Good morning, everyone. My name is Parag Agrawal, and I am Vice President of Investor Relations and Corporate Development at ON Semiconductor. Let me be the first to welcome you to our 2019 Analyst Day. We know that you have very busy schedule and we appreciate your taking time to be here. We also extend a warm welcome to the people who are on the webcast.
Let me walk you through the agenda for the morning. We will kick off the presentations with Keith Jackson, who will provide a strategic overview of the company. And Keith will be followed by David Somo, who is SVP of Strategic Marketing and he will provide an in-depth view of our end markets. David Somo will be followed by the 3 business units, who will give an insight into how they address the various end markets. So we'll start with Vince Hopkins, who heads up our analog solutions group.
He will be followed by Tanner Ocelik, who heads up our intelligent sensing group. And finally by Simon Keaton, who heads up our Power Solutions Group. The BU will be followed by Bishram, who is our COO, and he will tell you how he ships and manufactures about 80,000,000,000 units per year. Finally, and most importantly, Bernard Gutmann will wrap everything up into financial metrics and he will explain how we create value for the shareholders. Now let me turn your attention to the most important slide in the slide deck.
This is a forward looking statement. You can read this at your leisure, but let me summarize it for you. During the course of this presentation, we will be making forward looking statements and these forward looking statements are subject to certain risks and uncertainties, which are detailed in our Form 10 Ks, 10 Qs and other filings with SEC. With that, it is my privilege to welcome Keith Jackson, our President and CEO.
Thank you and good morning. Good to see everyone here. I'm going to start with a summary and then you're going to hear the details behind all the key points. But hopefully the message will resound very clear about the transformation of ON Semiconductor and our progress accelerating toward the vision we've set out. We met 2 years ago for the Analyst Day with some objectives.
I think we have shown that not only is that model been an accurate model, but it's one where we're starting to accelerate the performance and prove out the market trends that we were expecting. We are participating and not only participating, but enabling the technologies that are required for the fast growth in disruptive markets. And we think we've got a business plan, a business model that is highly defensible and a competitive positioning and scale that is going to enable us to continue to make progress. And then lastly, we've been demonstrating very strong execution and very strong and consistent performance to expectations. So this is a great formula.
We think the company is great in a great position and we'll kind of go through some of those pieces here today. From the megatrends, of course, all of you are very aware of what's going on in the megatrends. In the automotive industry, there is On semiconductor is a key component in all three of those areas. On semiconductor is a key component in all three of those areas, not just a participant, but a key component. Further, if you look at those marketplaces, even with flat unit growth, you're going to be seeing accelerating semiconductor content growth.
And so we're very well positioned. Since our last meeting, we've added cloud power to the areas where we see growth and we include the 5 gs infrastructure into that. For us, it's the same type of power components and that now has been a great position for us with lots of deployment coming up in providing another leg to industrial and automotive for us. From a competitive basis, we've now got scale, which gives us great cost structures, gives us great supply chain, gives us great reach in our sales and marketing side. But we also have differentiated products, which is where the investments go to make sure we continue to have the best performance.
And we have a very diversified customer base. One of the keys to our success is supplying all of the leaders in each of these markets, not just one of them or 2 of them, but all of them giving us the optionality or the opportunity to grow with the market and with the winners. From a cash flow perspective, you've all seen the numbers, but over a 3.5 times increase in free cash flow over the last few years, 3.7 times on the non GAAP EPS and of course significant gross margin expansion and operating gross margin expansion. So what are those things in the megatrends? We talked about ADAS.
All of you know us for the image sensors. Since we've last met, we've added LIDAR and radar and we're spending our time with T OEMs looking at how we fuse all of these sensing technologies into an intelligent platform that will save them money, save them weight and save them data rates. On the power side, we're enabling the electric vehicles, whether that's with power modules that use silicon based solutions or with silicon carbide, But that is just again a huge increase for us over the next few years in content. And then in power management, we've got for automotive, all of the power management you're going to need. And in electric vehicles, you don't have the alternator belt driving things, you need new solutions and ON Semiconductor has all of them.
From an industrial perspective, we're leaders in the sensing for machine vision, looking at the expansion of robotics and automation in industrial, great application for us. From a power perspective, all of the robotic arms, all of the things that go on in industrial need to be driven with power. And of course, we've got the most energy efficient products there. And then from management, power management for the IoT, which we think is the connectivity that are required to modernize all the factories were there as well. Cloud power, we've got the CPU power, whether that's a 5 gs base station or a cloud server.
We've got the MOSFETs that do all of the other powering to those systems. And we've also got all the power you need in your data centers. So basically we've got it all, but more importantly, we have the best performance in each of those categories making us the preferred supplier. This is our market breakout. One of the things I will draw your attention to, we have now split out the infrastructure piece of communications and combine that with the cloud portion or the server portion of computing, put that into one sector that we call cloud power.
So what you've now got for communications is our handset business and what you've got for the computing side is the client computing side. We made that split out specifically because we are seeing that breakthrough growth in cloud power and 5 gs stations. We think that's going to be significant and give you a great indicator on the progress we're going to make towards our model. Of course, our position in automotive and industrial continues to grow. When you look at the combination of those, you get most of our business and the margins in those businesses are the best for the company.
Competitively, I mentioned participation. It's nice to be a player in these high growth markets, but we have spent our time and energy and investments in being the best in class performance. For things like power, that means the highest efficiency. For things like connectivity, it means the lowest power consumption. And for things like imaging, it's going to be the best resolution and data.
And we've got all of that accomplished in our markets. We're going to continue to invest at the rate we've been investing. We think this will keep us at the front edge of that. So even if there are other competitors with similar products, we're going to be able to drive the stronger market share. Beyond that, if you look at what it takes for industrial and automotive, these are markets that are very sensitive to reliability and dependability.
They need a track record on semiconductors had that. We've been there for decades. We understand those markets and we provide the highest qualities and reliability out there. So breaking in and changing that from a competition perspective is going to be very difficult. From a lifecycle on the products, again, the markets we've been focusing have long lifecycles.
I'll talk a little bit more about what that means to us, but it does enable us to get good returns on the investment we make in R and D and it gives us good returns on the investments we make from a marketing and applications perspective. Our life our products are diversified from a customer base perspective. I mentioned earlier our strategy here is to make sure we're able to get into each of the leaders in each of the markets. We've been able to do that. And so there kind of de risk the portfolio on which of our customers is the winner.
And then lastly, looking at someone who might, there's many times people wonder about will China or somebody come in and displace you. Company's got over 84,000 products in hundreds of different technologies. This is not something that's easy to duplicate. In fact, it's extremely difficult and takes a very long time. So we do feel very good in the longer run about our portfolio.
And we talked about the sticky products. You can see our growth rates there on your left hand side. So you can see we are growing. And then you can look at the pie chart and see the age of the products that are driving that growth. And it's perhaps oversimplification to say that things like cell phones drive all of the under 3 year old products.
But clearly the consumer, the client computing and the handset piece have more content there. And the automotive and industrial tend to be in the 3 plus year range. This gives us a lot of stability from a market return perspective and a great stability from an R and D return perspective. So when we look at our R and D investments and we say what's the right amount, you have to understand the markets that you're in and the technologies that you're driving, and we believe we found the sweet spot on that. Mention customers several time and the diversity.
Cycles come, cycles go, customers win, customers lose. We like to support all of them. Our biggest customer is about 5%, and it takes the top 20 of them just to get to 36 percent of the revenue. We don't list names for obvious reasons, but in this top 20 list you will find the top suppliers in each of our key focus markets. And so therefore, we believe we're very well positioned to take care of geographic disruptions, market share changes, M and A that might occur, etcetera.
So we believe it's a very good place to be in sustaining our model going forward. And these are the results we've seen. If you look at the gross margins getting over 400 basis points in the last few years, faster growth at 500 basis and your non GAAP operating margins may mean showing the efficiency of scale in our support structures. Looking at free cash flow going up very strongly last year over nearly $760,000,000 and we think there's a lot of leverage left on that. And of course, the earnings per share will follow.
Furthermore, in that time, we consistently either met or exceeded the consensus estimates out there. So it is demonstrating that this model is not only repeatable, but sustainable. Where are some of the margin expansions come from? You're going to hear Bernard talk about the future later on, But this is what's happened over the last few years. If you look at the light blue at the bottom of this chart, those are things that are less than corporate average margins.
And you can see there's very little growth there. That middle kind of amber color is things that are at corporate margins. You can see the median, you can see the growth has been very substantial there. And then the dark blue at the top are things that are above corporate averages and you can see the significant growth there. The mix element and the focus on markets is what's driving that along with the efficiencies in manufacturing that Bill will talk to you about in our pricing structures and efforts that we've had in our sales and marketing channels.
So again, we're seeing great progress there and we think the leadership products that we've got will command a premium to the marketplace in each of those fast growing markets and you'll continue to see this trend accelerate in that dark blue area. What are we trying to accomplish? We want to be and we believe we are leading in power, analog and sensors. If you look at the applications we focused on in those markets, we've got the highest efficiencies, the best performance specs, and we continue to invest there to make sure we maintain that lead. We think this will allow us to enable disruption in those markets.
It will allow us to become the preferred supplier in those markets. And so we should be able to maintain market share throughout the cycles as well. Focus on execution is going to help us be consistent and make consistent progress toward these goals. And of course the results for all of that are the improved margins, the capital efficiency that we get in the free cash flow. Bernard will go through capital allocation.
One of the most popular questions we get asked is what are we doing about M and A and how does that fit into your plan. I'll address the M and A thought process now and Bernard will get
the rest of
that. From an M and A perspective, we know we can create value for our shareholders with that. We've demonstrated it many times. And so there is an opportunity for M and A and creating shareholder wealth that we are aware of and we are not afraid of taking that. We're looking for the right opportunities, however, and it's not just M and A to gain size.
If you look at the things we've done, every one of them has been to gain technology capabilities to make us more important to the customers in our key focus segments. And that is a continuing theme that you will see is becoming more important. There is something we think leverages and accelerates our growth because as we become more important to a customer, we get more of the share of his bill of materials. Scale can be a reason for doing expansion. Certainly, we've gotten a lot of scale with our acquisitions over time, and there's a great deal of cost benefits on that and operating synergies.
From a transaction perspective, all the things we've done when we do them look like they are significantly above our cost of capital and also significantly accretive. So that is the pattern you should be looking for. We do think that consolidation in the industry is going to continue and on semiconductor wants to be a profitable part of that process, not just a consolidator for the sake of consolidation. We borrowed this slide from Morgan Stanley. It's their thoughts on where we are in the industry consolidation.
I will tell you that we echo their thoughts. The industry is going to continue to consolidate. There are large numbers of inefficient suppliers out there, does not make sense for the consumers at all. And so we see that continuing. There's been a pause after a great deal of activity.
You're all aware of that. I think the conditions are becoming ripe again for the next round of that. But we are again intending to be a portion of looking at how we can most profitably make that happen. We became a Fortune 500 company last year, having grown very significantly in the last number of years. And of course that entails a great deal of different responsibilities other than just P and L and return of cash flow.
I'll let you know you should have great comfort in on semiconductor as a company to invest in. This is the 4th year in a row we've been recognized as one of the world's most ethical companies. And we are one of only 3 that gets that designation. We've also been added to the Dow Jones Sustainability Index and recognized by Barron's as a sustainable company. We are a founding member of what's called CSR.
This is a group of companies that are looking at how enterprises should be practicing their corporate social responsibility, very active in that. And then from an ecology or an environmental perspective, we have always been a great advocate for the environment, providing products that eliminate waste in energy, but also within our own factories, from an overall company perspective, we think we're also a great place to invest. I'm going to let Bernard go through the details on the model. When we showed you this a couple of years ago, we had a goal in 2020 of having a 40% gross margin and a 19% operating margin. We think we have demonstrated that we're ahead of that plan, that we will be able to achieve those objectives handily and we haven't said what's coming up next.
And we'll go through the details with you as I mentioned, but we do see margin expansion continued to be enabled by the focus of high performance products, high growth and disruptive markets. We think that is a great formula. We still have a manufacturing infrastructure and network that can get some more optimization and that will be a contributor. And just the fall through on revenue growth, which we now back a couple of years ago thought was low single digits, we now believe are mid single digits of growth facing the company over the next few years. So with that, we're looking to create over $1,200,000,000 of free cash flow each year and make sure we can deploy that for our shareholders.
With that, I'm going to give you a summary of kind of where I started. I believe the company is pretty compelling right now. We are excited about the acceleration of what we're doing from a transformation in the company, really becoming a leading technology company, power, analog and sensing focused on markets that value that and have high growth behind it. The technologies that we've got are enabling a lot of this change. And we believe that the investments we're making in R and D are going to further uncover disruptive technologies.
The model that we've got we believe is defensible and also executable so that we can meet those objectives. And then lastly, our demonstrated financial performance we expect to be repeated into the future with significant free cash flow. I'm going to open it up now for some brief Q and A. And then again remember that we've got all the folks who will be coming after me. Ross, you've got a mic behind you.
Thanks, Keith. I want to go
back to that longevity slide that you had. It's clear that focusing on different end markets is going to help on that factor, but investors tend to look it on as something that's hypercyclical and very, very volatile, not only in the revenue side, but also on the margin side of things. Talk a little bit about if it's something just driven by mix changing, less PCs, less handsets or are there areas where you're actually becoming more of a partner with your customer instead of just a supplier?
Yes, I think there's a few trends there that are very significant. One of them is we've been working very hard with our customers and now OEMs in some cases in automotive to provide a larger value to them rather than just components. And that's taking the shape of doing modules. And when we get into a module relationship with these customers, it does change the pricing dynamic because we're providing a lot more value to them. It also changes the percentage of the BOM that we get.
And it also gets us out of the deciding pricing annually or quarterly for a large list of components. And so it changes the whole dynamics there. We're seeing that impact. Our module business has got the fastest growth in the company, growing 100 of percent a year and is becoming a pretty exciting part of that. And once you get those modules in place, they are not easily duplicable or sourceable from other places, which again changes the dynamic in the pricing space.
So we see that bringing more stability. Further, from a capital efficiency perspective, we're looking at participating in the multi sourced markets where it makes sense and not investing in a lot of expansion there where it doesn't make sense. And so we think we can continue to drive margins up by becoming more efficient even in those multi source markets.
Craig?
Thanks, Keith, and congratulations on the progress on the prior model. You showed on, I believe it was Slide 16, margin mix within the portfolio. The question is with regard to the below corporate average bucket, what's the potential for further strategic mix out in the portfolio? And what's assumed in the new target financial model? And is there any M and A that's part of the new target financial model?
So the new target model is organic. There's no M and A in it. And from a specific assumptions on kind of the bands there, the assumptions are as follows. All of those bands are going to benefit from continued manufacturing efficiency. So we're going to get cost reductions in our manufacturing supply network that will hit all of those.
Naturally, as we're driving that midpoint up, all those basis points, that means you need to be dragging the things at the bottom with you. And so the cost reduction piece is a big part of that formula. The pricing mix that I talked about earlier by controlling the types of businesses go after, so the market focus on those is an important part at the low end of that scale. But really it's the innovation and the enabling technologies that drive the high end of the scale that gets you the most leverage.
Thanks. Hi, Keith. Vijay here. Just on that pie chart that you show on the same slide, if you look at 2022 on that
at that $7,000,000,000 top line, what should this pie chart look like, especially the cloud power and automotive? Thanks. So the automotive, industrial will continue to grow somewhat because the entire pie is going to grow.
I would expect automotive to take a larger share, maybe in the 33%, 34% range at that stage. And what you'll see is that cloud power piece should become double digits. And then the remaining piece of the pie will shrink accordingly.
Hey, Keith. Good morning. Abhirish from BMO. I just wanted to get back to the M and A metrics you have. Clearly, the last two acquisitions you have made, you've done a pretty good job in kind of reversing the prior to that.
But could you please share with us the metrics on how you measure those 2? So specifically when you said return greater than WACC, what timeframe do you look at and can you share with us what those last 2 have done? Thank you. Yeah, Bernard will
go into more details on the specific models, but in essence there, we look at our average, Bernard, 1 year, 2 year, 2 year is the answer to that. The average 2 year WACC is what's used on that. And the accretive place, I hope is pretty self evident. So it's the question there is how much and then we compare that frankly the options on buying back your own stock or dividends are considered as well and make the decisions as to whether it's significant lead better or not.
Keith, just going back to the 2022 target on top line. If you look at the CAGR you need to get there, it's sub-five percent. The glass half full interpretation is that seems like a pretty low bar, especially with the consolidation going on in the industry and some of the past comments you've made on your expectations around pricing. I wonder if you could just give us the broad strokes of how you built that kind of rev CAGR bottoms up? And as you discuss it from an end market perspective, I'd also love to hear kind of just your thoughts on pricing.
Yes. So when we do that, we do our we really haven't changed the pricing model, even though we've demonstrated better performance recently. We still look at it and say there's about a 6% ASP decline a year. Anything we do better than that will be upside. We look at about a 5% CAGR to your point and that was based on what we looked at when we're thinking about 5 gs expansion and deployment, looked at the cloud server expansion that we're expecting.
We looked at the automotive with a relatively, I think it's like a 1% SAR kind of number. And we looked at all of the IoT, the industrial IoT portion as we saw conversion there. And so we built that up. We saw a number of greater than 5%, but frankly judged it down and said those markets alone should drive us the opportunity to grow in those rates.
Thanks. Keith, just another follow on on the top line. And it's a faster growth rate than you were a bit too a bit too conservative? Do you think that the end markets are perhaps healthier, there's more fertile ground for you to grow or maybe it's a combination of 2, maybe combination of the end markets that you're seeking to serve?
Well, I guess history and data says we were a bit too conservative, having grown over 9% last year versus the 3% in the model. But I think right now when we're looking forward, the big changes at that stage, we didn't have a 5 gs deployment that we were looking at as a new driver. And we frankly did not factor in how fast the adoption in ADAS of companies trying to drive their models from level 1 to level 3 to level 4. And we think that's a lot faster. So 5 gs plus faster implementation on the safety side are both things that we were not looking at in 2017.
Hey, Keith. Thanks for hosting this great day. On the M and A front, how do you see the number of potential targets out there for you guys now? And how do you kind of judge valuations out there in the market? And then lastly, would you buy more commodity player at the right price or are you really more focused on moving up that value stack?
Yes, I think so I'll attack in reverse order. We are clearly committed to providing more value to our customers And that is really the key thing we look at first. And so we're after becoming very important to customers in that automotive, industrial, cloud areas. And so the question is, is an acquisition going to help us there or is it going to hurt us there or not help us there. And so that's kind of the first one.
The gross margin piece, you see us driving that hard. That is important aspect of what's going on. And so we would not be looking at looking to acquire things that we couldn't drive to those gross margins. So they do matter, the gross margin side. But the strategic fit, the accretion and then the ability to drive those gross margins are all factors.
Thanks for your questions. So with that, we will have David Somo. Thank you.
Good morning and thank you for being with us today. So before I begin, I just want to thank Parag for setting up a Q and A after Keith's session before the rest of us present. I don't think I have anything to present now. So we can move on very quickly. Now, before diving into the details of my presentation, I want to provide you with the key themes that you're going to hear from me, echoing some of the commentary that you heard from Keith during his presentation.
1st, on semiconductors, key enabling technologies that we've built into our portfolio over the past several years with R and D investments and through M and A are leading to enabling some of the key megatrends that we see in these secular growth markets. Today, about 65% of our business is exposed to the high growth markets in automotive, industrial and cloud power. Those key secular growth drivers for the company are really enabling outsized growth as we look at a market today where if the units are flat, the units are up, we'll be able to outperform in any scenario. In the automotive sector, there are really 2 key growth drivers. These are related to increasing fuel efficiency, improving emissions and passenger and pedestrian safety.
In automated driving systems or drive active safety systems and vehicle electrification. In the industrial sector, the overarching theme here is to improve energy efficiency. If you look at electric motors as an example, they consume about 40% of the world's electricity. There is a strong push from governments to improve the environment by reducing and conserving energy. Energy efficiency is a key part of that and you'll hear that throughout the presentations today.
On the automation side, ON Semiconductor is very well positioned with the breadth of portfolio that we have. And I'll elaborate on exactly what that means a bit later in my presentation. And then finally, in the area of cloud power, 2 key growth drivers there. The first is around the 5 gs infrastructure deployment that kicked off in earnest last year and will continue to grow over the next several years. The second is in the servers that go into the data centers that are storing, processing and distributing this information that goes along with the 5 gs infrastructure.
Looking first at the automotive sector, ON Semiconductor has a long legacy decades of experience in automotive. As you know, automotive is safety critical. Having that experience positions us very strongly with our customer base. That is the traditional Tier 1s. The investments we've made in key technologies in automated driving systems like our sensor portfolio, in our power solutions technology for vehicle electrification have made us increasingly important to the OEM customers.
We now have broad engagements where we are actively working on R and D with both the Tier 1s and the automotive OEMs to be able to enable their next generation of applications. Keith mentioned some of the key enabling technologies we have. I'll cover that a bit later, but the types of applications we're going into are around electrifying the vehicle, whether that be the traction inverter onboard chargers, 48 volt systems, plug in hybrid or mild electric vehicles. Then multiple different types of active safety features that are being adopted rapidly in the vehicles moving from level 2 on to level 4 over the next several years. On the industrial side of the business, here again, we have very broad, long standing relationships with our customer base.
ON Semiconductor has been known for some time for its power solutions and we're one of the few suppliers who provide the complete range from discrete devices, IGBTs, MOSFETs, diodes through modules and then later moving to silicon carbide and other wide band gap materials to address the importance of energy efficiency in our customers' applications. The other side is around industrial automation and here we're one of the few who has an extensive portfolio consisting of the power management solutions, connectivity, sensors and embedded computing capabilities that are needed by our customers to put together complete solutions for the emerging automated applications. On the cloud power side, 2 key drivers there. The first around 5 gs infrastructure, we have relationships with all the major equipment manufacturers to enable their next generation 5 gs systems that are being deployed. And on the server side, that's a new area for us that we've moved into over the past couple of years, leveraging the position that we've had and expertise acquired in the client computing space to provide power solutions for the high growth that we've seen in servers that are supporting all the data we as a society produce and consume.
Looking first at the automotive sector, if you go back over the past 5 years, ON Semiconductor has nearly doubled our sales to automotive customers through the capabilities we brought with enabling technologies for vehicle electrification and automated driving systems. On the vehicle electrification side, I'll give you more details in upcoming slides about the content opportunity for us, but it increases significantly up to about $500 On the automated driving side, not only do we see high growth that's approaching 20% as advanced safety features are increasingly being adopted, but we have a unique position in providing all of the sensor modalities that are needed to get to level 4 and level 5 autonomous vehicles and image sensors, radar, lidar and ultrasonic sensor interface, as well as the power management that goes behind those sensors and for the ADAS processors that go into vehicles. Looking at our addressable content, if I use 2018 as a baseline and as a representative model use a level 2 enabled vehicle from an automated driving perspective, that is a battery electric vehicle. Our content today is about $150 on the advanced safety side. So typically, these are going to be features like adaptive cruise control, automatic emergency braking, lane keeping systems and similar technologies that are showing up in the first wave of level 2 vehicles that rolled out in 2018.
On the electrification side, there's about $200 of content in that model vehicle. Those are typically IGBTs and IGBT modules that are going into applications like the main traction inverter and onboard chargers going into these vehicles. If I flash forward over the next 4 years to 2022, you see our content increasing nearly 3x over that time. The drivers of those increases are in vehicle electrification increasing by more than 2x to $500 and that's primarily moving from IGBTs and IGBT modules to new wide bandgap materials that provide greater power efficiency, power density and reduce the size of weight of the different systems within the vehicle. In addition, our content for automated driving systems increases to about $1,000 I have more detail in upcoming slides, but that essentially is around more sensors, more radar elements, solid state lidar distributed throughout the vehicle and more cameras as well as the power and connectivity that supports those systems.
So in total, that leads us to up to nearly $1800 of addressable content in a level 4 battery electric vehicle. Looking specifically at each of the key applications areas now, vehicle electrification, if you go back to the internal combustion engine or the diesel engine, on semiconductors power content is about $40 of opportunity. As we look at a plug in hybrid electric vehicle or a pure battery electric vehicle, our content increases up to about $500 The presenters that follow me, particularly Simon Keene, are going to share with you some of the details around the ON Semiconductor products that go in here, but the applications areas that these products go into include the onboard charger, the main traction inverter that comes off the battery, their auxiliary inverters that are feeding the increasing content in electrification in motors that go into blowers, pumps, fans, seats and the like. The starter generator and then the high voltage to 48 volt down to 12 volt DC DC conversion. So our power products go into these areas as well as having motor drive and the connectivity for the communications buses between the different ECUs to enable all these elements to be controlled throughout the vehicle.
Let's look at automated driving, going back to level 0, passive safety systems, seat belts and airbags, little content opportunity there for on semiconductor or semiconductor companies. It's about $10 As we go to the level 2 systems today, which primarily are consisting of a number of image sensors, a front facing surround view, the backup cam, You see supporting and a front long range radar that supports adaptive cruise in your automatic emergency braking functions. The cameras are supporting things like lane departure keeping systems. We have about $150 of content between the sensors and the power for those sensors. Going forward to a level 4 system, you increase the number of radar elements that go into a vehicle, so it's not just the front facing long range radar.
You have medium range radar, millimeter wave that's looking around the vehicle. You add multiple LIDAR elements as well that have higher resolution to be able to look around the vehicle and are less interfered with by weather. And to have a level 4 vehicle where you're allowing the car to make all the driving decisions, it's important to have redundancy in the vehicle and hence you're going to see a large number of image sensors, radar and lidar elements together with ultrasonic sensors in these Level 4 vehicles. In addition to that, we do the power management devices that support each one of the sensors that go into the vehicle. We also are doing the power for the central processors that make the decisions around what to do to steer and navigate the vehicle through its environment.
So as an example, a couple of years ago at CS Intel shared that we were doing the power management for the Intel Go platform for their processor, Denverton, that's going into some of these automated driving systems. Looking at lighting, this is an area that's important for both aesthetics, where the vehicle manufacturers differentiate their vehicles from a look standpoint, as well as a safety factor. In the traditional lighting, HID and incandescent lamps, we had about $3 a content. If you look at the newer vehicles that have advanced front lighting interior LED lighting and rear LED lighting, we now have up to $30 of content in these applications. As an example, there's a North American SUV that's in the market today.
We have content that goes into the advanced front lighting system. Each one of those advanced headlamps has about $12 of our content that spans from the LED controllers for the LED strings that do high beam, low beam and the turn indicators to the motor drivers that position the lighting, whether it's moving and tilting up and down or swivel side to side to follow the road. And then we have the connectivity, the CAN bus and LIN bus that goes with it. So in total throughout the vehicle, we increased our content up to about $30 Switching over to the industrial market, 2 major areas to talk about here. 1 is primarily around power and improving energy efficiency in the different devices that are used in industrial applications, motors, power supplies and other areas.
The second area is around automation. Each has strong growth potential and as you can see, our revenue here has more than doubled over the past 5 years due to the portfolio and key enabling technologies that we've put together from a product standpoint feeding into these high growth applications. On the energy infrastructure side, we have both renewable energy that's in solar and wind and then we have things like EV charging stations that are coming in to support the growing adoption of electric vehicles. Industrial power for motors is another key growth driver. And finally, industrial automation, areas like robotics, machine vision, these devices are becoming increasingly intelligent, connected and able to sense the environment they're operating in and our portfolio spans each of those key foundational building blocks.
So some examples of where the content increases are coming from in the areas of energy infrastructure, coal power plant, not only is it bad for the environment, it's not good for semiconductors. There's very little semiconductor content there. But when we go into areas like distributed solar, we can have up to $6.50 of content between the inverter and boost converter stages. Today, those are supported primarily by IGBTs and our power integrated modules using IGBTs. Looking forward in time, those will transition to higher efficiency solutions like silicon carbide and silicon carbide modules.
On the charging side, if you look at the traditional gas pump, nothing there for us to speak of. When you move to an EV charging station, that is fast charging. So this is one that can top off your battery in about 20 minutes. There is nearly $500 of content available to us between the DC DC conversion and the power factor correction that takes place, as well as having the onboard charger content inside the vehicle that works together with this. So significant content increases in both of these areas for the company, driven by the move to higher efficiency and the new types of vehicles or power generation facilities we're putting in place.
Industrial power and motors, if you look at the traditional AC induction drive, almost no content there for semiconductors. When you move to variable speed drives that are much higher efficiency, and these are becoming more broadly adopted now across a number of areas, Our content can go up to about $40 so a significant increase there led by power conversion and power factor correction as well as the inverter motor power supply going back from DC to AC to drive the motor. So you hear a bit more about the details behind this during upcoming presentations, but it is one of the key areas for us where the with motors consuming so much of the world's electricity, about 40% again, it's important to make them more efficient. On the industrial automation side, here again, if you go back to human manufacturing, almost nothing there. When you look at the transition that's occurring to robotic manufacturing or industrial automation, there is significant content there.
ON Semiconductor is very well positioned with the investments we've made in our power portfolio, building our connectivity and sensor portfolios, as well as providing some embedded processing for these automated manufacturing systems and other types of robots. If you look at China for an example, this is one of the key trends within China. They realized their labor is no longer the lowest cost in the world. So how do they maintain their manufacturing base? They move to more automation.
So this is a trend that we see increasing from a manufacturing standpoint. We're very well positioned with the ability to offer a complete breadth of products spanning the connectivity, sensing, power and embedded computing technologies. Now switching over to cloud power, the 2 key growth drivers here from a content standpoint for On Semiconductor are in 5 gs infrastructure. We're showing back again 2016 to 2018 over a 3 year period. We more than doubled our sales in this space.
With 2018 being the 1st year that we've really seen strong contribution from our entry into the server market. Here in 5 gs infrastructure with the architecture they're putting in place, 5 gs new radios, small cells to augment the macro base stations that are put in place. We have significantly more content and more base stations to be able to address with our power technologies like our medium voltage FETs. On the server side of the business, this is an area where we've taken our expertise in providing power management for the core processing and memory in client computing and move that into the server environment. And now we have substantial content opportunity in servers that is rapidly growing as we produce and consume more data that needs to be stored and moved throughout these data centers.
So, first looking at 5 gs networks, if you go back to 4 gs, it was not millimeter wave technology. You had your base station architecture. You had MIMO, but it was more early stage MIMO in these base stations and our total content opportunity was about $9 As we move into 5 gs technologies using millimeter wave signaling, there are now going to be more cells, more radios to bounce these signals around at higher speeds to achieve the 10x higher data rate communications. They're going to use massively MIMO architectures and beamforming, which in total provides more baseband processing, more radio power requirements and more power supplies for the additional base stations that are being put in place. Our portfolio of medium voltage FETs as well as power management for some of the network processing is very well suited for these applications.
Efficiency is critical here as you deploy these networks and that's where we have an advantage with our technologies that will be shared in the upcoming presentations, providing up to about $170 of content in these systems. And on the server side, if you go back to VR12, we actually did not have a solution yet when we looked at those servers. Going into VR13, following the Fairchild acquisition, they had developed a smart power stage that was very efficient in providing power for the core processor as well as the memory subsystem in these servers. Each successive generation of Intel platform around their server architecture has increased the power requirements to support the processing. Increasing the power requirements has driven for higher current.
The higher current drives more phases that need to be managed and the more phases drive more controllers and more power stage content opportunity for ON Semiconductor. So as you hear in Vince's presentation that's following me, he'll talk about what his group has done to address this opportunity. We have in the VR-thirteen server platforms per lead that are now in production about $45 a content opportunity only in the power stage and some of the LDOs that are used. As we move into VR13 HC, the power goes up again. We do not have the controller.
We do have the power stage and it represents about $60 of opportunity in the VR13HC platform. And as we move into VR14, that's where we introduce our 1st controller, our digital controller that goes with our power stage, further increasing our opportunity, as well as once again the power requirements moving up for a total of about $75 of content opportunity in these Intel server based platforms. Now another aspect of this that I didn't address here, but is becoming increasing trend is with AI and doing machine learning or deep learning in the cloud, as well as some of the inference that is distributed between the cloud and moving out to the gateway or to the edge with intelligent devices, they're going to be increasing adoption of accelerators. So the power management that we put in place not only supports the main processing that takes place in these servers, but there's an opportunity that will increase over time with the accelerators that are being used for specific frameworks and workloads to be able to provide new types of computing that are being envisioned. So with that, I'd like to wrap it up by saying, as you heard from Keith in his presentation and mine on semiconductor with its broad based portfolio and the investments we've made in key enabling technologies has significantly increased our exposure to these key megatrends in the secular growth markets of automotive, industrial and cloud power.
That has the ability to drive outsized growth for ON Semiconductor in markets where irrespective of the unit growth, we can certainly outperform. In the automotive side, the 2 key trends around automated driving and vehicle electrification are taking advantage of our capabilities in our sensor portfolio and our highly efficient power solutions to enable the next generation of technologies that are supporting these trends. On the industrial side, we're supporting the energy efficiency drive that's taking place on a global basis, supported by government regulatory actions and mandates to reduce electricity consumption, while also uniquely providing the complete portfolio of foundational semiconductor building blocks in sensing, connectivity, powering and embedded processing for the automation of these systems like robotics. And in the cloud, both 5 gs infrastructure with our power solutions and servers present a strong growth opportunity as content increasingly goes up as they deploy these systems. So with that, thank you for your time and I'll bring up Vince Hopkins.
Thank you, David. So hopefully, you can all hear me okay. Vince Hopkin, and I manage the Analog Solutions Group. So I'm going to start off the same way that both Keith and David did and just talk about the takeaways and hopefully some things that you'll remember about the Analog Solutions Group and what our focus is. So we're going to talk about how we're driving secular growth in ASG and really how we're seeing a lot of content increase.
So we've talked about automotive, industrial and cloud power. Those are the key areas that we're focused on in my group. And we do that through differentiation. And a key element of that is ultra low power consumption, integration and high reliability. So Keith also mentioned efficiency around power.
That's certainly one of those as well. On the integration side, we do a lot of integration within my group. We build parts that are very small, but we also build highly complex solutions that are system and package, for instance. So a lot of integration that we use for our customers, especially in the three markets that we're focused on. Of course, we leverage analog power management across the group.
It benefits us in a big way in automotive and cloud. I'll talk about that more. And then we'll talk a little bit about margin expansion and what we're doing in my group to improve margins over time. So first off, what's Analog Solutions Group? Basically, this is the mix of the revenue in my group.
Automotive is 31%, so definitely a key market for us, the largest market. And you heard David talk about some of the areas we're leading in, so LED front lighting, sensor interface ICs for things like ultrasonic sensors, we'll talk about that in more detail. And then ADAS, power management, is a key focus for us adding a lot of content growth moving forward. For industrial, it's 2nd largest at 24%. We have a broad portfolio of products around industrial, and we are leading in things like power conversion, so AC to DC conversion products there, power safety, and then a lot of different industrial ASIC and ASSP applications in markets that are industrial, but things like medical, medical imaging, implantables, hearing aids, Mil Aero, so a lot of different products with very high value, high margin type products that we service in this market.
2nd or third, we're going to talk about cloud power. It's only 6% of our business today. If you looked at this a couple of years ago, it was much less than that. So we've seen a lot of growth. We'll talk about that.
David mentioned in some detail the products and what we're doing there to focus on that market, and I'll go through that as well. We do have a client computing business, and we're still involved in that, and we're still number 1 in things like desktop controllers, for instance, around power, but we're leveraging that capability around multi phase into the server space. So over the last couple of years, one of the ways that we've been able to move fast and really see a lot of growth on the server side is taking our capabilities around multi phase digital control and power and really taking it from the traditional computing business we had and leveraging it around computing and also around automotive for ADAS, because we're also developing a large PMICs that are multi phase as well. So that's how we're going to market with cloud. Lastly, we do have a wireless communications business as well.
Same thing, we have a lot of DC to DC type products in that business that we leverage into some of the other key areas that we're focused on. So strategic intent, we look at the areas that we're focused on here. Obviously, we're investing heavily in analog power management. We'll continue to do that for automotive, connectivity, ultra low power is critical for us. So that's where our investments are and that's where we're going to keep focused around.
We leverage a lot of the synergies between my group and Simon's group around PSG and Tanner for ISG. You'll hear Simon talk about some of the same market areas that I'm focused on like cloud and ADAS, for instance, and EV, and we leverage a lot of our products together to go to market with those for a total solution. And then of course for image sensing in an ADAS type solution, every one of those image sensors have power that goes along with them. So we leverage heavily around Tanner's solutions for ADAS, for instance, as well. So key part of our strategy moving forward.
We are definitely focused around margin expansion. Last year from 2017, we improved by about 800 basis points in ASG. So definitely focused there, and I'll talk about that a little bit more. I have a slide on that. And then lastly, Keith mentioned this, but enabling disruption, a lot of our products today are driven by either 5 gs in that megatrend or ADAS in that megatrend.
So a lot of those megatrends are really what we're focused on to drive our product development and make sure we have products that complement and drive different types of applications with those megatrends. So how do we win? So of course, obviously, we participate in product categories that we have a competitive advantage in. We are and have been traditionally a high volume analog business. So DC to DC, AC to DC type products, very high volume, very high volume power switching products.
We do precision analog as well. But really, the core of our business has been on the high volume side, and we leverage heavily the manufacturing capabilities that ON has, very low cost. Bilsch Sturm will talk about that in more detail. It really allows us to win in those markets. The other thing is, obviously, high efficiency around power.
So Keith mentioned that. When you're doing power stage type products for servers, robustness is really critical for customers. And we have some of the best solutions in the market. And because we have very good low efficiency products, pricing is not the big driver for product selection with our customers. It's more around how efficient are your products and what's the quality level of those products.
So we tend to see good margin around that business for that reason. We do leverage differentiation in low power and power efficiency, I talked about that already, and high reliability. So a lot of the businesses that are in ASG require high levels of reliability. So we have automotive, obviously, that's 31% of our business. But in the industrial space, we also have medical, which is also a high rail market, and we have Milero, which is a high rail market.
So the culture of our group is focused around very high reliability, very high quality, and that culture is something we take very serious with our customers, and I think it really allows us to win even in markets that don't require those kind of levels of reliability. We are focused on automotive, industrial and cloud. That is the key area, and that's where we tend to double down on from an investment perspective. Why are we focused there? Well, obviously, there's barriers to market entry.
Longevity and lifetime cycles of these products are over 5 years in a lot of cases. In ASG, we have a lot of a high percentage of sole source products. So we take that very seriously with our customers. Obviously, if we're their only supplier, we have to make sure we do supply. And that is how we win because we're very good at that.
And we've been supporting that type of product with customers for many years. So the other part of that is just you tend to get better margin with the approach that we have here in the markets that we're involved in. So lastly, I talked about manufacturing. So we leverage our internal capabilities as much as we can. This gives us stable supply, gives us controllable supplies.
We have a lot more control over what we're doing and the destiny that we have. It provides lower cost for us and higher quality. So we're going to continue to leverage that moving forward in ASG. We do use outside supply as well as a complement, but we'll continue to use and leverage Okay. So growth opportunities in the strategic markets that we have.
So the previous slide I showed, we had 31% of our business in automotive. It's a large TAM, dollars 30,000,000,000 For the areas that we participate in, it's over 6 percent CAGR when you look at 2017 to 2022. The key solutions that we're investing in are listed there. I'll talk about that in more details in a few minutes. For industrial, again, large TAM, this is a widespread market.
There's a lot of different things going on. We are focused around things like ultra low power wireless connectivity, advanced motor drivers. So you heard David talk about motors in general. We have a concentrated effort around things like 48 volt VLDC, for instance. So this is an area we're going to invest in.
And then embedded MCUs are being used heavily in industrial to really improve time to market. So for instance, we'll take an ARM based core, we'll add some analog functionality to that, we'll add memory, we'll then create or develop known good dye and manufacture them to where they're fully tested. And we'll develop an analog front end in the system in package with that known good die to really speed up our time to market for customers for specific applications. So that's an example of what we're doing with embedded MCUs. And then on the cloud power side, the key solutions, David's talked about some of this, but we started off with Smart Power Stage a couple of years ago, and we've been very successful.
You'll see some of that growth. And as David mentioned, we're definitely going to be in the So we're going to we're definitely going to be in the market for VR-fourteen with a controller and power stage moving forward. We also support point of load within this market as well. So there's investment going on with that type of product as well. So just to talk about some of the markets in more detail.
So our revenue, you can see the growth on the left side there. So we've had very good growth over the last few years for the automotive business that we have. Number 1 supplier in LED lighting, so we're seeing a lot of content growth here with our LED drivers and our stepper motors. And as you see more animated backlighting with companies that are doing that today. With automobiles and matrix lighting, we're seeing a lot of demand for our drivers and our controllers that are involved in those kind of solutions.
So we see this as a key area of growth for ultrasonic. We also see this as a key area of growth. I think 5 years ago, there might have been debate about ultrasonic and whether it was going to be growing or not moving forward. But really, ADAS has kind of stimulated growth for ultrasonic to where now we're seeing most automobiles in the future having over 10 ultrasonic sensors per automobile, in some cases, 20 to 30. So we see a large growth opportunity here, more than 20% growth in content there for us, and we had 35% growth last year in this type of product.
So very good area for us. And then lastly, ADAS Power, we've talked about it a lot. David had mentioned that we have
little bit
more detail around that. But we are one of the only ASIL management companies or we have a product that's ACIL certified today. We're the only company that has that. It's a very complex PMIC power management IC that goes into solutions for companies like Intel for ADAS. So I'll talk about that in a little bit more detail.
So some of the growth drivers here, if you look at the TAM, I had a chart earlier that talked about the TAM being over 6%. Front, interior and convenience sliding, content wise, it's about $25 per car for us. David had some of that data as well. Safety and drivetrain sensing, it's about $50 per car, so a lot of content there for us. And in ADAS power, it's a minimum of $40 per car just for multi phase when you look at power there.
So this is an example of how the content is growing for ADAS. So if you look at what automobiles have had for limited assistance and probably some of the cars that we're driving today or you're driving, we had about $50 of content when you look at the ultrasonic sensors that were in those cars or are in those cars today and the safety and comfort around LED lighting. But what's changing when we move to advanced driver assistance is a lot more sensors around ultrasonic, pressure sensors, inductive position sensors. In vehicle networking is definitely growing there. We're doing more integrated solutions for system basis chips there where we're combining functionality with the LIN and CAN in a lot of cases and power.
So we're seeing content growth that way and value growth with an IC. And then of course, LED lighting is growing with the things like matrix lighting that's happening. But lastly, ADAS Power is a huge part of our growth. So we see a high end of $500 of content with some of the applications out there that are fully automated. This is a Power Tree that I mentioned.
And if you look at this, just the block diagram and we've kind of drawn lines around a bunch of different products there. But really for ON and if you take again the Intel Go example, we take one of their CPUs, basically we have the PMIC and we're the only company that has the PMIC that's fully ACEL certified that goes along with those CPUs in the system and it's fully integrated. So it takes all of those functions like the control, the power stage, the sensing and the monitoring, and it's all contained within one chip. So very, very complex. We're really excited about this.
We'll start seeing some limited production next year on this with various different applications in cars. And we're also leveraging our battery connected power capabilities with some of the other solutions around this. So outside of the PMIC, we also have things like pre regulation, post regulation. There's a sub PMIC in this. So a lot of content here, you can see it's $40 for each processor with the power that we provide.
There's $2 on the sensor there. There's $5 for the module. So we see this as a key area for us moving forward. So switching to industrial for a minute, another area that we are really excited about, ultra low power, power connectivity is one of those investment areas and we have products today that are really gaining a lot of traction in the market. One of those is called RSL-ten.
It's a low power BLE device, Multiple different applications, we actually targeted that for industrial, industrial IoT, medical health applications. An example would be someone who needed to measure their to use continuous glucose management type devices for diabetes. And today, you can actually have a needle that's injected into your skin with a small chip in it that has BLE that allows you to look at your phone or your smartwatch on any time you want and see where your levels are. So that's a typical type of application for this product. We're really excited about this.
It's already in the market today. We're building out a roadmap for it. We also have sub-one gigahertz transceivers for those type of radios, for proprietary protocol type applications, seeing a lot of opportunities, especially in Industry 4.0 there and also industrial IoT. We see a growth in industrial around high speed interfaces. So we have a business today for USB 3.x type applications.
We have total solutions around that with things like re drivers and data switchers and audio switchers, interface protection, those types of things and we're really leveraging that into the industrial space today, Dew, and we see a large growth area for us there. And then I talked about embedded processing already, but we're going to continue to invest there with known good die around various different ARM processing based platforms with security built in to target industrial the industrial business for us. You can see how we've grown. It's a good market for us. We're going to keep going with that.
These are the growth drivers. So I talked about productivity. That's about a 15% CAGR for us over the next few years, and that's what we're modeling actually and forecasting. If you look at smart building and home control, we see about 8% CAGR with that for us. It's a big opportunity.
It's a big market. The high speed data side, especially around industrial, is going to grow quite a bit with things like smart connectors. So we see that as greater than 30%. And lastly, Industry 4.0 is a key area for us that we're focused on. We see content growth of more than 30% there.
And David had mentioned some of the applications around robotics, industrial automation. And so we provide products that have embedded microcontrollers in them that do actuation and positioning, sensing, so that the robots actually know where they are and large TAM, as you can see on the left side. So I'm going to talk about cloud now. This is probably the hyper growth area for us over the last couple of years and definitely moving forward. You can see what our revenue has done.
So we've really doubled our revenue in a couple of years. That started with the smart power stage. We are definitely a player there now. We're a leader. And it's really around the efficiency of our medium voltage FETs and our power stage devices.
The multi phase power control piece, I talked about the digital controller that we've developed, That's going to add a lot more opportunity for us. So we see about $600,000,000 of new opportunity just this year alone if we had the digital controller and greater than $75 of content when you take PowerStage alone. So a big opportunity for us when we take David mentioned AI accelerators, that's certainly an area that we're also focused on. We can see up to $150 of content in those type of products. We also have point of load.
We've been doing and supporting point of load products for multiple different markets, including automotive, mobile, computing. We're leveraging that into the base stations and the servers now, so you can see the content there. And then there is a push in the market to move toward 48 volt solutions, so customers can improve efficiency and reduce power even more. So we see that as an expansion of our SAM in this market as well. So key drivers, this is the TAM on the left side.
So you can kind of see what the market says about TAM. Talked about AI accelerators. What we're counting on for our growth is we're going to see over 100% growth in with those type of products. We're going to see 100 and 10% growth on 5 gs and data networking that's being driven really by the deployment of 5 gs. We're going to see greater than 70% growth for us around large scale storage and 40% growth around graphics cards and workstations and those type of products.
So David touched on this, and this is just more of a graphical view of why we're seeing content growth with the new Intel every time Intel comes out with a new platform. So there was the last platform that really finished up last year was VR12.5, excuse me, and it was called Brantley. And that is now finished. So we've really been ramping up on VR13, and that started last year. And you can kind of see that the peak power requirements and that's the for the Perley, that's the 4 10 watt measurement there.
You can see that doubles when we move to VR13. Hc. That's an intermediate step to VR14. So basically Intel came out with that after they initially came out with a VR-thirteen spec and a VR-fourteen spec. So it added a lot of opportunity really for us because every time that peak power requirement goes up, there's more power stage requirements that we need to be able to support.
So you can see what happens and you can see the content there. So when we get to VR-fourteen, we're really up to $76 there of content with just power alone. So this is what's driving the server content growth. And if you see just some other pictures here that kind of give you the story, that's a typical GPU board there. And it's got the processor there.
There's 16 phases on that board. So that means there's 16 PowerStage devices, which you can see the arrows pointing to, which we have. And there's a controller on the backside of that board. So the content with just one board is quite substantial, and there's multiple boards in a server. And then when you look in a rack system, there's a ton of these things in a rack system.
So when you look at a typical hyperscale data center, 3,000,000 servers and customers are really looking for what the effectiveness is of your power usage. And that's where our products are superior when it comes to efficiency. We're able to save customers 1,000,000 and 1,000,000 of dollars over the lifetime of those servers and the prediction is up to about $40,000,000 So you can see with just a little bit of efficiency improvement, they save a lot of money. That's why the price sensitivity around these products is not as large as it would be for other types of products in the computer industry. So just talk about margin focus for a minute.
Clearly, we're focused here. We saw improvement last year. We are the mix of our business changed last year. So you saw how much we grew with our cloud products. Margins are better there obviously than what we would have seen in traditional computing type products.
So we are focused on high value investments. We're going to continue doing that. We're going to reduce investment around areas that we don't see as driving the value that we need. The key areas that we're going to focus on for secular growth are all areas that are going to help us with margins. So we've talked about those.
And with cloud power, connectivity, embedded solutions, medical is another area that has very high margin that we can see a lot of growth potential in some areas. And then Bill Strum is going to talk about operations, but the scale that Keith mentioned as a company allows us not only to drive costs down within our own manufacturing, but also with our suppliers. So we have a lot better buying power. We're seeing benefits of that today. That's going to help us continue to improve margin moving forward.
And of course, we're going to continue with our technology advancements. So it's a treadmill. We're going to keep working on those. For instance, medium voltage FETs, we're going to keep working on technology advancements there. So we have best in class when it comes to efficiency.
So this is a key part of our margin focus. It's just one slide, but there's a lot of work that goes behind this and we're going to keep focused on that. So just to summarize, we talked about the secular growth areas that ASG is going to drive and automotive, industrial and cloud are those key areas. We're really excited about those market trends within those areas. And I think it's going to really help us over the next few years with our growth.
We're going to drive toward products in developing solutions around areas that we can lead in, so that's a pretty obvious statement. But I think that the areas that we are not a leader in, we want to either deemphasize the investment there or we want to leverage that technology capability into areas that we can win at. So that's really going to be what we focus there on. We're going to see more traction, especially in cloud power and ADAS. So you're going to hear a lot of that from ON over the next few years, especially from ASG.
And lastly, we're going to continue to show margin expansion. So thanks for your attention. I think we have a break now. So I think that's it.
Thanks, Vince. We'll take a 15 minute break and we will reconvene at 9:40. Thank you. All right, folks. Let's get started back.
All right. Are we on? Okay. Good morning, everybody. My name is Tanner Oslik.
If we can flip to my slide. Not yet? Product we're not starting yet or are we? I have the flicker here, yes. Yes.
Okay, good. All right, so Intelligent Sensing Group, so I want to start by highlighting the change in our name. As you know, we have been calling ourselves Image Sensor Group in the last 4 years since inception of this group. Late last year, we have decided to change the name to Intelligent Sensing Group. So there are two reasons for that.
Number 1, all of the products that we're basically developing and shipping are finding themselves in systems that are powered by artificial intelligence and AI is, in my opinion, one of the biggest transformations in our industry and it's actually infusing and demanding that our sensors become more intelligent and that's what we're doing as I will show you. Number 2, we're not just doing image sensors anymore. We are we have incorporated radar and lidar technologies through acquisitions. And we're also excited about those as I'll share with you in a minute. So I'll talk to you about 3 things.
Number 1, automotive number 2, mission vision and number 3, edge AI. But at the end of the session, I'd like to leave you with basically and convince you hopefully on these four points. Number 1, we have a pretty formidable position in ADAS, which is the fastest growing market for us, which is growing at about 30% per year. And not only that, we have amassed a pretty sizable design win funnel. In automotive, we have about $2,500,000,000 in the funnel, which is about 5 times revenue of the automotive business as you'll see and more than half of that is in ADAS.
We have and we take great pride in implementing and innovating in the technology and keeping ourselves as a technology leaders in the mission critical applications that we are attacking, so automotive and industrial. The technology that you put in these sensors can be the difference between saving somebody or killing that for somebody. So we take great pride in doing that and I think we have about 3 times more R and D in automotive than any of our competitors out there in the market. We're also bringing new sensor modalities, as I'll explain with radar and LIDAR because we believe in sensor fusion in the future. We are as a core focus now investing heavily in also machine vision and robotics.
We have number one position in the market as I'll share with you, We have 30% market share and those two markets, the robotics and machine vision is also growing at the rate of about 30%, which is basically same as ADAS. And the last one is that, as I promised you last time we met here, we wanted to focus on our gross margins because at that time the gross margins were not as healthy as the rest of the year. Well, I can tell you today that as you've probably seen in the presentation, we have improved our gross margins to the level that we said we would by 2020 already by the end of 2018. And we, I think, still have a little bit more room to go there and I'll explain how we're going to do that. ISG is made up of 3 divisions.
Automotive is more than half of the business now, it's about $800,000,000 business in total. In ISG, automotive is about a $500,000,000 business, slightly shy of that. Industrial is about a third of it and edge AI is about 15% of it. We have number 1 market share in automotive as I said. We're basically number 1 in all segments that we play in automotive as I'll explain in a little bit.
We have the technology leadership and brought us to product and customer portfolio. In industrial, number 1 in machine vision, we're growing there as well. And we serve all the markets that you see there, scanning, retail, in edge AI, security, robotics and flat panel inspection as well as factory automation. Edge AI happens to be a term that we obviously here in the industry and what I refer to it as basically any application that's either consumer or industrial, kind of like a smart IoT, where the intelligence is at the edge, where also of course you're coupling that with the sensors that we build. Our gross margin, as we highlight here last year was 41%.
We are focusing on 4 things. So these are the 4 goals that we have. Number 1, we are going to continue to invest heavily in what we believe in mission critical applications. So automotive and mission vision, those are the 2 most important things for us And we will do that to not only to keep the formidable share that we have that you'll see in a minute, but also to extend it hopefully. The second thing is that in ADAS, it's very important what you put in terms of technology.
So we're doubling down there and I'll give you some example. We're pushing for instance the boundaries of dynamic range up to about 160 dB, which is not achieved by anybody else in the industry but us and I'll show you some example of that as well. We're able to do that, ultra high dynamic range with what's called an industry LED flicker mitigation, which is basically a really hard thing to do. As human eye, you do see the LED lights continuously, you don't have the artifacts that you normally have in sensors, but we're now able to do that together, which also puts us as the only company that can do that. We have the best low light in the industry right now.
As Vince mentioned, we have also a big focus on ASIL, which is the safety and integrity levels, which is basically functional safety, which is very important for ADAS and we have the leadership there by I think about 1 year to 2 years. And we have put and led the industry in things like cybersecurity. God forbid, if you have these fleet of cars that are autonomously driving and somebody hacks into them, it can have a havoc in your hand. So we have led the industry to put cybersecurity in our sensors and everything that we built in ADAS comes out with that. And like you see there as well, we're investing in radar and LiDAR.
In radar, we have a great proposition. I'll mention that in a minute. In lidar, basically what we're trying to do is democratize lidar so we can build modules or our Tier 1 customers can build modules at about $200 a piece. We think we can get to about $100 bomb with all of the components that we supply from on and they can build it and sell it for about $200 Those modules today go for about 1,000 of dollars just to give you a perspective of the transformation that can happen there with our technology. We're focused on gross margins.
There are basically a few things that we've done there. Number 1, we have improved the mix. We have decided to deemphasize the businesses that were not generating enough return investments, so we exited them as I'll share with you in a little bit and we have also invested heavily on mission critical applications in markets like automotive and industrial. We've done a lot of operational improvements. We do think there's still more room there and I'll explain that.
And of course, of the best way of expanding your gross margins is that you are the leader in technology. If you can do that and consistently, you can gain the price premium, which is what we're doing at the moment and we intend to continue to do that. And the last thing I want to say is that the name of the game in imaging or any sensing technologies that you are the leader in that in technology. So we continue to look out for opportunities in products and applications where we can lead, particularly driven by artificial intelligence. So let me show you something.
So I have a sensor in my hand that I taped it. It's basically I think I call it Ninja, it's the world's I don't know if you can see it, it's like it flicking my finger like a blood clot. It's the world's smallest sensor, which is about 1 tenth of what you have in your cell phones, which consumes about 100 of a typical sensor, which is about 3 megawatts and can run up to about 400 frames a second. No one else has this technology and we're seeing incredible traction for that in the industry in all kinds of markets. ISG, right, I know a lot of you think, okay, so you do great products, why hasn't your revenue grown with the rest of the company or with the rest of the industry?
But there's only one simple answer to that is basically the mix. In 2016 2017, we decided to exit mobile. Mobile was part of our business. And we said that basically that that's become a race to the bottom kind of a market and the margins were completely unacceptable. So we exited that.
In 2017, which took us until about 2018, mid of 2018, we also decided to get out of the low end security market, which also became a low margin product. But if you look at the growth markets right there, which is automotive, which is in blue and in machine vision and in edge AI, in amber, we have grown pretty well over the course of the last 3 years. And the growth in automotive particularly is pretty impressive, which is about 24% in the last 4 years since 2014, since ISG's inception. How are we going to win? So I know you guys are worried about, so I'm going to talk about our competitive space.
Newcomers are coming into it and I'll show you more details in the rest of the presentation on this, but I wanted to summarize it here. Number 1, we have a 1st mover advantage. It matters in automotive, particularly, but also in industrial. We have been selling and developing these sensors for about 12 years now. We have basically relationships and system know how of almost all Tier 1s in the world and all OEMs in the world.
And it's incredibly hard for a newcomer to break into this network of trustful relationships, system know how if you don't have that over the course of the year. So it will take for anybody to come in and for those who want to switch, there's a high switching cost because all of their engineers are trained on our products, they use our tools and by the way, since the entire industry is going towards AI and AI is very dependent on the type of sources that you get for training data and given the fact that we have basically about 100,000,000 sensors already shipped in ADAS to date, all of those systems are trained on our sensors data, which is important to know. So that's number 1. Number 2, we're investing in, like I said, in technology. Low light was one source spot for us until about middle of last year.
We have doubled down on that. We have partnered with our manufacturing partners. We have increased our resource tremendously. And now I can claim squarely that we are the low light leader in the industry as well, both in terms of read noise and this is technical jargon, I apologize, and the SNU, which is dark signal non uniformity and you combine both of them, you get to low light performance. At high temperature and low temperature, we are clearly number 1 in the world in that.
The 3rd piece is the expansion of the portfolio through LiDAR and radar. RADAR brings us quite a bit of SAM expansion as I'll share in a little bit and LIDAR is something we acquired through Sensil acquisition last year and they have basically developed for 14 years a mature way of sensing depth using near IR frequency on silicon on CMOS. Why is that important? Because all of the LiDAR systems that you have or almost all of them use APDs or avalanche photodiodes, which are basically built on 3, 5 material, non silicon, costs about 10 times. Now we're able to bring down the cost and achieve the performance.
So they have the best performance in SNR, in distance, in uniformity and also what's called photon detection efficiency, which is basically how well you detect the photons and convert them into electrons, which basically gives you the timing information, which is the depth. So I'm quite excited about their technology. Since we acquired them, our design funnel has increased tremendously and we're making a lot of products there. I can tell you we're basically in all of the leading LiDAR solid state solutions in terms of designs today in the world that we know of at least. And the last one is Machine Vision at AGI, we're investing quite a bit there.
You'll see that you'll see some of the products there. So I won't steal the thunder there as well. But one thing I want to point out in that is that we are number 1 there. We have the best technology also. This is against Sony, which is our number 1 competitor at this point.
And we believe we have a great winning proposition there as well as we invest in that heavily also. Three growth drivers for the group. Number 1, automotive. In automotive, of course, as you know, we're in everywhere, ADAS being the fastest, but viewing is also growing. And cabin is in fact even faster than anybody else, any of those, which is in the form of either driver monitoring systems where you take the driver picture of the driver and see if it's drowsiness, it's drowsing or not and occupancy monitoring systems for detecting child, so you can deploy, for instance, their airbag with the right pressure and so forth.
Viewing with surround view and rearview cameras is still a growth market for us and then of course radar and LiDAR. In radar and LiDAR, our market share is small obviously. So the growth there is tremendous, as I'll share with you. All of these drivers have double digits, as you see there at the top. In industrial, what I list there is basically the drivers, but the top 3, the robotics, intelligent transport systems and mission vision are the most important ones and are the fastest growing ones for us.
And in AJI, we find these applications everywhere. So give you an example, Amazon has a store called Amazon Go. I think you guys obviously probably know about that or maybe even have been in it if you've I think in San Francisco and it's fairly phenomenal. You go in there, you grab anything you want, you just walk out and just charges you as soon as you walk out. It's all powered by our sensors.
That's a perfect application or example of what I call edge AI. Another one is I'll get to it. Another one is drones and other robotics devices that we have with customers like Amazon. Automotive market and the automotive business itself has been growing tremendously. So the last 4 years, as I said, is about 24%.
Our market shares are there, as you see, quite good. And we are focused on ADAS. And as I mentioned, the installed base is hugely important. So the fact that they have written software for us for the 100,000,000 sensors and training it and have trained it creates a major advantage for us. And the customer base and the fact that we work with them and we know everybody at these companies, including OEMs and Tier 1s at the highest levels at tops of the electronics divisions allows us to accelerate LiDAR adoption, which is what happened since the Sensil acquisition in May, as well as bringing some completely new device from on for the first time in radar, which is what we're doing also and there's a lot of attention because of that as well.
The growth drivers are everywhere, like viewing is growing, ADAS is growing at about 30%, in cabin camera monitoring systems, which are electronic mirrors is growing because of all sorts of obviously benefits on the outside, you have the air drag being less, so you can have more fuel economy. Even in the case of electrification or electric cars. Inside the car, you have better benefits as well. Autonomous driving, we do have I think I don't have this in the slide deck, but practically about 90% market share in autonomous driving. So we have shipped our sensors to everybody out there and they're using so all of the robotaxis, all of the car companies, truck companies, China, U.
S. Are using practically our sensors there. One other thing that I want to mention is that I believe we do have quite a bit of innovation and this is something I have here, which is called Iveq in my hand, which is basically a complete module in itself for in cabin applications. So it stands for in vehicle camera, in vehicle experience camera and it has everything in it. So it has the refouble, lens elements, it has the whole package of course, it has the sensor and has the SPU or the processor to do whatever you want to do with it.
It's mainly for pointed applications, right, for driver monitoring, for occupancy monitoring, for potentially for backup camera. And we have it. We're now deploying it and so that means like I think Vince and Keith mentioned, we're now going up in modules, so this will increase our ASPs as well. So I'm quite excited about that product line coming out as well. SAM expansion, we said we want to invest in the new sensor modalities, as I mentioned, because we believe in sensor fusion.
And we believe that doing that at the edge is going to be also beneficial to all the systems that we are building products for, including automotive and industrial. We expect about 10x as you see here from level 2 to level 4. David mentioned some of this. This is average is what we see in terms of SAM expansion. David mentioned what it can happen if you have all the large sensors that we're building for autonomous machines and we do have that in development.
But on the right, I highlight in green all the components that we're basically building for that sensor fusion platform. So not only the camera itself, of course, the radar transceivers, but also the LIDAR detector, all combined through this SPU. That SPU is also now designed by An in my group, which is basically a tiny little processor that has a neural network in it, nothing like Mobileye, nothing like NVIDIA. It's just a glorified ISP with a neural network engine. So you can fuse everything better.
The benefits of doing this is at the bottom. It's highly energy efficient. You can use basically 1 processor instead of 2 or 3. You can have a much less cable size or weight wise, there's a pretty tremendous benefit. You can increase the bandwidth, like Keith mentioned also.
And then of course, you have better sensing because you can leverage all modalities of the sensors at the same time. So radar can help lidar, likewise with radar helping imaging. So how are we doing? Last time I showed you what I have here on the left. So overall market share was about 49% in automotive.
In sensing overall in ADAS and autonomous driving, it was about 63%. While we have grown quite a bit, So in 2018, our market share overall was 62%, so that's a 13% expansion and in ADAS it was 81%. That's another 18% expansion. Now that's of course rearview mirror data. We're not resting on this and what's important is what we're building for the future.
So let me show you that. We have a lot of competitors. Some of them are newcomers. A better one is here, Sony, a competitor is Omnivision. The product breadth is something that we take pride in.
We have I think the broadest portfolio in automotive industrial markets that we targeted. We are the leader in 8,000 AD autonomous driving. We have the leadership in system expertise, which none of our competitors have. We have the global shutter technology that again is the best in class. But one area, as I mentioned before, that we were lacking was the image quality at high temperature particularly, but also in low light, but we fixed that.
So last year, as I said, we fixed that in both the SNU and read noise. I turned that green as well. And customer support is something we also take pride in. At ON, we have about I think over 1,000 sales professionals around the world, incredible sales network and all of them can sell our products, which none of our, I think, competitors have as an asset. The tools that we have helped, of course, but the people and the technical support teams are phenomenal.
This is an example of why high dynamic range matters. This is basically one of the most challenging tunnels that you have in U. S. Where autonomous driving tests are done. And as I said, we have the best in dynamic range technology.
And what you can see there is that compared to our competitors, the tunnel end of tunnel is not washed out. So you can actually clearly see the end of the tunnel and also the details inside the tunnel itself. Now I want to highlight that this technology came from another business that I have, which is basically cinematography. We build cameras for a company called ARRI, which is Arnold Richter, and it's based in Germany. It is maybe you don't know, now the number one movie camera maker in the world and they are basically used to shot all of the most of the Academy Award winners like Life of High or Reverence and so on and so forth and directors like Oliver Stone, Lee Yang love their cameras and it's all based on semi technology and we brought that in and they just won another Academy Award for Best Cinematography I think a week ago and the last 7 out of the 8 Best Cinematography Awards went to ARES cameras, which are again all powered by on semiconductor sensors.
So we're proud of that. And that technology now is in automotive. So this is how we benchmark ourselves compared to our top competitor today. And you can see that the breadth of what we do in both technology and product is extensive. I mean, you need to have all of this to be successful in automotive, whether it's in viewing and sensing or ADAS, autonomous driving, global shutter for in cabin applications, you need to have all of that.
And then the SBU is a completely new category of device that we're now establishing, which is the sensor processing units, which are these little chips that I mentioned. And just to show you the size, it's also quite tiny, like I have it in my hand here, which is basically same as what I showed you with the IVEC. It's a BGA package, simple edge neural network device that helps with the sensor fusion as well as point of application. And then of course, we have the new radar modalities that none of our competitors have, right. Radar, LiDAR will be quite important and how you fuse them will be quite important.
And I think there's a lot of innovation there that we're going to bring to the table. In radar, I know you've been asking about this on the calls. I can tell you we're making a lot of progress. So we just sampled the 1st radar device at CES. It's a Siggi, silicon germanium transceiver.
It receives a lot of attention, in fact, more than I thought it would. The reason for that is basically it's not only higher performing, highest performing now in the market compared to any of our competitors, NXP, TI, Infineon because we have the longest range because it's a long range radar product, but also it can actually do a 2 in one use case where you can reduce because of the performance that we have through that MIMO plus technology and reduce the BOM by 50%. So you can either use it as a longer range higher performing device or just cut down 2 out of the 4 and save on BOM by about 50%. That we think is going to help us grow about $90 maybe to $100 of content in the car, which is quite significant expansion And as the industry transitions from level 2, level 3 to level 4 and level 5, our relevance will become even more because we decided to enter at the very high end of the market and we intend to hopefully lead to continue to lead in that regard. There's a lot of design activity now.
We think the first revenue for that will be in 2021. That's because of course, as you know, the cycles in automotive for design are quite long. Let me show you this. This is kind of what we do in machine vision and edge AI that I keep talking about. So the first one on the top left is Amazon's new robot called Scout.
These are basically devices that they just launched it about a couple of weeks ago. A truck comes to your neighborhood full of these robots, I think 20, 30, 40 of them, they unleash these robots and of course they all have your package in them and just goes and delivers to the neighbourhood, brings them back, get in the truck and go to the next neighbourhood. We're in that. We have 14 of our sensors in that, which is total of about $50 of BOM content. In the middle, you have basically the drone makers, a product called Mavic Pro 2.
This is DJI, largest drone maker in the world. We have 4 sensors in that doing simultaneous localization and mapping, which is one of the most challenging things to do in that. Flat panel inspection on the middle bottom is something that we dominate. I think basically every mobile phone company, Apple, Samsung, Huawei, Oppo, Xiaomi, Vivo use our sensors to detect if their flat panel is accurate, is high resolution enough or whatever. We're in, of course, ring like devices.
And I didn't show it here and one thing I can tell you is that in Amazon, there is a division called Kiva. Kiva used to be another company, Amazon acquired them, we're in all of those robots also. These are warehouse robots. So in these robots, you basically pick up the racks at sometimes £750 and then bring them to you in these warehouses as opposed to going and picking things as somebody orders them. So we're in that about 2 to 5 sensors per robot and that's a fast growing market too.
That's about 30% sometimes depending on the customer 40% growth per year. Aquify on the top right is a company that does factory automation through visual cameras based on our sensors, they can detect if a product on the assembly line is high quality or not, for instance, sneakers or count the number of items on this factory floor for factory inventory automation. And on the bottom right is Bossa Nova and that's a customer of ours as well that has about 15 of our sensors and they do inventory stock reshelping and restocking, sorry, and inventory management in general. And I think 50 Walmart stores now have that robot as well. So we see quite a bit of growth there.
In industrial, again, shift, mix change and restructuring of the businesses, so low end security, we took it out and decided not to exit on it and not to emphasize it. And then on the CCD, some markets, we started the deemphasizing. That's why you see the downward trend on the revenue on the top line, but what we focus on is the growth markets like machine vision is growing and that's what matters to us at this point. TAM wise, it's everywhere. You see this, it's basically in all of the markets that I mentioned here, future retail, smart building, smart building is for instance, I'll give you an example, we're working with a company called Cree, who has which has about 25%, I think, of the industrial lighting systems in the world and they're now implementing our sensors in their lighting fixtures, so you can detect who is in the room, how many people are in the room and adjust the energy flow so they can save on power through HVAC.
So a lot of innovation, it's about a $3,000,000,000 TAM that you see by 2022, which is basically higher than automotive and the growth rate is also quite phenomenal. So that's the reason that we're focusing in this. And in machine vision, as you know, you probably know this, in small format pixels, we have unambiguous leadership. There's basically small format and large format segments of the market. Small format is 3 micron pixel size and lower and then outwards is called large format.
We had an ambiguous leadership in small format, but we didn't have that in the large format. So we decided to focus on it. So we rolled out Python about a couple of years ago and we started attacking our competitor, in this case, Sony with essentially speed. So capture speed was the highest people bought on it and then we priced it so that we can grow that business. It doubled in about 2 years.
Now that showed us that there's enough opportunity there. So we now are now launching XGS, which is our next generation basically Python and that now improves on the noise as I see here on the top right, which also now makes us either on power better than Sony and we think we're going to take even further share. It's important because it will allow us to get into intelligent transport systems with the same product line also, which is a very fast growing market, just as fast growing as machine vision, just as large as it, but we have no almost no presence there. So I'm quite excited about it and then all the rest on the bottom there that you see the 29 by 29, the portfolio breadth and support and system is what Sony is trying to copy us now because we already had that in the first launch with Python about a couple of years ago. Margin wise, we've done good, I think, as hopefully you can agree, but there's more to do.
So continue to do work on yield where we think we can have quite good returns there, test time improvements, package cost reductions, the back end, we ship close to about 400,000,000 sensors a year. So we have quite a bit of scale, so we can I think have economies of scale there that will work for us? And of course, ON is an internal manufacturing company. One thing I want to note is that in ISG, it's kind of the opposite of ON. We have 80% of our products manufactured outside, whereas in ON, it's the other way around.
So that means that we have a lot of opportunity to in source and where it's beneficial and improve the cost structure quite dramatically. And the most important thing is what you see there in the middle, which is mix, the healthy mix. So doing and building products that can command high margins like radar, like lidar and of course technology leader products in imaging that will help us drive the margins up. So I'm excited about the margin plan. And then finally, in summary, in ADAS, which is what matters in automotive the most right now, highest and fast biggest market and the fastest growing, we have a leadership position and we have I think a great wealth of technologies there and in total at ISG, just to leave you with some other number, we have about $3,500,000,000 of funnel in total, about $2,500,000,000 of that is automotive and about more than half of that is in ADAS, so about $1,500,000,000 So we feel good about that at this point.
The other number maybe I can leave you with in verbally here is that we have won basically 10 out of the 10 next generation ADAS systems out there. And we're also working on next, next generation and so far the scorecard there is about 4 out of 4. We're pretty proud of that also and you know of course who we are competing. We have great leadership. We're bringing new sensor modalities to our technology portfolio.
We are doubling down on machine vision. We have, again, a leadership position there and we're going to go after it like we have been going after automotive. And I think we have quite a bit of room to go do more margin from so we can give the value back to our shareholders. Thank you.
All right. Good morning, everybody. My name is Simon Keaton. I run the Power Solutions Group for ON Semiconductor. I thought I'd start off today talking about the importance of power semiconductors.
And to do that, I wanted to share something with you that I indeed have a problem. And in fact, I share this problem with you. And if you have kids or grandkids, you also share this problem. And that problem comes with climate change, and it comes with air pollution, and it comes with energy scarcity. So these are major world issues that we're dealing with.
If you've looked at the latest reports from the United Nations or the World Health Organization or the U. S. Government, in fact, you'll see it's a fairly damning picture of where we sit today. Now thankfully, there are several major market disruptions that are happening at the same time that are going to help with this. So it's the electrification of the vehicle.
It's the use of alternative energy. And the use of more efficient motors will significantly help in the reduction of the use of fossil fuels. And if you think about it, at the very heart of these market disruptions are power semiconductors. They're the very heart and soul of how these things operate and these technological advances that solve these big major problems. And I think that ON Semiconductor has a very unique position in that we will enable these market disruptions for two reasons.
Number 1 is our vertical integration and number 2 is our breadth of product. So let's take those 1 at a time. The vertical integration, we manufacture the vast majority of what we sell. And when I say manufacturer, I mean from seed crystal to wafer substrate epi, packaged device whether it's a discrete or a module and tested end product the full scale of vertical integration. And that gives us control over things like supply, quality, robustness, scalability, cost.
You couple that with our breadth of portfolio technology. And if you look at those market disruptions that I just talked about, you're going to need a variety of power semiconductor technologies to address those issues. You will need silicon, superjunction MOSFETs, insulated gate bipolar transistors, silicon carbide and gallium nitride. You will need all 5 at the same time to effectively address these issues. And ON has all 5 of those technologies internal in house that we're continually reinventing ourselves on.
And the net result is twofold. Number 1, it's a very new, very large and rapidly growing revenue stream of sticky high margin product. So that's great. That's great financially, which is good. And number 2, frankly, we get to solve some of the world's biggest issues.
And I know it sounds perhaps trite, but we mean it. We have a very diverse group, a very highly motivated individuals working on these technologies and applying them to these world problems and having a significant impact. That level of motivation to me honestly is a competitive advantage for us in the industry.
So let's take a look
in a little bit more detail what we're up to. So key takeaways. We have established leadership in power solutions. We are number 2 in market share in power semiconductors. If you look at the 3rd guy, we're more than 2x their market share.
So a very strong position and certainly focused on these secular hypergrowth markets that we've talked about today in automotive, industrial and cloud power. ON's power modules will enable the electrification of the vehicle. We're engaged with more than a dozen different Tier 1 and OEMs worldwide across all regions with our power modules for the vehicle, starting with IGBT based solutions, moving towards silicon carbide based solutions in the future. IGBT based modules go to production this year for the automobile and then next year with silicon carbide. Looking at the power content in industrial and cloud applications, we see really the push to Industry 4.0 moving towards much more efficient motors, significant savings in electricity usage.
Those motors that are out there today consume 45% of the world's electricity and they're terribly inefficient. The way that they work to think about it in the most simplest fashion, it would be like driving your car with your foot all the way on the accelerator. And when you want to slow down, you keep the foot on the accelerator and you just hit the brake at the same time. That's how they work today. They're not very intelligent.
So as the world moves to variable speed devices, it becomes much more intelligent and become much more efficient and we save lots of energy. And on cloud applications, we're seeing that build out of the 5 gs network. This will be happening in 2019. We'll double in size again in 2020 and we'll double in size again in 2021. So a very rapid expansion.
As I talked about very much headwind for margin improvement. Why? Because we're in that critical path. The very heart of these disruptive applications is driven by on semiconductor power solutions. So this is a very robust margin environment for us.
Power Solutions Group 2018, we surpassed $3,000,000,000 of revenue at 37% gross margin. If you look at those three focus areas of automotive and industrial and cloud power, combined it's about 60% of our revenue. Not surprisingly, it's also where the vast majority of our R and D spend is and subsequently the vast majority of our revenue growth over the next 3 to 5 years. But even in 2018, if you look at the growth from 2017 to 2018 in automotive, industrial and cloud power, it grew at about 15% year over year for us compared to the rest of the markets that grew at only 7%, so a significant rate of acceleration of growth. We have leadership positions in automotive and industrial and cloud power, which is frankly a testament to the technological capability and the breadth of our portfolio, allowing customers to put us in that leading position.
Strategic intent and goals. I'd say, 1st and foremost, our priority is relentless innovation in power semiconductors and packages. So if you look at it again in the power semiconductor areas, we're innovating. It's silicon, superjunction MOSFETs, insulated gate bipolar transistors, silicon carbide and GaN at the same time. We're not content to rest where we are.
We're always working on the next and n plus 2 generations of those products concurrently. You can imagine the scale and size of that engineering effort. It is a competitive advantage that not many other people can do. Additionally, I talked about that vertical integration applied to it that allows us to control things like our cost structure, supply, quality and reliability. Then we get into the packaging side of it.
You think about the modules that we have. We shipped over 100,000,000 modules in 2018 alone, everything from 25 watt to 150 kilowatt modules. And that's before I even start talking about the world's largest and most technologically advanced discrete product portfolio in the world that we can apply to our power products. We are seeing an inflection point in power semiconductor technology with the advent of silicon carbide and gallium nitride wide bandgap capability. Again, we have a playbook of how we apply technologies to our manufacturing space.
We like that vertical integration. We are most likely going to do that same thing there, including substrates and epi. We're very strongly positioned for this explosive growth in the electrification of the vehicle. Yes, we talk about the traction inverter that is the very heart and soul of performance of the end car. But there's more than just the traction inverter.
There's onboard charger. There's DC to DC. There's auxiliary motors. There's battery management. And so to effectively cover that entire electrification powertrain, you need all 5 of those power semiconductor technologies, silicon, superjunction, IGBT, silicon carbide and GaN.
And we have all 5 that is a unique competitive advantage. At the end of the day, I've talked about these very disruptive markets. ON is not only going to participate, but we in fact enable those. And that's why we're driving so hard with our rapid expansion of the portfolio. So how we win?
It boils down to technological capability that is extremely important. We have the world's best IGBT and MOSFET based solutions. On top of that, it's having the supply of those products, it's having the quality, the reliability and the availability as well. We have an extremely broad product range in the Power Solutions Group, over 21,000 different products, a broad range. Let me just show you a quick visual.
I'll salt shaker. There's 500,000 in these. You can just shake them out in your hand. You can salt shaker. There's 500,000 in these.
You can just shake them out in your hand. You can come up afterwards and shake them out and kind of dust them off. This isn't the smallest device that we have. We have something called an X4 DFN. It's 0.18 millimeters by 0.2x0.4.
You put them in this regular salt shaker, €4,000,000 The reason I didn't do it, I couldn't figure out how to get them in there and I was concerned I was going to inhale a few on the way, so I didn't do it, but extremely small devices. And then let's go for the other direction and let me pull out one of our traction power modules. So this is for those on the webcast probably a 6 inches by 4 inches power semiconductor module. This is 150 kilowatt module that would go into a traction inverter. This uses an industry standard footprint, right?
So there's basically industry standard as a gel filled module. These go into production later this year with IGBTs and then next year with silicon carbide. We're already sampling silicon carbide today, these modules to end customers. The interesting thing so this is 150 kilowatt module. And depending on the performance of the car, let's say you have a 300 kilowatt motor that you're driving to, you would need to stack these next to each other in series.
And it's pretty heavy. You can come up and check it out. That's about an 8 by-eight inches sort of area. It's pretty large. And so one of the things that we're doing besides just developing these industry standard models is we're coming up with something called a dual site cool module.
This is equivalent to 2 of these. So this uses what we call a dual side cooled module. It's the industry's most dense, smallest electric vehicle module on the planet. So if you have 2 of these next to each other, this is equivalent. It's about, I'd say, 40% lighter and it's about 70% less area.
Nobody's ever done it on the planet. This is 150 kilo this is actually a 300 kilowatt module. That's over 150 kilowatt in these dual site cool modules. So bringing that relentless innovation to play. We'll have the industry standard.
We'll have something unique. And then we'll reinvent ourselves again and again and again. And in that way, our end customers can make a unique solution on the market. And you're welcome to come out and play with these later on. Manufacturing footprint and scale.
I won't steal too much from Bill Schrom, but we shipped 66,000,000,000 devices last year in PSG alone. And again, we have that vertical integration to boot. Then lastly, the focus on the critical applications in automotive, industrial and cloud power. We have a very large power lab here in Phoenix, as you can imagine. If you want to walk through, I'd love to take you through there.
But we don't just have the power lab in Phoenix. We have that power lab in Detroit. We have that power lab in Shanghai. We have that power lab in Munich. And then guess what, we have 15 joint engineering labs within customers in automotive and industrial space.
That will be up to 20 by the end of this year. So what's the purpose of those joint engineering labs? There's two reasons. Number 1, these are very complex products, right? They're not easy to design in.
Customers want to get them designed in quickly and efficiently. So we work in the same labs at our end customers to help them design in the products. Number 2, and almost more importantly, we get the future vision from the customer of where they want to go. So this allows us to future proof our development. Why?
Because we have the relationship with those end customers. We're in their labs and we're thinking about where do you want to be 5 years from now. And then we translate that into solutions for their problems. That is a unique advantage for ON Semiconductor and something that we're continuing to scale up 15 today, 20 by the end of the year, it will probably double again within 2 to 3 years. In terms of the markets, automotive, industrial and cloud power.
Automotive is about 25% of our revenue today, but is growing very quickly. I'd say within 3 years, that will be north of 30 3%. Dollars 8,000,000,000 worth of TAM growth rate at 7%. Primary growth driver here will be the electrification of the vehicle, for sure. On the industrial side, we're looking at 28% of our revenue growing to about 30% over the next 3 years, TAM CAGR of 9%.
All three of those are key growth applications, alternative energy, efficient motors and EV charging stations. Lastly, on power, 7% of PSG revenue today growing to about 11%, TAM CAGR of 7.5%. Key application here will be 5 gs. That infrastructure is being built out in 2019. It's not an if, it's not a when, it's happening today.
We have the orders on the books. If you look at the transformation, over a very short period of time, PSG has transformed quite radically. 2016 power was only 45% of our revenue. Now through organic and inorganic acquisitions like Fairchild, we change fairly dramatically in a short period of time. Fast forward to 2018, 67% of what we sell is in the power arena.
And here, we've seen an incredible pull for our products. Again, we have all those power switching technologies. Now we've gotten pretty heavily into modules. 2018 alone, over 100,000,000 modules shipped by ON Semiconductor. There will be even more module expansion by 2020 2, and you see power at 76% of our revenue.
Now what you may be thinking here is, well, what about nonpower? Are you pushing that to the side? Absolutely not. We're still innovating in that area. And in fact, the absolute dollars in non power are growing.
Yes, the percentage of share of revenue is shrinking, but absolute dollars is actually growing. It's just power is growing that much faster. But we're going to innovate across both. Automotive business. So from 2017 to 2018, the number of cars sold actually declined by about 0.5% to 1%.
At the same time, PSG's revenue in automotive went up by 9%. It talks about the expansion of our product within the automobile. Certainly in Body and Comfort, where our medium voltage FETs go into the brushless motors for things like seats, windows and pumps. And then you think about Tanner's products, all those sensors, we provide all the supporting componentry, things like power management, things like signal quality, things like protection. So about $5 to $15 of content per sensor that Tanner makes available, we wrap around.
And so it's a complete holistic solution. We've been in automotive for a long time. It is our legacy all the way back to Motorola days. PSG alone shipped 20,000,000,000 devices for automotive last year. And these are relatively small devices, probably a couple of millimeters each, 2 to 3 millimeters.
Dollars 20,000,000,000 of those wrap around the earth twice. We're talking about huge volume production of quality products, and that's what we do. Looking at the growth, certainly driven a lot by the electrification of the vehicle. As David showed earlier, going from $40 of content to north of $400 of content per EV. We're also seeing growth in body and comfort.
We have number 1 share for electronic power steering modules and discrete devices. As cars move to that level 3 and level 4 based automation, you're going to get replication of those as a safety backup driven by the ISO standard, ISO 26,262, and that's driving massive MOSFET increasing growth there. Diving in a little bit more, the specific components within the car, dollars 400 of content in that complete powertrain. We look at the major portions of that powertrain. Traction inverter is the biggest and the one that people are most familiar with, but you also have the onboard chargers, you have auxiliary motors and you have DC to DC.
And what I'm showing within those blocks are the types of products that you have to have in order to service the complete electrification of the vehicle. It is silicon. It is superjunction. It is IGBT. It is silicon carbide and it is gallium nitride.
You need all 5 to have that full suite of a solution and that's what AANG customers are looking for. So we're seeing very good traction, I guess pun intended, in the traction inverter module at both silicon today for IGBT and then tomorrow with silicon carbide where we're already sampling over 10 customers with silicon carbide modules for traction inverters today, both the gel fill, those big white ones and then the transfer mode of the smaller black ones. So within the electrification of the vehicle, if you look at power semiconductor content, I'd say about twothree of the content will end up in that traction inverter. But even through now through 2025, we think the vast majority of those traction inversions will still be IGBT based. So 80% in IGBTs, probably 20% in silicon carbide.
The nice thing is that we are fairly technology agnostic. We will have both. We will help customers make the transition and that's exactly what we're doing. The customers that go to production today with IGBT, we're already sampling them with the same module with silicon carbide, and we enable a very smooth transition for that end customer from IGBT to silicon carbide. If you look at silicon carbide in a bit more detail by 2025, about $1,000,000,000 market.
There's some intrinsic material properties with silicon carbide that make it very good for power products. The energy density is higher. It's faster switching. It's a higher breakdown voltage.
At the end
of the day, you need to apply both a discrete and a module solution to this. So by the end of this year, we'll have one 120 different silicon carbide diodes, both 6 50 volt and 1200 volt. We've already released our 1200 volt silicon carbide FETs. We'll have 900 volt by the end of this year. We'll have 1700 volt silicon carbide dials by the end of this year.
We're already sampling to customers an automotive 1200 volt and 900 volt silicon carbide modules. So we have a very large portfolio that we're expanding on very rapidly, and we will help customers make this transition at the rate of change that they want to make. And guess what? We're sitting in their labs helping them calculate the efficiency that they can get from silicon carbide. And depending on the architecture of the car, it will be different for different customers and their approach will be different.
But having all the power technologies allows us to create that unique value position for that specific end customer. Okay. Switching gears to the industrial side of the business. We've seen rapid growth here, specifically from alternative energy. We have 30% share in the solar inverter market with our solar modules.
Going into motor efficiency, this is where we have our intelligent power modules going into both commercial and residential HVAC applications. We sold over 80,000,000 of these intelligent power module devices last year alone. Looking into sorry, jumped ahead. Looking into the future, on the left, we're showing where we were and where the systems were. On the the systems were.
On the right,
we're showing where they're going. Things like coal to
solar, grid stabilization, gas pump to EV charging station, industrial motors to variable speed drives, coal furnaces to commercial HVAC and warehouse to robotics. Interesting to note on the left where we were, we had zero content. There was no need for the power semiconductors that we have. We have over $2,000 worth of content in this new infrastructure revolution. And I'll pick on a couple just as an example.
Those industrial motors I talked about at the top right, dollars 40 of semiconductor content, They consume 45% of the world's electricity, 60% more efficient if you go to variable frequency drive. It equates to about $1,700,000,000,000 of savings if you could snap your fingers and everyone changed today. If you look at the amount of savings of energy, if everyone magically changed to these variable speed devices, it would negate the need for any nuclear power plant on the planet, right? We're talking about significant environmental change here that is going to happen over a period of time, and we're helping that move forward. These things also tie together very closely.
If you think about it, the number of electric vehicles is going to significantly increase. As a result, you need more EV charging stations. Those charging stations, especially when you get into the high power charging station, 2 50 kilowatt and above, over $500 worth of content. Well, guess what? You put those charging stations into the grid, especially when you go to China, you're going to have grid issues.
And so now you need grid stabilization. There's over $8.36 of energy storage content for ON Semiconductor with that grid stabilization. So we benefit from the car. We benefit from the charging. We benefit from the grid stabilization.
And they all build upon one another. So you get this recursive event of growth, which will push us significantly higher. Silicon Carbide Industrial Applications. The good news here, I talked about that number one position that we have for the solar inverter. Those today are IGBT based modules.
Now we're seeing the use of hybrid modules that uses an IGBT with a silicon carbide diode for better efficiency performance. Those IGBTs will move to silicon carbide. We've already sampled our first silicon carbide inverter for the industrial application for solar and those go to production in late Q3 of this year. We're also doing the same for power factor correction moving from a traditional silicon rectification to silicon carbide. It allows you to get to that titanium level status of efficiency.
Switching gears once again into cloud power. For 2018, we're seeing significant growth here on the server side. And also 2018, I would say, was establishing the base footprint for the 5 gs expansion. So we're working with all of the top 4 major 5 gs suppliers that are going to supply about 85% to 90% of the market. So we've got very close and tight relationships there.
Massive demand in that architecture for our MOSFETs, and I'll talk about that in just a second. Huge growth rate. So we're seeing 2 47 percent growth in 5 gs. That's primarily on the medium voltage MOSFET. We expect that revenue stream to double from 2019 to 2020 and double again from 2020 to 2021.
At the same time, in the server power supply, people are using superjunction FETs today. That's a very high voltage MOSFET with particular performance capabilities. That will transition to silicon carbide over the next 3 years as well. Looking at 5 gs, I know David spent a little bit of time on this. I'll just talk about 2 things.
2 big transitions going from 4 gs to 5 gs. 1 is that grid array. Instead of being 2 by 2, it goes to something called massive multiple input, multiple output. It's just a fancy way of saying I'm going from 2 by 2 up to 64 by 64. As you do that, each one of those will need MOSFET content, so it grows dramatically.
Then also to get the latency using line of sight and beamforming with 5 gs. That means more cells per square area than you had with 4 gs. So you've got more cells and then each one of those has this grid array structure means a significant increase in content going from $9 to $144 And we really didn't participate in those 4 gs networks. But we're at the vanguard and leading in 5 gs. So at the end of the day, what happens to the margin is significant improvement.
Why? We are at the critical heart of these very disruptive applications, right? If you look at the car, the electrification of the vehicle, you look at the solar network, you look at the efficiency of the motor, 100% that is determined that capability by the power semiconductor. So it allows us a very rich mix of product. We're also going to take advantage of 2 things, a movement to more and more modules.
I've shown you our internal capability for modules and how we have a unique value proposition. We also have the internal capability as we move to wide band gap, both silicon, carbide and gallium nitride at a higher gross margin as well. Talked about that vertical integration, that manufacturing capability. And again, not to steal Bill's thunder, but we don't just innovate on end products. We actually look at how we make those products and we innovate every step along the way.
One great example of this is using plasma etching. This allows us to significantly increase the die per wafer reduces our cost. So up to a 35% increase in die per wafer using plasma etch. Just one example of one part of the manufacturing process that we completely control and then we relentlessly innovate on the manufacturing process as well, not just the end device. So key takeaways.
We have established leadership in Power Semiconductors. We are very strong number 2 and growing very quickly. Massive opportunity in the electrification of the vehicle, very well positioned today with our modules, both IGBT and silicon carbide, but it goes beyond just the module. You need all 5 of those switching technologies to actively engage in the complete powertrain electrification. Then power content besides automotive continues to grow in industry due to Industry 4.0 and cloud applications like 5 gs.
Lastly, because we have those critical products, certainly a lot of headroom here for margin growth and expansion over the next several years. Okay. With that, I believe we have our next Q and A session.
So this is a Q and A session for business units. So if David and all the business units guys can step up to the podium, please?
I've got these parts that come out everywhere. Speaking to my chest.
So two quick questions for Tanner specifically. First, I guess a simple one. Is the pruning done? And how do you combat the fact that more and more products there's always people chasing you and things can commoditize at the bottom end of your stack because the promise of growth that you have is great in your target markets. But as you said before, some of the mix changes in pruning impacted that.
And then the second question in your business, how important is the non CMOS image sensor business to your customers? Do they view on as primarily CMOS image sensor company and then they still think of others with the radar and the LiDAR side of things? Or is those two technologies exceedingly important for ON's business going forward?
Okay. So the first one, pruning is not completely done, but largely done. There's a few more things that we have to do. Nothing compared to what we have done so far. So I think it'll be more easily manageable than we have done in the past.
As to how do you compete with the bottom feed, basically, I can tell you Ross that we have not had the scale even though our market share is great in these two markets, automotive and industrial. We didn't have the economies of scale working for us in the last maybe 3 or 4 years ago. But we have built that scale now, and we now can compete with basically cost and price all the way to the bottom of those markets. So we're winning in viewing, rearview cameras still. We have the best product there and there's a lot of innovation there as well because if you implement the system level innovation, they save tremendous amount of cost.
It's not just the sensor anymore. So we're putting in, for instance, SoC features, the de warping features that they want. We're putting in some little processor in there and also enabling them to do a much easier cabling to the rest of the car. So they can reduce the cost and weight and EMI and so forth. So I think we're in a much better position to fight the scale and the bottom feeders in these two markets, automotive, industrial than we have ever been before.
Radar, LIDAR being important or not, it's absolutely incredibly important, because people believe that in fact, the reaction shows that we were nobody of course in radar, people were questioning whether we can do it or not. There's 2 big guys out there, how can you compete, even the third one who had some shares kind of disappearing. While we told them we're going to just innovate, if you like innovation, we're going to show you what we can do and by God, I mean we have I think surprised the market quite a bit. So everybody that we talk to in the ecosystem LiDAR, incredibly important as well, because I believe in sensor fusion. I think we're going to have to have all three sensor modalities to be able to completely autonomously drive cars in the future.
Many people are implementing like for instance even Mobileye as you might have heard Amnon talk about it in the industry, fault tolerant systems too. So you don't have to necessarily rely on all of them coming together, but people are now implementing, for instance, what they do with cameras completely and solely with radar and LiDAR, so they can completely duplicate what they have done in camera. So having 3 sensor modalities and being able to combine them at the edge, like I explained, is going to be very important. And one thought I leave you and I can't mention this completely here, but we have a vision to be able to combine particularly LiDAR, which is all on CMOS, if you will. That's one of the reasons that we acquired Sensil because they have built a perfect process for 14 years that works and they have achieved the comms of scale with imaging, which is also in CMOS.
So if we can do that, the fastest we can do that, then I think we'll be in a much better position. So that means basically having depth and color at the same time from a single sensor, which I think we can do.
Craig Ellis, B. Riley, FBR. Simon, my question is for you. Clear message on the strategic import of auto and other businesses. But my question was really around Communications and PC.
Those in the past have been strategic growth business for ON. Can you clarify their role in the portfolio? Should we expect them to be growth drivers for the company? Are they really evolving into businesses that would be more cash cows and managed for strategic decline? Thank you.
All right.
So I can certainly address that for PSG and then David, maybe you want to talk about broader on. So within PSG, we have our focus areas and that's really driving the technology development. What I've seen is that a lot of the capability that we get from that investment is a trickle down effect.
At the end of
the day, what people want from PSG and Power Solutions specifically is 99% of the time is better efficiency. So when we do something for the 5 gs network with our MOSFET, so we have world's best MOSFET, 25% less power loss than the next closest competitor. Guess what? That capability can then be applied to things like computing. So there is a trickle down effect that happens, and we naturally get some growth from that R and D investment in another area.
So I'm not investing specifically in that computing area, very focused, but I do get that trickle down effect, and I expect to at least maintain share, if not grow, just from the fact that I've got the best MOSFET on the planet, why wouldn't I grow there as well?
Yes. And just to finish up on the question, I'll take the mobile side of it and turn it over to Vince for the client computing side. On the mobile side, we have continued content increases that we'll continue to have products for. It cuts across our DC DC converters, our power management ICs. We do the optical image stabilization and the autofocus technologies that go into the increasing number of cameras that you see in smartphones as well as USB Type C that's now proliferating across smartphones.
So we'll continue to, in those focus areas, deliver solutions that keep us in mobile and benefit from some of the content gains that we're seeing.
Okay. So for the client computing, I mentioned in my presentation that we're actually leveraging the multi phase capability with that business into the cloud business. So really that's how we were able to move so fast the last couple of years, especially around the digital controller effort is because we took part of that team and that expertise and applied them toward cloud. So we're going to maintain market share around desktop and notebook, but we did deemphasize a little investment, put it toward cloud, leveraging the capabilities that we get from supporting the computing market. So really a benefit for us.
We're going to keep supporting it that way and keep leveraging it between it and cloud. A lot of the customers tend to be similar as well. So it helps
us that way. Hi. Just a couple of quick follow ups on ISG. Just if you go back 2 years and you excluded the effect of the mix and the pruning from the portfolio, what was the underlying growth rate of the business, taking that into consideration? And as a follow-up to that, can you talk about the pace of cost reductions versus ASPs in that business and what that implies for the margins of that business going forward?
Yes, the growth rate, I think I showed it in the slide. So the legacy was grayed out. And if you take out the legacy is what you see there. I think it's approximately 20% in that range. And second question, my apologies.
Margin, yes, price reduction. So we do see ASP effects, obviously. I mean, it's automotive. It's now becoming a scale business. We ship, just so you know, about 100,000,000 sensors now to automotive.
But it's nothing out of the ordinary that ON doesn't know how to deal with. It's in the same range, I would say. We obviously track this every month, every review that we do. I don't see any measure there, problems there, but I also want to point out again that we have a lot of cost reduction opportunity ahead of us. Again, 80% outsource, that's a major game changer if we do that gradually.
And we as we scale up and with 60% market share, now you can comment from your supply chain better pricing, better cost improvements as well, which is also what we're seeing. So as I answered Ross, the scale starts to matter now and that I think is one way to mitigate that risk that you mentioned, which is anyway in line with what the rest of the market is seeing.
Question here. Harlan Sur from JPMorgan. Question for Simon. You did a great job of outlining the opportunities for some of the wide band gap products, right, silicon carbide, gallium nitride. You also at the same time highlighted the differentiation of your segment because of the vertical integration.
So, on the wide band gap material, my understanding is that most of the substrate and epitaxial deposition is procured. And so I'd love to get your thoughts on bringing all of that capability in house and over what time?
Yes, okay. So as you know, we've been doing semiconductor substrates now for decades, right, all the way back through Motorola legacy. It's something that we know how to do very, very well. We are looking at bringing silicon carbide substrate capability, and currently, we're already working on that development. Now the timing of which is still yet to be determined.
It really depends on the economics. But I can tell you today, if you come in on semiconductor, we can show you silicon carbide bulls at this gentleman right here. That's Hans. He's our CTO. His team is already making silicon carbide boules.
Now we are sourcing from the outside today. And even if you look at silicon, we have a combination of internally sourced and externally sourced. We'll probably have that model continuing forward in the future. It's just the right economic timing to do so. But technology wise, we're already working on that development.
Great. Thank you.
It's John Pitzer with Credit Suisse. It's probably best to ask Dave this question because it kind of just spins across a couple of product groups. But just on the 5 gs, cloud power is a relatively new segment breakout for you guys. I'm just kind of curious how we should think about the rate of adoption or your penetration growth into that business. You've got guys like ADI and Xilinx.
ADI's comms business is much larger. It grew at 44% year over year. Xilinx's wireless revenue infrastructure revenue is higher today than it was at the 4 gs peak. Can you just help us understand kind of how we should think about the inflection of growth in your business? And maybe talk about in terms of base station OEMs that you have penetration with or how should we think about from a service provider spend perspective?
Sure. I'll take that and then they can add in if they'd like. If I just look at our growth year on year and what we consider cloud power from the telecom infrastructure side, primarily driven by 5 gs participation and in the server side of the business, 2018 over 2017, we were in the neighborhood of better than 40 percent. So we saw a significant lift in that business. With 5 gs just being early in its deployments, a lot of that came from the server side of the business.
As we look forward, we think it will accelerate in 5 gs because of the architectural changes that Simon and I described during our presentations as well as Vince. So I think that could accelerate and being new for us where we didn't have much participation in 4 gs, it's going to be a substantial content driver and increase in revenue opportunity for the company. So that's primarily where we see it coming in for us. So we actually have supply agreements with all the major telecom equipment makers and the server companies as they look at the content the need for the power devices, particularly in Simon's side of the business around medium voltage FETs. They want to be certain that they have the supply continuity to support their investments in infrastructure.
So the only thing I would add is that with the VR13 release for our Power Stage, we as you know, we didn't have a digital controller for that. So we're leveraging other companies' digital controllers. So we pretty much have been a second source in a lot of those applications. Moving into VR-fourteen, our strategy is to be 1st source. So we're going to see a big jump as far as demand goes for us with that capability having both the controller and the power stage.
A question on the discrete in your business and how that can kind of shape for when we look within the context of the 2022 model? And maybe related to that, a lot of some of your auto customers are looking to get into the module business as well. Can you talk just about that, how much of a threat that is? And maybe lastly then, maybe could you give us an idea on the margin difference between the bare die or the squeeze versus the module? Thanks.
Okay. Maybe I'll take it a little bit
in reverse. I'm not going to comment on specific margins. I'd say the margin is robust in both areas, whether it's wafer sales or module sales. If I look at the modules in terms of revenue today, again, it's dominated primarily in industrial. We're seeing some in automotive.
But in those electronic power steerings, it's probably 20% -ish sort of revenue. That range will grow significantly as we get into really the electrification of the vehicle and more into the things like the traction inverter and onboard charger. Certainly, the traction inverter, it is going to be heavily dominated by modules, and we're talking about 100 of 1,000,000 of dollars over the next 5 years. So it will be significant. Was there a last portion of the question that I missed on that?
Yes. So it's obviously, I think people are trying to figure out their position at the end of the day of how they're going to play in this business. One of the things we subscribe to is what our competitive advantage is, which is that vertical integration, right? There's people that just have the module and they don't have the semiconductor supply or there's people that have the semiconductor, but they really don't have the module. We think both of those are going to be tough sell in the future.
People want somebody that's completely vertically integrated. And doing semiconductors is tough, right? People ask me questions. Are the car guys going to get into semiconductors? Who knows?
This is not an easy market, as Keith mentioned, and we've got decades of experience. So we'll see. We think we've got a strong competitive advantage, and we're prepared to win with that.
So folks, with this, we will end this round of Q and A. We can take a 10 minute break and reconvene at 11:10. Thank you. Hey, Bert, we're getting started. Why don't you walk through the podium and get everybody rounded up All right, folks.
Let's get started now. Now we have our COO, Bill Schram.
Good morning, everyone. So I'm Bill Schram, our Chief Operating Officer here at ON Semiconductor. You've heard from my colleagues this morning of all the great opportunities, the great products we're delivering, the great products we're going to deliver in the future. And so now you get to hear from the team that has to build it and ship it. And so that's what we do.
So consistent with the other presentations, we have the some of the 4 key themes we'd like to leave with you today. Keith mentioned quite a bit about the prowess of on semiconductor from a manufacturing standpoint. We are a semiconductor manufacturing company. We believe that we have decades of experience and we believe that gives us a true competitive advantage. So if you look at the breadth of technologies, processes, packages and products, 84,000 SKUs that were mentioned earlier, that's incredible strength we bring to customers.
It makes us very important to customers. And as we grow, that importance to customers just keeps getting bigger and bigger. Combined with the scale of manufacturing we have, it just really makes it very difficult for someone to try to duplicate. So we believe this is a true competitive advantage. I'm going to talk about 300 millimeters.
We get asked that question a lot, what's on think about 300 millimeters wafer production. I will talk to that. But in a nutshell, we think to derive the benefits of a from a 300 millimeter fab, it only happens in the right economic model. And I'll talk about that in future slides. And finally, you're going to see here that we're investing capital, not just in capacity, but investing in the tools and systems and people to be able to keep that advantage going in the future and deliver best in class quality, service and cost.
So why do we think manufacturing is a competitive advantage? Well, of course, there's the scale. So we have 12 wafer fabs, 9 assembly test sites. We produce billions of units every week. You've seen these units, as Simon showed, that are smaller than an ant to the big modules that we're working on and delivering and everything in between.
And so it's quite an impressive scale. I'll talk more about the factories in a minute. But one of the key things about being a manufacturing company is we have control of our destiny. We don't have to worry and we don't about the are we getting enough mind share in our foundry or a subcon assembly test site. We have 25,000 manufacturing employees thinking about on semiconductor.
So with this control, we have the ability to and flexibility to add capacity in our network in multiple places and source products from multiple factories. So what does that do? It gives us flexibility in servicing customers in especially with quick upsides, which we see, but also gives some risk mitigation protection from a dual sourcing standpoint to customers. It gives us control around how we operate our factories and optimize for costs, optimize for delivery and optimize for quality. And we're playing in the automotive, industrial, cloud server market.
These are harsh environments for semiconductors. So quality and reliability is key. And so if we look at our world class team last year, we broke another record for quality from in terms of quality incidents per 1,000,000,000 unit shipped, we operated at 138 parts per 1,000,000,000. Now it's not 0, which is our goal, but it's very competitive at that level of quality. The other thing with our internal processes we're allowed to do from a new product development standpoint, we can accelerate our cycle times, so that we can turn our engineering development efforts faster, do more of them and allow us to accelerate our release of new products.
Last year, the company introduced over 400 new products. Think about that, that's over one product a day on average being introduced into the market. And finally, we're able to fine tune and optimize processes, especially in our analog BCD flows, but even in our power MOSFET and IGBT flows to really give that distinct advantage that you heard Simon and Vince talk about is eking out better power efficiency. We're able to do that with our internal flows. So here you see a pictorial view of the factories.
We have 25,000 people working in these factories. The footprint here represents about 9,000,000 over 9,000,000 square feet of space. And these are sites that can be expanded, which is one of the advantages we have and then how we compete going forward. We shipped 76,000,000,000 units last year. We're capable of building more than that.
In that 76,000,000,000 units that we shipped, about we'll say 80% of it was done internal. But even when you think about the 20% that's external, it may be that we used a foundry for the wafers, but we assembled and tested them in our site. And maybe that we produced the wafers even though they went assembly and test in an external site. So the influence on our manufacturing is heavily dominated internally. And at the end of the day, all these units come back to us for packaging and shipment to our customers and our distribution hubs.
Now these factories, you'll notice if you look at some of the locations here, they're in very low cost locations. And that is something that as we go forward, we'll continue to look at how we transition to the most efficient low cost factories going forward. These front ends here have the capability to run our advanced analog BCD flows, our IGBTs, advanced IGBT flows, advanced power MOSFETs. The back ends, we can produce over 1,400,000,000 units a week. So very, very high scale, highly efficient manufacturing.
Now if you look under where it says front end and substrate facilities, in that top right corner, you see it says Czech Republic substrates. That's the facility where we internally produce 150 millimeter and 200 millimeter silicon substrates. So over the last couple of years, you've seen all semiconductor companies have seen shortages of supply of substrates and increased prices. ON was not immune to increased prices on substrates over the last 2 years. However, I believe we're the last semiconductor manufacturer that's producing its own silicon substrates.
That means growing the crystal, slicing the wafers, shaping, polishing, growing epi on top of it. If I believe we're the last we're certainly the last at this size and scale. We can produce 40% to 50% of our internal silicon needs internally. So last year on that portion of our needs, not only did we not see a price increase, we were able to further reduce costs as we do every year. So this is a very specific advantage to ON Semiconductor that we have that vertical integration.
So we talk about the leading cost structure. What are some of the things we do in the back end? Well, 1st, we talk about test methodology, both in wafer probe as well as final test. And a key component there is shrinking test times and parallelism of tests. So instead of testing one device at a time, testing multiple devices at the same time and that reduces your overall test unit cost.
And we can in some cases go up to testing 256 units at the same time. So that becomes highly efficient from a test standpoint. Vertical integration, I just talked about the silicon. Simon mentioned the modules. So there are module makers out there that don't have their own die.
They have to go around and buy the die. There's others that have die, but don't have the modules. We have both. Many of our products module products, we can manufacture all the dye that's needed to produce the module. In addition, we have in house DBC direct bonded copper substrates, which go in the modules.
We're able to do that in house as well. So that vertical integration we think will give us continue to give us an advantage in modules. High density lead frames, so they're made of copper. We produce or we consume almost 1,000 tons of copper every month. So it's billions of units.
It's a lot of copper. The lead frame is usually in strip form and you have die mounted on the lead frame, then wire is connected and then they're molded, okay, and then singulated. So the more units you can put on a lead frame, the lower your cost, the less copper you're consuming. And so we have a very keen technology we've developed over time and high density lead frames and it gives us a very, very competitive advantage. Simon mentioned the technology development that we do around processes.
So he showed you the plasma die separation. Other example would be our wafer thinning process. So that's every wafer gets thin near the end of its production process. We have our own internal low cost patented technology that we believe gives us the lowest cost thinning methodology in the industry. And finally, there's just the sheer scale that we have, not only the scale of internal, but the scale that we use for leveraging our suppliers and leveraging external partners.
We see when we benchmark our costs against what it would cost to buy the same product from a subcon, for example, we see 30%, 40%, as much as 70% lower cost when we do it ourselves. And so that's a keen advantage we have when we manufacture internally. I'm going to talk a little bit about some capability that we've been putting in place at ON Semiconductor. We don't just spend our money on capacity. We also spend our money on capability and improvement.
This is our advanced integrated manufacturing roadmap. It began some time ago, but began with really benchmarking and identifying gaps to how do we become even better and with the ultimate vision to be best in class for service, quality and cost. And so in this particular area, we look at a lot of tools that we would invest in to provide outstanding control, control around how we move products on the shop floor, optimizing cycle times, optimizing delivery, yield improvement tools and software, productivity improvement tools and software. So we have all these this investment in the tools and we're going through that implementation now. And the next thing is now you have all these tools that give you data.
For example, FDC is fault detect control. So this is monitoring equipment and ideally it tells you when there's a change and when perhaps you're going to have equipment failure, so downtime or maybe it's going to produce scrap. And so also now we have all this control data that our engineers and technicians are working to improve processes and make good decisions. But the next step then leads me to the next slide around Industry 4.0. So this is where we take that control information and now we interconnect it and how we correlate it with the products we make.
So Industry 4.0 is really the automation and the data exchange of information. And some people will also refer to it, you'll hear smart factory. So it involves the interconnection of manufacturing equipment, sensors, devices and people. It involves the information transparency because now with all those control tools I've told you we were investing in, now they are we want to interconnect them. It provides all this information that was latent that we didn't have before.
Now it's transparent to us. So now how do you then take that and provide technical assistance so we make good decisions in the manufacturing floor. And the ultimate then is then into the artificial intelligence where we get to decentralized decision making or autonomous decision making. So and on semiconductor, we're building those technologies. We'll be implementing pilots over the next few years.
A key one that we have going right now is what we call predictive yield. So think about it, you've got equipment that's monitoring or tools that are monitoring how your equipment is performing, how your process is performing, identifying defectivity and you correlate that data then to yield downstream. And now you can feed that back and now you get to where you make autonomous decision making around what you do with the equipment to prevent scrap, improve our yield, improve our costs, improve our productivity. It also will improve our quality because you want to get to, we believe we can identify what we call those walking wounded where product that goes out test, test good, but ultimately fails for some defect in the field, we believe this will drive quality improvements as well. Okay.
Thoughts on 300 millimeters. We get asked a lot about 300 millimeter wafer fabs. We've done modeling where we look at a greenfield, you're talking about $1,500,000,000 to $2,000,000,000 investment. We don't believe that that's going to drive the type of return that we want to see it on semiconductor. So we are very open to acquiring a 300 millimeter wafer fab.
We see benefits for doing that, but we only see the benefits on the right economic model. So our 300 millimeter strategy is all based around on the right economics and we will have patience and discipline around that. And someday I would expect we'll see a 300 millimeter wafer fab in the network, but only under the right economic model. In the meantime, we have a very competitive cost structure that will that we can continue to expand and make more competitive. We're not I think Keith mentioned we have more optimization we can do and we plan to do that.
And so the model that you've seen and what Bernard will be talking about, whether we have a 300 millimeter wafer fab in the network or not, we will achieve that model. We are competing today against companies that have had 100 millimeter wafer fabs for years and we're competing successfully. So again, we'll achieve our model whether we have a 300 millimeter wafer fab or not. So what are some of the manufacturing gross margin drivers? If you look at the model, we'll improve gross margin from revenue growth, from product mix, from portfolio management and manufacturing efficiencies.
So we're planning for about 130 basis points in improvement in margin from manufacturing efficiencies. So of course, driving continued scale will help with absorption. We target all our factories target every year internal cost reductions above the ASP declines. So that will add to our efficiencies. We'll be looking of course, we talked about the materials, in house materials, but a lot of these internal cost reductions are driven by productivity gains, cheaper materials, leveraging our supply base with increasing purchasing power that we get.
Every time we do an acquisition, every time we find tremendous opportunity and synergies around our purchasing of materials, chemicals, gases, other supplies and that doesn't happen just the year of the acquisition. We usually drive that over a 2 to 3 year period. So still more to be done there. Expansion in low cost sites, we'll be expanding our facilities that we have in low cost sites, not only the footprint within a building, but even in some cases building expansion, we'll also be transitioning to more efficient facilities, whether it be 150 millimeter to 200 millimeter conversions or just optimizing in low cost locations. So key areas of investment from a capital standpoint, I think in 2019, it was about 9% of sales from capital intensity.
Our model going forward is to be in that 8% range. We're focusing most of our investments in cost effective capacity around automotive, industrial and cloud power. So the 8% is the average for those type of products in those markets that I just mentioned, you're probably looking at double digit, not probably, you are looking at double digit CapEx as a percent of sales. And then other parts of the business would be less. But on average, we're looking at 8% model.
So we'll be investing in the power MOSFETs, IGBTs, silicon carbide capacity, not only for the wafer fab capacity, also silicon carbide epi, which we already do today. So a lot of growth there. Cloud power, the other area I mentioned. I showed you the tools before. We're continuing to invest in new tools to be able to extend our advantages and improve our productivity.
That's not going to change. From a back end standpoint, we have 2 building expansions, 1 in the Philippines, 1 in Malaysia. So again, concentrating in low cost locations. And then we'll continue to invest in our internal substrate capability over the next couple of years. Outsourcing, so I talked about outsourcing earlier.
Again, it's about an eightytwenty model. We're very comfortable with that model. We think it works. We outsource where we think it makes sense. So it doesn't make sense for us to put in a bunch of deep submicron capacity with plenty of foundries that can support that very cost effectively.
Image sensors are for the most part external today. I can see that changing in the future. But the first part of it will be more likely the back end where we'll we have opportunities, as Tanner mentioned, to in source some of the assembly and test. And so we'll be working on that over the next couple of years. And then packages that are sunset or packages that are niche low volume and processes that are low volume we'll use outside manufacturing for that.
The other advantage we have with that eightytwenty model is especially very highly cyclic markets like smartphones, which peak and trough in the same year, this capacity gives us a certain amount of flex. So we're not putting internal capacity in place to catch the peak. We have a certain amount we put in, but have flex to get the big increase. And the other advantage that we have is in dual sourcing for risk mitigation, especially in the automotive space, server space, that's a key requirement from our customers. So in summary, the same four notes, again, we think we are a very formidable manufacturing company.
No one's going to duplicate what we have after years of experience. The scale absolutely matters. It's what drives our efficiencies and we'll continue to grow. 300 millimeter, again, we think there are benefits to a 300 millimeter wafer fab, but our objective is to implement a strategy at a very cost effective economical way. And then and service.
And that takes me to the end. So, and service. And that takes me to the end. So next is Bernard Gutman.
Thank you, Bill, and good morning, and thank you, everyone, again for joining us today. I will be covering the last portion of this Analyst Day. We'll give you 4 big takeaways. First one, the most important one is after you saw what Keith and my peers presented, we have a lot of very exciting plans and that will result in us raising our financial target for our new 2022 target model that will be aligned with our market and profitability outlook. And I'll go into a lot more detail, several pages on that during my presentation.
Now before I do that, I will probably I'll spend a few minutes in covering what we have done historically and what we have done in comparison to what we presented 2 years ago in our March 17 Analyst Day. 3rd topic I'll cover is basically committing to an efficient deployment of shareholder capital with intent to maximizing returns and shareholder value. And last, also very important for all of us, what we're presenting today is not the end game. This is a milestone. We are a company in transformation, and we believe that this is just milestones that we keep adding every time we present to you.
So having said that, let me talk about our progress against our 2020 target model presented in 20 17. As you can see, we delivered 2018 revenues of €5,900,000,000 already above what we had earmarked for our 2020 model. 2 very good years, 2017 on a pro form a basis at about 10% growth and 2018, 9%, all fueled by all the secular drivers that have been presented today. We also had a very robust 310 basis points improvement from our base in gross margin, achieving an average gross margin of 38.1% in the year, well on our way to achieve the 40% target. I'll talk about that in a little bit more detail later.
We keep making progress on our operating expenses. In 20 16 when we initiated our goal. We were at 22.7%. We are now well on our way, 40 bps away from achieving our 21% target that we put ourselves under for the 2020 model. Result of that is that we have a 16.7 percent operating margin, also well on our way to achieving the 19% that we have as a goal.
Tax rate has been as per plan. And with that, the resulting non GAAP EPS in 2018 was only $0.04 shy of our 2020 goal, dollars 1.96 Actually, if we look at the performance in our most recent few quarters and you annualize those, they are well above the $2 price a $2 non GAAP goal. Free cash flow has grown from EUR 370,000,000 in 20.16 to EUR759,000,000 last year. Again, we're on our way to achieve the EUR 900,000,000 target we have or we had set in 2017. There is always pluses and minuses.
And here is some description of some of the pluses and minuses of what happened throughout the last couple of or several quarters. Revenue, we always said, fueled by strong market and good secular drivers was much better than we thought. Pricing has also been more benign. Market has been more disciplined in terms of adding capacities and we I think also have a better process for pricing. So we have been doing better here.
But we have been also having to fight a few negatives compared to what we had in the model. In the model, we had committed to doing some consolidation in our factory network. The fact that we were running much higher than what we had anticipated didn't allow us to do that. The benefit of that is we still got the 50% fall through on the incremental revenue. So we traded one benefit versus another.
The mix was very good. We and I'll show you another chart on that. We are getting very nice contributions and tailwinds from the wind. However, in our Analyst Day model of 2017, we had assumed that consumer and computing were going to decline in a much bigger way than really occurred. As was explained earlier, we are getting the side benefits of designing in our core areas and getting some better traction and better revenues in consumer and computing.
So that was a little bit of a lowering of the mixed impact. Nevertheless, we still had a very substantial mixed impact. We already talked about earlier about raw material increases. We have been facing raw material increases that also were a little bit of a headwind. And we also, as Bill mentioned and we announced publicly that we are stepping up our capacity investments to 8% to 9% for 2018 2019 had a little bit more negative impact on free cash flow on that.
Nevertheless, our scorecard, we believe, is very robust. We did achieve and we're on our way to achieving our target that we set ourselves for in 2017. Now turning over to the new model, which Keith already gave you a preview and I'll go into more details on next several slides on that. We are going for a $7,100,000,000 2022 revenue, which implies about a 5% CAGR and I'll show the details on that, improving gross margin by another 500 basis points compared to the existing 38.1% to 43%, keeping OpEx at the about 21%, which is the same level we have in our 2020 model. And the resulting EBIT is 22%, which translates into a non GAAP EPS of $3 The free cash flow is $1,200,000,000 so a very nice improvement compared even to previous target model.
And we now go into a little bit more details into the revenue. Questions was asked before, how is our 5% CAGR made up? And here is the made up. Our biggest dollar in the market, automotive is expected to grow at 9% CAGR. Industrial at 6 percent, very similar to the growth rates we had earmarked in the last Analyst Day and in line with what we have been doing historically.
Cloud power is a new one we added and that one we're expecting with 5 gs and with the server cloud server growth to grow at a very nice pace, although in dollars it's a slightly it's a much smaller piece, but it's a 13% growth. And the other 3 end markets, communications, we are expecting to basically be flat. We believe we can still benefit from content increases and tailwinds from R and D investments in our core areas. But the units themselves, we believe, are maturing and will be will only afford us to be at about a 0% over the 4 year horizon. And client computer and consumer areas where we're focusing very little or actually none of our R and D dollars, we're expecting to decline at the 2% giga over the next couple of years over the next 4 years.
The path to gross margin, we said it's 4.90 basis points improvement in gross margin, 120 basis points will continue coming from mix. We have been very successful and have been able to prove that that is the case and I'll show you a chart in a second on that. As Bill mentioned in his presentation, we are counting on 130 basis points from manufacturing optimization throughout the 4 year horizon. We keep using the same 50% fall through an incremental revenue. This is a little bit bigger contribution than we had in last Analyst Day, mainly because our CAGR on the revenue is 5% versus previously 3%.
And we will also continue with our discipline of looking at low performing non strategic businesses and either end of life in them or potentially divesting from them. So when we look at that, we do see and have a bridge that shows how we can achieve our plan to improve to the 43% gross margin. And a good portion of it actually, the majority of it does not depend on the top line. Here is the example on mix. This is comparing our starting point when we did the previous model, the previous Analyst Day versus what we did in 2018.
And you can see that the dark blue, which is the better than average corporate average gross margin has substantially increased affording us to show the mix improvements. And we expect that, that will continue. And you heard about all the secular drivers in all of these areas in automotive, industrial and cloud power. I think we had questions on that earlier. Also, this is the comparison of our end markets today in 2018 where 64% of our end markets are in the 3 focus areas.
And you can see in 2022, it becomes 72% with automotive continuously still being the biggest end market at 36 percent, industrial at 28% and cloud power at 8%. Operating margin, the biggest contribution margin and another 40 basis points from the operating expenses. On operating expenses, we pretty much have earmarked and decided that spending about 21% is the right level for the product portfolio that we have with technologies like EV, silicon carbide, sensors, all of the radar, lidar. We believe that some of these new markets or newer markets require a higher R and D intensity. But overall, we think we can fit it still and improve to achieve a 21% and remain at that level.
What does this target model translate into a free cash flow? You can see here all the statistics. Our operating cash flow goes up to $1,800,000,000 We will keep paying down debt, although we'll not get to a 0 level. And you can see our interest expense will go down in a to about $50,000,000 by the 2022 year. Another important point is cash taxes will rise and this is more a function of us using up all of our NOLs.
Once that occurs, our tax rate will be in that 15% to 20% rate. We use the midpoint, which is 17.5%. And as Bill mentioned, capital expenditures, €575,000,000 that's an 8% rate. Our traditional historical model was 6% to 7%. Then we increased it to 8% to 9% for the 2018 2019 years.
And then after that, we are stabilizing it at 8%, which is higher than what we have today, and I'll show you some details in a second. The resulting free cash flow of all of that, as I mentioned earlier, is $1,200,000,000 so a very nice amount of cash to do stuff with, which I'll talk in a second. Here, capital expenditures, you can see our historical track record. We have increased substantially in 2017 2018, particularly in 2018. The growth areas do need more CapEx and they are a little bit more capital intensive.
So EV, cloud power sensors, power discretes in general are areas that we call a little bit more capital intensity. And we have also focused on investing a little bit in the materials area to lower the input costs as we can source it cheaper inside. So bottom line, the message is we
are going
to spend about 8% of our cash on CapEx. Let me switch over now to capital allocation. So our first big tenant of it is really continue to accelerate the virtuous cycle. We believe that investing in our own business, improving our competitive position will allow us to continually increase our free cash flow, which we can then reinvest in the business and have this nice virtuous cycle. We are really committed to capital efficiency.
So capital will be deployed in a manner that really maximizes the return for shareholders. And along the same lines, we will be balancing the risk rewards in the capital allocation process. Four areas of capital deployment. We talked about the first one on the top left corner extensively today, which is the organic growth of our business, reinvesting our cash in R and D, in sales and marketing and in CapEx to continually generate increasing amounts of free cash flow. Keith talked about the M and A area.
It is an area that's in that we are that we will continue being a critical component of our strategy. Obviously, we talked about the fact that we have high hurdles from the both the financial metrics as well as the strategic metrics in terms of deciding to do it or not to do it, but it is a part of our overall capital deployment strategy. We do have share repurchases. We do have a commitment to returning capital to shareholders, and we'll do it via share repurchases. We did exhaust the or had an expired plan in last December and had our board reapprove a new $1,500,000,000 4 year plan.
So that's the means that we can use to return that money to shareholders. And again, we used already $700,000,000 of the previous plan when before it expired. We will continue paying down debt. It's not a means of we are at the right level of leverage, but more there's we want to keep optionality. And the last thing we want is to have cash on our balance sheet sitting idly.
So we'll be finding ways to make good use of it by any of these other 3 and to some extent the debt repurchase the debt repayment. Obviously, we don't know what we don't know about the future. We do believe strongly about our growth plans, but here is a chart that shows our different metrics under different revenue sensitivities, different revenue levels. So you can see that even in the conditions of extremely low 2% growth, we still saw some pretty robust growth in all of our metrics. And obviously, if we go the other side, if our revenue growth is 8%, you can see some extremely high numbers on all metrics.
So it was just to give you a perspective of what happens if revenue is different than the 5% CAGR. So now in summary, basically, you saw today presentations from all of our groups and from Keith. And our outlook and our plans are very robust. We feel that we have a very strong revenue growth coupled with really good plans to improve our margins and accelerating our free cash flow. So it is an exciting time for us and we're ready to continue making progress to that.
We will have a sharp focus on capital deployment. Goal is again to maximize returns and shareholder value. We think we have the specific drivers in place to do that and the specific plans to do that. A lot of the margin improvement is independent of revenue. So you don't have to depend on just the macro and the external world to make it happen.
And as I mentioned at the beginning, this is a journey. We have a company that is in significant transformation. We have always seen a significant amount of transformation and we expect to continue in our new 2022 target is a milestone and not the end game. And with that, it's all the material I've covered. We'll open it up now for questions to Bill and I or to anybody else.
Keith? Yes. So it
is for the whole of the ON executive team. So all the presenters, if you could please step on the podium.
We need a mic.
Mark Lipacis from Jefferies. Thanks a lot for the great presentations. It really is impressive. A question for Keith. Having come to these analyst days of yours for the last 10 or 12 years, it seems to me that one of the things that strikes me the most is the today, you have so much more conviction in your visibility than you did 10 years ago.
And I guess what I was hoping that you could share with us, Keith, a perspective here like versus 10 years ago, what percentage of your the sockets that you were winning 10 years ago, you had to fight for again every year or every other year versus the sockets that you win that you have to only fight for every 5 years or 9 years. And over that 10 year time period, maybe on top of that, if you could share with us your view on what are the biggest changes at ON that has occurred over the last 10 years? Thank you. Sure.
Yes, I think there's a couple of perspectives. 1, if you do go back 10 years, you'll find out that computing and consumer were the bulk of the business, nearly half the business. Those are turnover every year. And also they are classic classically known for having 2 or 3 sources for every component that they've got. That's the way they drove their economics.
And then today, you can see the profile very, very different. It's mostly industrial, automotive and the data center or 5 gs related things. Big decisions that need to be made. The product sets that we have within those are providing critical value from an energy perspective and or performance perspective. And the customer base, of course, a lot more stable.
So it's a the whole market focus piece is huge. The amount of what I'll call multi sourced products has gone down very dramatically. Most of the things that we sell today will have very few second sources, if any. And most of the things again that we sell today have a performance advantage over things that may have the same pin outs. So that's pretty dramatic transformation and does give us a lot more visibility in the future than we had 10 years ago.
Thanks, guys. So I think one of the key takeaways for me at least from today is that you guys are investing to integrate more vertically in new materials and stuff like silicon carbide and epi. These can be kind of difficult things. So I guess for Bernard, talk about the R and D investment, why you think you can kind of keep that under control because these are some pretty difficult tasks? And then on the CapEx side, 8%, there is quite a bit of capacity coming on.
What gives you confidence that you can keep that at 8%? And are there any opportunities after this kind of large initial ramp to bring that back down?
Very good. So on the OpEx, we always will have leverage on the G and A portion. As we have a 5% CAGR, I expect that our G and A will be continually decreasing as a percent of revenue, which allows us to spend a little bit more on the R and D front for on the R and D and the S and M front to really cover those high growth end markets and net net still be at about 21%. On the equipment front, we believe that 8% is the right level to at this stage to serve these high growth end markets that are also more capital intensive. Building modules is more difficult than building a SOD23.
So you need a little bit more capital intensity. We will continue with our mode of looking at used equipment for the front end mostly and that will help us keep the intensity of our CapEx to these manageable levels. And obviously, we'll look at what how the market performs to decide whether these numbers need to be adjusted. But at this moment, we believe that 8% is the right level.
Thanks. Chris Caso from Raymond James. A 2 part question, one on capacity and utilization for both ON as well as the industry. And obviously, we've come off a period of exceptionally tight utilization levels. Given the CapEx that you're spending and some of your competitors, what do you think that does to the balance going forward?
Do we stay tight? Do we normalize? What's your view? And then following on that, in terms of ASPs, as your mix increasingly shifts towards industrial automotive away from handsets and some of the computing business. Does that change the typical ASP reductions that you've seen?
And you obviously had a tailwind on that last couple of years. Is that something we should expect to be more permanent as that mix changes?
I'll handle the capacity piece of it. The industry is still short on power products. Lead times are still extended. Yes, there's expansion going on. No question about that.
That expansion comes on very slowly because the equipment industry is also behind. And so we do expect some improvement on time, but frankly based on the numbers you saw with the guys today, whether it's electric vehicles or 5 gs or the cloud, we think that demand there continues to stay ahead of the amount of capacity it can be added to the industry. So we feel pretty good in the power area. The balance of the business is going to continue to have cycles, particularly things like handsets. Each season, they go up pretty dramatically.
So there'll be swing there. But the bulk of where we're headed, we think the 8% number kind of keeps us up and we don't see anything in the industry that scares us relative to oversupply. On the ASPs, we mentioned earlier, we're continuing to model traditional down. We've had a great year last year and entering this year, things are still quite tame even with all the noise and reduction in volumes. So we're hopeful it can be better, but we didn't model that.
Thanks. Vijay, it's Mizuho. Bernard, a question on the tax rate. You said it goes up to 17.5%. When do you see that happening?
And also in terms of the buybacks, is there a target percent of free cash flow you want to put to buybacks there? And last question on the PSG side, I know you mentioned you're taking with the on the cloud power, VR-thirteen, VR-fourteen, you have some new controllers coming in. Where are you taking share from? Thanks.
Okay. So what was the first question? 2017 tax. Oh, the 2017 tax. We pretty much believe that the tax the NOLs will be used up by 2020.
So in 2021, we'll see we'll start seeing some of that increases and full phased into 2022. The second question, buybacks, we do have a $1,500,000,000 buyback program out there for 4 years. So that should be a good way to think about it. We have historically been opportunistic in the way we do this. So I expect that it will continue doing that as we always go through time.
And obviously, we'll be gauging that with other opportunities for the use of cash that might modulate the intensity that we have. But
and the other one? So on share gain on VR13 and 13. Hc and 2014, there's 2 things. First off, there's a lot of content increase. So there's a lot of growth there for the top 3 to 4 players anyway.
But traditionally, Infineon has been one of the largest suppliers in that area as well as TI. So we're definitely competing against those 2 suppliers.
Craig Ellis, B. Riley FBR. And thanks to everybody for all the detailed information on the business. I had 2 related questions. The first is to Bill.
Bill, you talked about the low cost position of the back end part of the factory network. Can you quantify the reduction in per unit cost or other cost parameters that you've seen recently? And what should we expect over the target model forecast period, not just for the back end part of the fab network, but the overall fab network? And then the second question would be for Bernard. Bernard, it looks like about 65% of the gross margin improvement in the target model is related to things that are volume dependent, either volume or mix.
Does that mean that we can expect if growth is fairly steady, a fairly linear progression to the new target financial model or are there some things that would create discontinuities in the forecast period? Thanks.
So from a cost standpoint, we target our factories to reduce on the order of 6% cost reductions per year. That's the challenge that we give them. That typically is going to at least meet or exceed traditional ASP reductions that we've seen. In addition, we have other structural things moving to more efficient facilities within the network that we'll continue to work on over the next few years, including in some cases, expanding from 6 inches to 8 inches in some of our factories.
So on the question about the linearity of the margin gross margin improvements, there will be a good element that will be linear, like you said, revenue based or mix based. Some of the rest, particularly in the manufacturing area is going to be more lumpy. And some of it depends on customer qualifications, which might push it into being a little bit further back end loaded than front loaded.
So with this, we conclude the Q and A session and also the formal part of the presentations. So we can disconnect the webcast at this point of time. And we have launched in the ORchid norm. So we will be very grateful if you could join us there. And then after lunch, we have golf.
Thank you everyone for coming. And we look forward to seeing you in future events.