll see us become more focused on in the next year.
YD: So you are still thinking of licensing your technology to other players?
BA: Yes, we’re very open to that. The reason we’ve been slow to develop is that when we started, we did license it to a few people, but what we came to realize is it wasn’t enough because we really had to put a lot more support behind it in terms of the technology, and by the time they did all that, we still had to have some kind of operational support for them. We can’t just license without putting any infrastructure together.
We’ve had big companies approach us, and then, for example two years ago, we announced a deal with a company called Sonion in Europe for a microphone where we developed the technology for them and they really stepped in and helped develop the technology and the product, because that’s not a focus market for us. But our technology is high-performance enough to apply to that market, and it’s been a great collaboration. So these are the kinds of models we’ve explored in terms of partnership. For example, for automotive safety, we haven’t announced who our partner is, but we’re partnering with someone in the sensor area in automotive safety. We really intend to be scalable at the same time, as the industry scales with our tools and technology.
YD: My last question is a broad one: what will happen in the coming months/years for InvenSense? What do you wish for the company’s future?
BA: My tactical wish is that the mobile market will get excited and buy more phones, and that IoT will happen faster. But on a more serious note, my wish is that the use-case for sensors becomes better recognized by consumers. I think the industry needs innovation, and the only way the industry can innovate (meaning companies like myself and my competitors) is to have scale and R&D dollars. Simple as that. I think a lot of companies are experiencing that but it’s not happening fast enough, and that’s when you see companies suffering. We’re one of the few companies that are profitable in terms of sensors; many companies aren’t. My wish is for my competitors to start focusing on use-cases and driving sensor adoption instead of competing in existing markets, because that leads to price erosion and destroys markets, and that’s what my competitors are doing today. So we should work better as an industry, even amongst competitors, and drive market adoption.
YD: We totally agree with you. This is something we tried to push within the industry group, without much success yet, because the only way business can increase is to help the customer integrate more functions with the sensor. And it’s a collective action to do that; to really increase the total business.
BA: Exactly. And that’s why I get a lot of support for the amount of investment we’re making, and I tell myself, “Look, my gross margins are much better than the industry average”. If you really dig into the gross margins of my competitors, you can see that we’re much higher, and that really drives the value, the adoption, and brings the end-to-end value. And OEM customers tell me, they come to me and say, “Hey, I’m surprised we’re making this much money, because we really rely on your technology - you and your competitors’ technology - and we’re surprised you guys can have such low prices”. They don’t have a choice; there’s competition. It’s not like they’re here to give me more money. But at the same time they’re surprised that the industry is unhealthy, which everybody recognizes.
Getting back to what my wish would be, it’s that we can really drive adoption. So if I see my competitors talk about image stabilization, navigation, and other things that really bring that value to the customer, I’d rather have them invest in that rather than just price reduction, because if we don’t do that then we kill the industry.
YD: We completely agree. Is there anything we didn’t mention that is really important to you?
BA: No, I think we have covered the key discussion points. I think that the main point of our company is…in one of your questions you talked about packaging. Packaging is important, but for us it is not as critical as everybody else because the fundamental part of our technology has so much value in itself and doesn’t rely on packaging as much as other people. So even though in some products we rely on great packaging, our focus has been on the bonding technology, and as we’ve done that we’ve been able to integrate more functionalities, such as MCUs and memories, and that really makes the vision of the smart sensor a reality.
Behrooz Abdi - President and CEO
Abdi, a member of InvenSense’s board of directors since June 2011, was appointed President and CEO in October 2012. Most recently, he served as Executive Vice President and General Manager of NetLogic Microsystems, Inc., a leader in intelligent semiconductor solutions for the Internet. Before this, he served as the President and Chief Executive Officer of RMI Corporation (also known as Raza Microelectronics Inc.), a fabless semiconductor company, from November 2007 to October 2009. He served as Senior Vice President and General Manager of CDMA Technologies (QCT) at Qualcomm, Inc., a provider of wireless technology and services, from March 2004 to November 2007. Prior to joining Qualcomm, he held leadership and engineering positions at Motorola, Inc. where he served as Vice President and General Manager for Motorola’s radio products division, in charge of RF and mixed-signal ICs for the wireless mobile market. He currently is a director at Tabula, Inc. and Exar Corporation. Abdi holds a BSEE from Montana State University and an MSEE from the Georgia Institute of Technology.
Amid a commodization paradox, Chinese players taking off, sales stabilization and new applications, how can the MEMS industry regenerate value?
Find more here
MEMS & Sensors Market: Current Challenges & Future Opportunities|
Discover the presentation held by Guillaume Girardin at InvenSense Developer Conference, in ...
Gas sensors market and technology trends|
Discover the presentation held by Eric Mounier at SEMICON Europe 2016
A new virtuous cycle for the MEMS industry - China focus|
Discover the presentation held by Claire Troadec at MSIG Asia, in Shanghai, in September 2016.