As machine learning “brains” get better at understanding the world, the numbers of sensors that feed information to those “brains” will increase dramatically. It’s probably a good idea to own some sensor companies for the next decade or so. One of the leading sensor companies is AMS AG. The company has guided for 2019 revenue of $2.7b and 30% adjusted EBIT margins. That works out to about CHF7 of EPS or a ~15X P/E at the current stock quote. I think that is a reasonable valuation for a company that will continue to grow earnings far beyond 2019.
Let’s get the big risk out of the way: Apple is ~50% of sales, so if you think Apple eventually nukes AMS like they did Imagination and Dialog, then obviously this stock is not for you. I don’t think Apple does since AMS’s expertise is in optics, light, and sensors and not in silicon design. In fact, Apple is funding 25% of AMS’s planned $600mm of 2018 capex. In any case, I could be wrong, so exercise your own judgment here.
AMS’s revenue doubled YoY to $1.2b in 2017 with the introduction of the iPhone X. AMS provides the color sensor module and dot projector optics that power Apple’s FaceID and TrueTone features. These technologies will proliferate throughout Apple’s portfolio in the coming years. In addition, the entire Android ecosystem is also racing to add a FaceID-like feature and AMS is ready to sell them what they need. AMS recently announced a 3D sensing design win with a large Asian smartphone OEM. The bridge to AMS’s $2.7b of 2019 revenue is driven primarily by 3D, color, and spectral sensing spreading into hundred of millions of smartphones.
Post-2019, there should be more waves of growth as AMS has a world of opportunity outside of smartphones. This second wave of growth will be 3D sensing and LIDAR for cars. Beyond smartphones and cars, it’s ridiculously easy to think of new applications for machine learning + sensing. My microwave has like 30 buttons. Why? Why can’t I just put yesterday’s taco in the microwave and have it warmed perfectly because the microwave can sense it’s a taco and measure its mass? I’m going to tweet this at Elon and hopefully Tesla will build me a taco-sensing microwave. I’m being silly with my taco example, but machine learning + sensing is the substrate on which thousands of applications we cannot even imagine today will be built.
Management Strategy & Competitive Advantages
The current CEO took over in early 2016 and has done an excellent job of focusing AMS on four sensing areas: optical, imaging, environmental, and audio. Some businesses were sold off because they did not fit (NFC, RFID, LED). Some businesses were acquired (Heptagon, Princeton) to fill in gaps in AMS’s sensing portfolio. AMS now has a leading position in all four sensor areas and a product breadth that no single competitor can match. With all the pieces now in place, AMS expects its addressable markets to grow from ~$5b in 2017 to ~$14b in 2021.
AMS groups its end markets into two segments: consumer & communications (C&C) and automotive, industrial, and medical (AIM). Revenue was split 50/50 between C&C and AIM in 2016, but the mix this year will be around 70% C&C and 30% AIM because of the giant Apple ramp. AMS targets an eventual mix of 60% C&C and 40% AIM. I mention this because selling electronic components into AIM applications is an inherently attractive business. Product lifecycles are long and design wins are sticky. Look at the analog semiconductors companies as proof of this statement.
On the other hand, consumer applications are fast cycle and suppliers must compete for sockets in each new generation of product. However, it’s not impossible to build a good business serving consumer applications. Examples include Broadcom and Qualcomm. These companies possess a broad product/technology portfolio that they can integrate into complete solutions for their customers, thus enabling performance that can only come from system-level integration and faster time to market for their customers. I think AMS has an advantage in sensing along two dimensions: vertical integration and product breadth.
I’ll use the supply chain of Apple’s FaceID module as an example of how these advantages could play out. To begin the authentication process, a VCSEL (vertical-cavity surface emitting laser) shoots out lasers that go through some optics and then forms a dot grid on your face. The VCSEL was likely made by Lumentum, who buys epi-wafers from IQE or VPEC and then contracts out wafer production to Win Semi. Optical elements were made by Himax and AMS. Then the entire dot projector module was put together by LG Innotek or Sharp. This complexity causedproduction difficulties for Apple.
Now if you’re an Android phone maker, AMS can save you from this supply chain headache and provide you with a working solution that accelerates your time to market. AMS can internally produce the VCSEL and wafer-level optics, then package everything into a fully-tuned dot projection and IR camera system with software algorithms for face authentication. No other supplier can do this. Huawei, now the third largest smartphone manufacturer, is rumored to be AMS’s 3D sensing design win.
AMS’s broad capabilities also allow it to build unique monolithic sensors that integrate many different sensing abilities. For example, AMS offers a single sensor that integrates a MEMS microphone, temperature sensor, and pressure sensor. Instead of separate holes for each sensor, all three functions can use one hole in the smartphone case. The ASP of a combined microphone/temperature/pressure sensor is 3-3.5X that of a microphone. This is just one example. No other company can match AMS’s broad portfolio and offer the same level of monolithic integration to create value for customers.
In summary, if I was stranded on an island for the next decade, I would feel comfortable owning AMS. The company can use its advantages to win designs in AIM applications, which are long cycle and sticky, and thus have inherently attractive economics. In short cycle consumer applications, I think AMS can be the Broadcom or Qualcomm of sensing with the ability to provide complete solutions as a competitive advantage.
I think AMS has a good line of sight to its $2.7b 2019 guidance based on the adoption of 3D sensing in smartphones. The next wave of growth will start in 2020/21, with the adoption of 3D sensing and LIDAR in automobiles. Whereas the smartphone opportunity is perhaps $4-$5 for components or $10 for full modules per phone, the opportunity in automobiles is several hundred dollars per car. I haven’t touched on other blue-sky opportunities like autonomous industrial robots or color matching or home sensing, but it’s crazy when you think of all the possibilities. Even more exciting are the applications and opportunities that will be dreamed up by smart people over the next few decades.
Machine learning is an enabling technology, like the relational database, that will spread throughout the world. I don’t think we fully understand the implications of computers that can understand the world through vision and sound yet. I expect AMS’s earnings to grow for a very long time and the stock is currently available for 15X 2019 earnings.
I’m going to warm up my taco now.
I do not hold a position with the issuer such as employment, directorship, or consultancy. I and/or others I advise hold a material investment in the issuer's securities.