Good morning, ladies and gentlemen. Stuart Chambers, chairman of Arm. It's my pleasure to welcome you to our 2015 results presentation. Let me just first kick off with a couple of board changes, which we've made since I saw you last, which was at the half year. Chris Kennedy, obviously, fully on board now. Also, Lawton Fitt has joined the board as a non-executive director, but also as a replacement to Kathleen O'Donovan as chairman of our audit committee. We've also had Stephen Pusey join us, who will bring a welcome global technology approach to our board. Those are the board changes. Let me now turn to the company. Briefly, last year we celebrated our 25th anniversary, Simon and the team had a whole series of events around our various locations, clearly thanking, if nothing else, our employees for the tremendous progress we've made.
In that 25 years, a lot, many very successful years of progress. Clearly, it is our intent to look back at our 50th anniversary at a second quarter century of similar success. To do that, of course, we have to do two things. We have to deliver today, tomorrow, next week, next month for our customers on the execution of our various technology roadmaps, we need to keep on doing that flawlessly. Secondly, it's very important, of course, we have a long-term perspective on strategy, we have long-term strategic objectives. As a part of that, at the investor day last year, we talked about accelerated investments in order to make more progress further down the road. Of course, those are very important things for us to be doing.
Without further ado, now, I'll hand over to Simon, who will cover both of those things, our performance in terms of 2015, but also importantly, progress against some of our medium and longer-term objectives. Simon.
Thanks, Stuart. Good morning, everyone. Thank you for joining us. Before I start, let me just refer you to the usual cautionary language, if you need any assistance with that, please see us later on. Here today to talk about our full year results, our Q4. Looking back on 2015, it was a very busy year for Arm and the Arm partnership. In that, I'm really thrilled with the progress that we've made on delivering to our long-term strategic objectives, which are around collaborating with business leaders to develop the technology that's going to be required for the products of the future. I think we made great progress on that in 2015. When we boil that down to the numbers, we'll talk more about that later, Q4 was very strong for us and ended a strong year of great performance.
We ended the year with nearly $1.5 billion of revenue. In Q4, our licensees shipped 4 billion chips containing Arm technology. The total volume of shipments through the year was roughly 15 billion chips. That's a huge number. Q4 was the first time that we've broken that 4 billion in a quarter barrier. Very large volume. That's resulted in strong royalty revenue growth. The volume growth in the year was about 23%, and the royalty revenue growth was 31%. The reason for that difference is the increased Arm content in some of those chips, and the increased royalty rate that we're able to command because of the value that our technology adds into those devices. The business performance led to strong cash generation that's enabling us to recommend a 25% increase in our dividend because of the profitability and performance of the business.
Those results happened because of the long-term investment and the long-term view that we take in growing the business and the investments that we've made for many, many years, resulting in products that are licensed in the year, and volumes of chips that are shipped in the year. This all comes about through the work that we do in our R&D. At the Capital Markets Day, we talked about some of those investments that we're going to make in the future. Arm really is a long-term business, and we're thinking today about products that will be shipped in five, 10, even 15, maybe 20 years' time. We have product shipping today by our licensees based on technology we were designing 20 years ago. This really is a long-term business, and we have to take a long-term view on investment.
I'll talk briefly about some of these areas now. Starting with mobile. Clearly, the growth of smartphones is slowing down. This is something we've been anticipating for a long time. It's obvious that this is going to happen. As we look to the future and how the mobile computing market plays out, we see continued opportunity for more compute performance, more Arm content in the mobile computing devices of the future. That's going to come about again through work we do within Arm. Crucially, it comes about through collaboration with our partners, the way we work with leading companies in this space to take what we do, to amplify what we do with all their own intellectual property and their own work.
That collaboration, which is a crucial part of Arm, continues to play a role in the evolution of our products in all spaces, but especially in mobile. In networking, in the enterprise segment, in networking and in servers, we've outlined before the opportunities for Arm and the Arm partnership to grow share, and we've seen good successes in that in 2015. What we're doing to accelerate that is further investment in the software ecosystem. There's a lot of software that's got to run on these processors in the data center, in the network. Our investment right now is about enabling that to be optimized for the Arm architecture to fuel that growth.
We're doing some of that work internally, we're partnering with leaders in this space to make sure that the right code is optimized in the right way for the right workloads, because our partners understand the details of that much better than we do. Partnering, again, very crucial for our growth in this space. Turning to embedded. All around us, we're seeing more and more devices become connected, more and more devices become increasingly intelligent, so that they can make sense of the world around us and share data with other devices to gain real insight. All of that is enabled by tiny, low-cost microprocessors embedded in microcontrollers, which sell for a very small amount of money. That low cost leads to very high volumes.
This trend of embedded computing in all sorts of applications is going to be a real driver of technology over the medium to long term. What we've seen is a migration in this space from very simple technology, 8 and 16-bit architectures produced by others, to more sophisticated 32-bit microprocessors. This is where Arm and the Arm partners have really taken a leap. We are continuing to enhance that lead through R&D in our products, again, partnering with others to bring the right technologies together to solve some of the security issues that arise when suddenly everything is connected. Three areas in which we're investing in both hardware and in software to grow the opportunity for Arm. As we do this, we look at the size of the markets, these are big, growing markets.
When you take just these three, obviously we're involved in other markets as well, we'll come onto some later. In terms of the silicon content, the devices that will be sold by our licensees, these three markets alone add up to $120 billion of silicon. That's $120 billion in 2020 growing from today. You can see, you can do the math there. The increase is about another $30 billion worth of silicon compared to 2015. These are big markets, they're growing markets, they're all markets where the computing requirements are growing. These are perfect markets for Arm and our partners to address. Back to mobile for a moment. The devices that you're all carrying today, I'm guessing that everybody in the room has a smartphone, at least one.
The devices that you see your friends, your families buying are all based on technology that we developed a long time ago. Through 2015, we've seen increased adoption of version 8 of the Arm architecture. Come the end of the year, when we look back over 2015 as a whole, around half the smartphones that were shipped in that time were based on version 8 of the architecture. That is technology that we started developing a decade ago. It's been in development a long time. We've produced processor designs based around V8. Our licensees produce chips based around this, now they're shipping in volume. You can see through the year that the volume has grown. In fact, in 2015, compared to what I said 12 months ago, the adoption of V8 in mobile was stronger than we expected. It's not just about V8.
We've seen increased attach rate of our Mali graphics processor. Very high volumes there, about 750 million Mali processors shipped by our partners in 2015. The attach rate of Mali going up. We're the number one graphics architecture. Increase in the number of cores, the amount of compute power packed into these mobile devices that we're carrying around. 2015 has really seen an increase in the compute performance of these devices. We've seen the growth of eight-core devices and even some 10-core devices now in smartphones, but all based on technology that we already developed. What's coming next? In Q4, our licensing into the mobile sector was very strong. I'm just going to grab a glass of water here, excuse me. Very strong demand for our next generation products.
There were nine CPUs licensed by our partners, which are products we haven't actually announced yet. These go by code names. They're cats and dogs and birds and flowers and things like that. Products we haven't actually launched yet. Very strong demand for our next generation technology to provide the computing for next generation devices that hopefully we'll all be using. Our continued investment in this sector is leading to that technology and leading to strong licensing in Q4. We've been evolving our technology for the high end. Everybody thinks about the flagship devices, these kind of hero products which have all this compute power in them. There's a lot of growth ahead in low-cost devices, in entry-level and mid-range smartphones. These are going to be the fast-growing sectors or segments within the smartphone sector.
We've been evolving our products through R&D investment to make sure that we are well suited to address that portion of the market as well. Through 2015, we launched Cortex-A35, a version 8 architecture processor optimized for the needs of a sub-$50 smartphone. That is a portion of the smartphone market that we think is going to grow very rapidly. We want to have the right technology for it. We're not just focused on the very high end. We're focused on this mid-range, these entry-level devices. We're focused on wearable devices, too, and the introduction of our Mali-470 with a GPU optimized for these small screen devices, smartwatches, IoT devices, small things with smaller screens. Again, this all comes about through our investment in R&D.
We're able to target our technology at different markets because we have the engineering manpower, the firepower, to create the optimized products for these different segments of the market. One of the other things that we've done is evolve the way that we're working with some of our licensees. As I mentioned, collaboration, partnership, these are the cornerstones of Arm. When you read through the release, you can see the announcement of a program called Built on Arm Cortex Technology. What we've seen over the years is some of our licensees have some different needs for what they want to do. Some of them have addressed that by taking an architecture license and making a huge investment in building a processor team to build their own CPU core.
The way we've evolved this slightly is working with some of our partners who are looking at just a little bit of difference compared to our standard offering. We're going to be working with them closely on new products as they're developed to create a slight variant for what they need. We're doing that engineering work in collaboration with them. We're making sure that the software ecosystem supports that variant and satisfying the needs of our customers who want specialization, differentiation. They want to tailor their products in slightly different ways. As the end market gets more sophisticated, we believe this is a great way of addressing those challenges. Qualcomm is the first partner that we've announced today that we're working with in this way. I'm sure there are going to be others in the future.
It's just a slightly new way of working with our partners, just for making sure that what we're doing is optimized for the needs of our customers. Now moving on to networking. We've spoken before about the changing needs of networking and the opportunity that that brings to Arm and the Arm partnership. A year ago, I was very proudly saying that our share of the enterprise networking space had grown to 10%, through 2015, that went up by 50%, and we now have a 15% share of this market. Very pleased with the progress in this space. In Q4, we saw a number of version 8 architecture chips announced by our partners. There were devices from Broadcom, from Marvell, from NXP.
Leaders in the networking space who are producing very sophisticated, very high performance, in many cases, very high core count devices to meet the computing needs of the next generation networking equipment. This is targeting a wide range of applications. There are low core count devices, six core count devices. The scalability of our architecture means that the Arm partnership can address the entire space represented by networking. Now we've been investing in the ecosystem too. Processors are useless without the software that runs on them. In networking, there are different requirements. Networks are becoming more software oriented. They're requiring more flexibility. They're requiring the capability to have different applications running on the same processor, but isolated from other applications. This is what virtualization does. It generates a container for an application to run in.
We've added features into our processes over the years to support that, we're now working with the software ecosystem, again in a partnership model, to create the software that the next generation networks will require. A key component of that is what's called OPNFV. I won't bore you with the technical details of that. If you think of that as a software platform to enable the functions that are going to be required by the next generation of networks, that is a very key component, we're optimizing that to run on the Arm architecture. Now at the Capital Markets Day, we set out a goal of achieving a 45% market share in the networking space by 2020.
The way in which we'll communicate our progress towards that is through what Ian called the SMARTIE chart, which is breaking the market down into different segments. Each of the SMARTIEs on the chart here represents a leading company in this space. If we have high success in all of those, we end up with an 80% market share. This is a way of segmenting the market and showing our progress within that. The color codes mean, red, we have nothing going on there yet. Amber means there's a design win. Green means somebody started shipping, and although there's no blue on there, when you see the dots turn blue, it means the majority of their product line is based on Arm. We will do updates on this as we go. For this last quarter, we've seen products shipping in the wireless access space.
One of the leading companies has turned green there. We have some design wins in two other areas, which are important in this market. I think in one quarter, that's pretty good progress, and we will give further updates as we get more design wins, as products start shipping, and hopefully, as more of these companies move more of their products towards Arm. The data center is a very important market for Arm. Clearly, a lot of computing is going on in the cloud, and we believe that there is a great opportunity for Arm and the Arm partners to target this space. The Arm technology is going to allow greater flexibility in the kind of processing that's done in the cloud, with the added benefit of greater energy efficiency.
This is an important market, and the Arm partnership is driving a lot of innovation in this space. Progress in 2015 was pretty good. We saw more chips designed by our licensees to target this space. We saw the shipments of Arm servers, and we saw a lot of work going on around the software ecosystem that's required. Just recently in January, AMD is another example, gave an update about their product roadmap, introducing more products based around the Arm architecture targeting this space. Again, it's all about software. Processors need software to run on them. For some time now, we've had a big investment in this space. Again, work we do, work we do through the ecosystem, work we do with our partners. Now all the major Linux distributions that are important for this market are optimized to run on Arm.
Not just about the operating system, you have to think about the workload, what the server is actually doing, there is lots of different software that runs in a data center. We're picking off the key workload packages. We're optimizing them so that somebody building a data center has choice about the technologies that they put in there. All of this work has led to real deployments. There are now three tier 1 companies operationally deploying Arm-based servers. This isn't kind of an experiment in the back room. This is real production, operational deployments of Arm technology. There are three tier 1s in different countries, and in fact different continents. There's one in China, there's one in North America, there's one here in Europe, using Arm technology to provide better flexibility around their service to their customers. We're really pleased with that progress.
I want to talk about embedded. Embedded is a very broad space with lots of different products in these, in most cases, very low-cost microcontrollers. I am very pleased with our progress there because Arm is the number one architecture for embedded. There has been a migration over the years from these simple 8 and 16-bit microcontrollers to more sophisticated 32-bit microcontrollers, and the Arm architecture in that 32-bit space, which is the fastest growing, does have number one market share. We are number one in 32-bit. When we look at the dollar value of that market in total, again, we have the number one market share. Our shipments into this space have gone up 25% year-over-year, and in that time, shipments of 8 and 16-bit based microcontrollers have gone down 5%. Again, we have been gaining share in this space.
What we have seen in 2015 is more sophisticated microcontrollers. Microcontrollers based around Cortex-M4. That is a processor that adds DSP capabilities into the microcontroller space. Products based on Cortex-M7. This is a superscalar processor. If you asked people 10 years ago, would you see a superscalar processor in a microcontroller? They would probably say you are mad. Our technology, given how efficient it is, how small it is in terms of die size on silicon, has enabled that to happen. That is delivering a lot of performance into a field which has typically relied on just the bare minimum you can possibly spend the transistors on.
This architectural shift, the work we are doing in our products, the work we are doing in our ecosystem, is enabling far greater performance into a category of electronics which, in the past, has been very constrained by the performance that you had available to you. We are continuing to add capabilities. We are thinking about how these devices become connected. During 2015, we launched the Cordio Radio. We saw our second license of that in Q4. We are bringing some of the security features that were previously only available in our application processes down into microcontrollers with the Armv8-M variant of the Arm architecture. Again, we are licensing that to people. A lot of progress in this space, and it is growing rapidly. Why, though, do you want a 32-bit processor? 8 and 16-bit micros ship by the billions.
They have been around forever and will probably ship for a very long time to come. Why do you need this shift to 32? As I said, 8-bit technology, great. It has been around for a long time. Volumes are very large. What they do, though, is they enable very simple devices. If you want to put a screen on something and have a little LCD with some numbers on it, an 8 and 16-bit micro is not a bad way of solving that problem. It is a very simple device. Hard to program. If you ever want to use the code for something else, if you want to find engineers who can program it, that is pretty hard. These devices tend to be written once, black box, and never touched again. The functionality that you can provide is relatively simple.
You see 8 and 16-bit micros in streetlights where maybe you've got a very simple sensor. It's just working out when it's dark and it turns the light on. In something like a blood glucose monitoring device, again, you might have a screen, stab yourself, and it says the answer is 5.7. I don't know if that's good or bad, but that is the answer. Do something. In cars, lots of microcontrollers in cars, helping the functions of unconnected devices. The speedo and the rev counter and the display, all separate devices which just need a little bit of control. Maintaining the code base of these devices and what you can subsequently do is very limited. We saw this shift to 32-bit coming.
We anticipated that you could build a 32-bit microcontroller in the future, that future is now, for the price of an 8 and 16-bit micro. Once you've got a 32-bit processor, there is just so much more you can do. You've got more performance. You can use modern programming languages. You can hire engineers who know how to code for it. You can run an operating system. You've got a lot more performance and capability available to you. As a result, these devices are becoming much more sophisticated. You can have a streetlight which is connected. Isn't just switching itself on and off, but if the bulb breaks, it dials up some help and somebody comes and fixes it. It can think about the environment. It can think about how it interacts with other streetlights and other infrastructure in a city.
The glucose monitoring system can now pair with a smartphone. The data that you capture can now become useful because it can get broadcast somewhere in the cloud, some analysis done. Your blood glucose was 5.7 on a Monday morning at 9:00. How does that compare with every Monday morning at 9:00 in the past? Data can now become more useful because you've got more processing power locally to do something. In the car, we're seeing a real revolution about of the technology in a car. You went around CES a few weeks ago, lots of floor space given over to automotive, and people are thinking really creatively about how to use embedded computing, given you've got more performance now, in creative ways.
Talking of cars, this for us is a really interesting market, it's one where the semiconductor content we anticipate is going to grow quite significantly. This, in 2020, in terms of a portion of the semiconductors in a car that Arm could address, grows to a $15 billion market. This is $15 billion worth of silicon, potentially all of which could contain Arm technology. That's a materially large market for us, and it's one that's growing. When you think about the number of cars that are going to be manufactured in 2020, and you divide that into 15 billion, you get an average silicon content per car of $150. That may not sound very much, but you compare that to smartphones. That's about seven or eight times the amount of semiconductor content relative to a smartphone.
The volumes might be lower, but this, and certainly in the higher-end cars, is going to be a place where there's a lot of computing power, there are a lot of semiconductor devices, and in all areas. Whether it's sensing whether the door's closed properly, sensing whether you're about to hit the car next to you as you open the door and doing that for you so it never happens, whether it's just unlocking the doors or with the plethora of cameras that you're going to see in cars, making sense of what is going on around you. There is a growing need for intelligence in a car. You can't defer all of this to the cloud because you need real-time response, and the latency through the network is not going to let you do that.
You're going to need a lot of processing in a car, as we're saying here on the slide, we anticipate a car really turning itself into a supercomputer on wheels. Lots of compute power in there. Lots of our licensees are looking at this space and looking at how to address that. A great example was shown by NVIDIA at CES. They launched their product called Drive PX2, this box here on the lower right of the slide. Big thing, it's water-cooled. It's got a lot of compute power in there. It's got a couple of chips that they have developed with a lot of microprocessing power. Within each of those chips are eight Cortex-A57s and four of NVIDIA's own implementations of the Arm architecture, they call this Denver. Four Denvers, eight Cortex-A57 times two.
That is a lot of computing power put into a car, but it's what you're going to need if you want to ultimately have a car that can drive itself, work out what's going on, work out is a pedestrian about to cross over, are they going to step out? What's this guy in the next lane about to do? What's the lane doing? That is a computationally really hard problem. Now, I've sat in a self-driving car, and you can just see how much more compute power is going to be required to really do that seamlessly and safely every day of the week in any certain circumstance that the car might come across. Semiconductors in the car going up. Again, it's a big growth market, and it's a great opportunity for Arm to address this with our innovative technology and our partnerships.
What this is going to result in is we think about 100X increase in the compute power in the car over the next five years. If that sounds like a lot, compare that to what's happened in smartphones. We look back over the last five years of smartphone evolution, and we see a similar growth in compute power. We've shown you some of the data before, but it's 80 to 100X over the last five years. Growth in compute power of something you carry around with you and run off a battery. Again, 20 years, that would have seemed infeasible 20 years ago. The smartphone did that, and we think that's going to happen in cars, too. Working with our partners, we're leading the way on that. We're working with people looking at in-vehicle infotainment into ADAS systems.
This is about providing the driver with more information, again anticipating what's happening, c hassis drive systems, t here's just semiconductors everywhere in cars. We're collaborating with OEMs, too. There's a lot of investment going in across the industry to address the challenges of next generation cars. As a company that thrives on partnership, this is a great place for us to be. We're going to work with these companies. People want to work with us to solve these next generation challenges. We're evolving our products, and we're evolving the way we design our products to make them suitable for safety-critical systems. Obviously, a car is a safety-critical system. It requires innovation in technology, innovation in our partnerships, innovation about the way we work. We're very focused on that since we see automotive as a big growth sector for the future.
As I think about the way the use of embedded technology is going to grow over the coming years, I can see lots of places where today's products work really well. Tomorrow's challenge is different. Tomorrow's challenge requires more compute power, more innovation in devices, more creative ways of bringing vision, cameras, computing, connectivity into the cloud together to provide the kind of compute power that we're going to need in our next generation devices, whether they're in our pocket, in our driveway, or just the infrastructure of the environment. The connectivity between a car and a traffic light, just the traffic light telling the car to stop instead of having to work out that the light's gone red. That's all about connectivity. It's all about embedded computing.
We want to make sure that we've got the right products for all of those technology spaces. That's why we're investing so much in our R&D capability to address that future need. We outlined that back in September. We're executing on that. The licensing that's being driven today comes from products that we've already built. That is the history. We're thinking about the future. We're only going to do that through our own work. Crucially, collaboration with our partners, building our ecosystem, working on the things that are going to help accelerate the gains. You saw us talk about investment in software, investment in the ecosystem. That's critical for us to address that market. We're going to keep focused on that. Finally, that translates into money. It translates into growing revenues. It translates into growing profit.
As we provide more value, we get paid for that. You've seen that effect in 2015 in the strong growth in our royalties. With that, let me hand over to Chris. He's going to go through some of the numbers. Then we'll be back a little later for some Q&A. Thank you.
Simon. Good morning, everybody. I'm going to start by taking you through the quarterly numbers. It was a strong quarter, as Simon said. Dollar revenues were up 14%, and that was driven by strong royalty growth from the adoption of all of that advanced technology in smartphones that Simon talked about, as well as the share gains we're making in markets beyond mobile. That was coupled with flat licensing off the back of a very strong 2014, and the royalties were also helped by a royalty catch-up from one of our partners who had underreported in prior years. We had a royalty catch-up of around $9 million. That 14% in dollar terms was helped by the strong dollar. Sterling revenues were up 19%, and operating costs were GBP 124 million.
That year-on-year increase was driven by an increase in R&D, which was 37% up in the quarter, year-on-year. As a result of all of that, normalized profits were up 17%, normalized DPS up 14%. I've lost this mouse somewhere. The full-year picture reflects the same trends as the quarter. Dollar revenue is up 15%, 22% in sterling terms. At the same time for the year, our normalized OpEx increased by 20%, and that's due to the investments we're making to maintain our pace of innovation and accelerate the market share gains in our target market. As a reminder, R&D investment accounts for half of our OpEx, and that grew 28% in the year. That means that the other costs, the other 50%, increased by significantly less, around 13%. They're growing at a lot less in sterling terms than our revenues are.
You can see from the chart, if you look at the headcount we've added in the year, we've added twice as many engineers. Our engineers are growing twice as fast as our non-engineering headcount. That OpEx, you need to think about it as running costs for the business, investment for the future, roughly half and half. Again, as a result of the strong revenue and performance, the normalized DPS grew by 25% to GBP 0.302. We've continued to benefit from the Patent Box regime that the U.K. tax authorities introduced. That's resulted in a normalized effective tax rate of 16.2%. All of that reflecting the high R&D investment we make. We've had a strong 2015. We've outperformed the overall semis market, and that is not a new thing. This is something that Arm has been doing over many years, and this chart just indicates that.
On the left-hand side, the light blue line is the total number of chips shipped every year. It excludes memory and analog. The green line is the number of those chips that have a processor in them, and the dark blue line is the number of chips that have an Arm processor in them. You've got the same data on the right-hand side of the chart expressed as a proportion of the total number of chips shipped. You can see that, as Simon said, that the amount of processing power is increasing every year. The number of chips with a processor in has gone from 25% of the overall market to 68% in 2015. The number of chips with an Arm-based processor in that time has gone from 5% to 21%. What does that mean in terms of the compound growth?
The overall market has grown at 5%, the overall market with a processor in has grown at double that rate. When you look at the number of chips with an Arm-based processor in, that's double that again. We've been growing at four times the overall semis market rate for a long time, and we don't see any reason for that trend not to continue. What does this trend mean for the addressable markets in 2020? These charts are something you'll be familiar with from roadshow slides and from the website. They show the addressable market for Arm-based chips, and we periodically update these target markets, both for the growth rates in the markets that we're in already, but also for the increasing range of markets that Arm technology has become suitable for.
As a result of the most recent review, we've upped our target market by around 10% in both volume and value terms. That's principally because we started to receive royalty checks from processors in battery controllers, flash memory controllers, and smart sensors. Those weren't markets that we had previously put in our addressable market targets. We're getting royalties on them. We think it's appropriate to include them now. We can get into the detail of these market forecasts. You can talk to Phil or to Ian or to myself, and we can go through it on the roadshow. Although we've increased the addressable market for 2020, there's clearly uncertainty over the short-term trajectory of smartphone handset volumes.
However, following our 2015 strong performance through the adoption of the advanced technologies, V8, Mali, octa-core, and the fact that we see further opportunities for our partners to adopt more of those technologies, we are very well placed to outperform the overall market. Looking further ahead, third party estimates for the smartphone market for 2020 have come down by around 5%. If you take those estimates, which equates to around a 6% compound growth from 2016 to 2020, factor in that increased adoption of Arm advanced technology, we're still confident that we can grow the royalties from the chips and smartphones by 15% compound from 2016 to 2020. At the same time, we're continuing to gain share in markets outside of the mobile market.
When you look at the royalty rate in all of our addressable markets, we still believe we can grow at 15 percentage points more than the overall growth in the semis market in that time period. Finally, we continue to expect licensing revenue to grow over the medium term by around 5%-10% per annum as well. As we set out in our Capital Markets Day in September, we're increasing our investment in R&D in 2016 to accelerate the opportunity to gain share in our target markets. We've also been investing inorganically as well. We've made 10 acquisitions in the last 10 quarters. That gives us the opportunity to further extend our growth.
Turning to the balance sheet, we're committed to having a net cash balance over the medium term, this reflects our commitment to maintain the investment that's necessary for our roadmap and our partners' roadmaps. It also ensures that we retain the flexibility to act quickly and decisively in what is a very fast-moving market, where we see opportunities to further extend growth. Given the expected rate of cash generation and the pipeline of opportunities that we can see today, I wouldn't expect us to resort to external financing markets for any acquisitions in the near future. The business remains really strongly cash generative. We generated GBP 361 million in cash in 2015. As Simon said, we're proposing an increase in the ordinary dividend of 25%, which is in line both with our EPS growth this year and the historic trends of dividend growth over the last five years.
The board remains committed to growing that ordinary dividend in line with the growth of the business and maintaining the policy of maintaining share count flat for maintaining that buyback program. I think the combination of the investment we're making in 2016 and the ability to increase that dividend really demonstrates Arm's ability to balance both the investment for the future and increasing cash returns for shareholders today. I'm going to conclude on the outlook. In revenue terms, we're well placed to outperform the overall semis market, driven by that further version 8 penetration in mobile and market share gains elsewhere. Based on our current view of the semis market, we estimate that full year revenues will be in line with market expectations. Although this does assume that a macroeconomic environment doesn't further impact the end market for our partners' products.
Finally, we expect normalized OpEx for Q1 2016 of between GBP 127 million to GBP 129 million as we continue the investment program that we outlined in September. On that note, I'll hand back to Simon to chair the Q&A.
Thanks, Chris. Just before we start, I would point out your expertise in asking compound questions. If we could start with a question, then we'll get around the room hopefully for another spin. Start down the front. Yeah.
Thanks. Kai from Merrill Lynch. I had a question on the near term. I appreciate you've highlighted that you continue to expect to outgrow the smartphone industry to obviously take share in other markets. Particularly about this year, it seems like octa-core penetration only 10%, there's still quite a lot of tailwinds, I think. I'm just wondering, nobody knows what the semiconductor industry is going to grow at this year. Let's say it's flat. What relative outperformance of your total royalties would you expect at this point in time? Thank you.
Our guidance is always to outperform the industry by about 15 percentage points, there's some fluctuation in that. That seems to have been the case for many years now, we expect that to remain the case into the future. As you say, we're in an interesting period right now where calling what the industry is going to do this year is very hard.
The overall sentiment out there around smartphone growth is lower than people maybe would have expected a year ago. As you say, there's a reasonable tailwind of more technology going into those devices for Arm based on what we have today, the long-term trends for more and more compute power in devices. As a working model, that 15 percentage point outperformance is what we have in mind, plus or minus through a given period. We don't tend to get too head up about the quarter, the next quarter, the next six months. It really is about the long-term trends.
Thank you.
Thank you. Andrew Gardiner from Barclays. Another question for you, Chris. Just on the dividend and the return policy relative to the investment in the business that you're talking about. Clearly, 25% growth in the dividend is nothing to be sneezed at, but your cash pile does continue to grow. You've talked about it, your predecessor's talked about it, that Arm doesn't need to continue to have an ever-increasing pile of cash on the books, given the business model. That said, you've increased investment. You're talking about a pipeline of potential M&A. You did four last year at sort of GBP 60 million-GBP 70 million worth. Are you suggesting that there's an increased appetite for inorganic growth here? What kind of sizing should we be looking at as to why you're keeping your powder so dry at the moment? Thank you.
As we said at the Capital Markets Day , we've been through our five-year planning process. We took a good hard look at the capital structure of the company. We compared that to our semi peers as well, we took all of that thinking to the board earlier this calendar year. I'll conclude that the first thing to say is we are not out of line with our semiconductor peers in terms of the amount of cash that's on the balance sheet. The second thing I would reiterate is our uses of cash are, it's first to fund growth, either organically or inorganically. Second, to maintain that strong balance sheet, and then increasing cash returns for shareholders as well.
I think the words in the statement sum it up really is we've done a good review in terms of the pipeline of potential opportunities, and they are potential. We're not signaling anything concrete around acquisitions. We're comfortable with the level of cash we have at the moment. Your point is well made. We don't want to sit on an ever-increasing cash pile. The cash growth this year was relatively modest due to the acquisitions we've made and the share buyback and the dividend we declared. We'll keep it under review. As we said, we'll take another look next year.
Thanks. Nick Jones from Unith. I guess I wanted to ask on the physical built-on Cortex product. That seems to be all the first customer you've announced has previously been an architecture license. Firstly, the first question was, is that for a product for mobile or for non-mobile? The second is the revenue potential of this new type of structure selling to them different or higher or less than what you've previously gained from that type of customer?
In terms of the specifics about what they're going to do with the technology, that's for them to talk about when the time is right. They are doing an analyst day, I believe, tomorrow. In terms of the revenue opportunity for us, these different licensing models at a high level have pretty much an equal royalty opportunity for us. There's no radical shift here.
Thank you.
Just keep going back the sound eventually.
Yeah.
Thanks. It's Achal from Credit Suisse. Simon, one question on the adoption of 64-bit and big.LITTLE. I think you gave some numbers for Q3, and now you've given numbers for Q4. I think it was 215 million 64-bit shipments in the last quarter, and now it's 240 million. At the same time, we've seen the big.LITTLE number go up from 70 to 75. Clearly, we're seeing a much faster adoption still on the V8 side, but big.LITTLE is still much slower. What are the things that are stopping your customers on getting on that big.LITTLE train much faster, like something we saw with V8 earlier last year? What are the things that they're possibly thinking about and
I think a mistake you're making in the way of looking at those numbers is to assume that every chip that an Arm processor goes into looks like a smartphone chip. Smartphones are somewhat unusual in that the workload that they're running at any moment in time varies. You can have applications that require lots and lots of performance. Everything you want to throw at it, you can. There are other applications which are very lightweight. Big.LITTLE is a great way of dealing with vastly different compute requirements in the same application. Some of those V8 chips are going into networking. Networking is about data going through at a constant rate. It's not doing one thing one minute, another thing the next, in the way that networks are designed today. Big.LITTLE doesn't apply in that kind of market. Similarly in servers.
These aren't applications which are switching around and need the sophistication of big.LITTLE because they're running on a battery. It's wrong to just go, V8 was growing at this, big.LITTLE was growing at this, and that's good or bad. V8 is a technology that can span lots and lots of markets. You see we're putting it into microcontrollers. Microcontrollers tend to have a single CPU, maybe a superscalar one with M7, but a single CPU. You're not going to see, Never say never, but that isn't the kind of sweet spot for big.LITTLE. It's about application and which technology you need for which application. What we're trying to do with our technology roadmap is to provide the ingredients. The ingredients that work really well together.
Based on what you're doing, you can choose the right amount of compute power, the right amount of everything else that you need, and build a chip that's optimal for that application. You can't just say that the two should grow at the same rate. It's much more complex than that.
Excellent. That's clear. Thank you.
Let's go across from this way.
Thank you. It's Brett Simpson at Arete. Simon, just a quick question on industry R&D spend. If you look at some of your big licensees, they're making substantial cuts in R&D. Qualcomm, I think they're cutting $1 billion off their OpEx, and they've just shut down their custom Arm CPU team on smartphones. Broadcom's doing similar things. You can look down the list of Freescale, NXP. We haven't really seen R&D cuts on this scale since the early 2000s. I'm trying to reconcile that with your outlook for licensing and why you think that we're going to see a growth year this year. Can you maybe just run through what the factors that get you there?
As you say, there's some consolidation going on, and there is R&D trimming. One plus one is turning into less than two as some of these big companies are being put together. There's a maturing of the semiconductor industry going on. I think that the trends that come from that play very well for our business, because people are going to look at outsourcing more. They're going to really focus on what is the value that their company brings. The dialogue we're having with our big customers is about what else can we do so that they can spend their R&D on the most effective areas. People are looking to stop doing the things that really don't differentiate their end product. As a business to whom the semiconductor industry outsources, that's a good trend for us.
That's one of the things that gives us confidence about the licensing in the near term and in the mid-term. To me, there is no doubt that the different types of processing elements, the way in which processors will be used in the future is just an expansion opportunity. We want to make sure we've got the right products so that we can be the guy that these big companies outsource to with the right product at the right time to address these growing markets.
Just to follow up on that. The industry is reducing R&D. You guys are raising R&D quite significantly. Can you just talk about two or three things as to why you think that's the right move for Arm? What's underpinning that increase in head count?
Well, as I was explaining in the presentation, it is about the opportunity that will mature over the next five to 10 years. Because of the long-term view we need to take on that. We can't throw a CPU design team together to build a processor in six weeks. It's just not like that. V8 has literally been being designed for 10 years. Now we're seeing the volume come through. We have to take that long-term look. It kind of related back to the balance sheet question as well. With the volatility that is around at the moment, with the uncertainty, now is the time to invest. We've seen this in cycles in the past where the companies that come out of these periods in a strong position, and we've done this before, are the ones that invest through them.
We have the capability to do that. Looking at this opportunity ahead, I believe it's the right thing for us to do.
Jaguar Bajwa, also Arete Research. Given the strong year you've had in smartphones, the increase in core counts, the move to V8, I was wondering if you could give a guide as how much of your royalty revenues is actually coming from smartphones this year.
You mean in 2015?
2015, also potentially the year-on-year move that you've seen.
To be honest, I don't keep that number in my head. We look at the split of the volume of Arm-based chips, and there's a table in there. It's about 45% of the volume is mobile. I think, and again, we can maybe follow up with the team here, it's about 60%+ of the dollars come out of mobile. We can follow up to see if we've got some more accuracy on that.
Okay, thanks.
Good morning. Amit Harchandani, Citigroup. My question pertains to the investments that you're making in the business right now. It pertains to share-based payments. It was up around more than GBP 75 million for this year, more than GBP 70 million last year. You spent about GBP 90 million to offset share-based dilution this year, if I remember correctly. Is there a case to be made that these figures really now need to be considered as normalized in terms of viewing the investments in your business? I just wanted your thoughts on the same. Thank you.
Yeah. Clearly we put out the IFRS figures alongside the normalized on the front page. It's completely transparent. One of the reasons that we exclude the share-based payment from the normalized is that it is quite volatile because it depends on the share price at the time as well. We believe the normalized view is a better way to be able to compare numbers over time and over multiple time periods. That's why we present them the way we do. The IFRS numbers are there for those who would like to use those as well.
Thanks. Gareth Jenkins, UBS. I just wanted to come back on the 15% outperformance versus the industry, which has remained unchanged this year through 2020 and despite having a very strong year last year for royalties, V8 going better than you thought, and the smartphone market expectations coming down, what has changed to offset that performance you put in last year and the smartphone market expectations? Is it line of sight on some of these new areas of growth for you? Is it networking and servers you feel more confident on? Was it just underlying prudence originally, and that maybe the compound growth rate should have been higher before?
Well, a combination of most of those factors. As we show here, our share in networking went up from 10% to 15%. That's a trajectory of growth that we're anticipating to continue. The increased content in the devices, as we talked about, the tailwinds of more Cortex, more Mali, more V8, are going to help that revenue growth and help what we believe will transpire into roughly that level of outperformance. Actually, the level of outperformance in Q4 was 17%, so the industry actually went backwards 3% while we grew 14%. There's variance in there. Obviously, most of this is out of our control, but we look at the pipeline of products coming out, the way in which we anticipate they're going to be used and products sold. We feel okay with that as a guideline for par, if you like, for our outperformance.
If you think about the addressable market slide that Simon put out, you've got three very large markets, one of which is mobile, which the estimates have come down by 5% for that market in 2020. You've only got a third of your addressable market that's coming down a small amount. We then recognize also the increased use of the Arm technology in smart sensors and battery controllers and the flash controllers. Within the mix, that addressable market is still very big. Everyone's focused on the mobile market and the immediate slowdown. Two things. One, over the longer term, that slowdown doesn't impact the overall CAGR to 2020 as much as you might imagine. Secondly, we've got the other markets to address.
Hi there. It's Owen from Numis. A question on the economics of the Built on Arm Cortex Technology announcement you put out this morning. Historically, the beauty of Arm was you design it once and you sell it many times, and the model is very scalable and very high margin. With this new model where you're taking more R&D and customizing for your customers, does that impact the margin potential of Arm? If you're taking the cost and you have to have an R&D team for Qualcomm and all your big customers, would that be a lower margin model longer term rather than one standard product for everybody?
Yeah. At one level, it sounds like a services business. "What would you like me to build, sir? Here's my team," and off we go, and we do it completely bespoke. It's not like that. We're talking about fundamentally a standard product which might be a little bit different based on how our licensee wants it integrated with their own technology on the same chip. The scale of changes that we're talking about are pretty small, but they're going to result in a more optimized end solution. The learning from that potentially can get rolled back into our future products. It's not completely, we're going to have a big CPU design team for every customer on the planet. It is about making some incremental changes, which we hope will result in bigger efficiency gains when the SoC is put together by our customer.
It does require some additional engineering to do that. I think we result in a better product, and hence hopefully greater adoption by people who are ultimately going to buy the product at the end of the day.
Just behind you.
Hi, it's David from UBS. I just wanted to come back to the automotive market, which you're talking quite a bit more about these days. When you've gone into other markets in the past, excuse me, networking, microcontrollers with embedded, you've often created software platforms and put a lot of investment on that side as well to establish yourselves in those markets. Is there something we should think of that you're doing along these lines to establish yourself in automotive, especially in things like safety and a lot of the new areas that are coming through?
I think over time, that might be something that we look at, particularly when we consider the challenges around securing all of these devices within a car. The car is becoming a sort of mini network, which in itself needs securing, and then a networked device into the broader wide area network. There are some security challenges around that, and we are thinking about the software content there. Maybe we'll do some of that. Maybe we'll work with the ecosystem. It's a problem that needs solving. Traditionally, a lot of the software that's run in cars has been a closed box. We've been in this market for a long time. When I did design work with a large customer in Germany who was thinking about two Arm processors and did they ever disagree with each other and turn a light on.
They were running their software, I bet you there's cars today with that system still in it. A lot of that software is homegrown and doesn't change very often. In the future, it's going to become more open. It's going to need more what has become the conventional way of developing code. You've got those security and safety issues to deal with. This isn't going to change overnight.
Thanks. It's Vidyanand from Merrill. I had a question on the networking market. Share went up to 15% from 10% last year.
My understanding is historically, most of your networking share is in the enterprise market with relatively limited penetration on the telco side. I wonder if you can explain or help us understand the dynamic in 2015, whether you made any more progress on the telco side and how you expect that to evolve, looking ahead. Thanks.
Yeah. What we showed there with the SMARTIE chart was about the penetration into some of those sub-segments of networking. Did we make progress in 2015? Absolutely. The telco side, the mobile operators are looking at how they deploy more flexible networks, how they deal with greater bandwidth, shorter latency, deal with the challenges of connecting billions of devices versus hundreds of millions of devices into a particular network. That has implications through the whole network from the edge where the device connects all the way back up into the cloud. We're anticipating a world where Today, there are kind of harsh boundaries between the end client device, the network, the cloud. The future, that becomes a much more smeared environment.
We are talking to everybody working through that chain, adopting our conventional partnership model of approaching the market about those challenges and thinking how that relates to our products, who we work with, what code we optimize, that's how that's coming together. We do believe that will result in penetration into sectors of the networking space that haven't traditionally used Arm in the past. I think it's going to become a necessity for those companies to think more about processor architecture, because so much more of what they do is going to be around the software and the hardware coming together, instead of a collection of closed boxes that get wired together and then optimized through some control level.
Yeah. Hi, it's Robert Sanders from Deutsche Bank. Just a question for Chris. Can you talk us through the increase in amounts recoverable under contract in 2015? It seemed to be quite a big move.
Are you talking about the total amounts receivable or?
The amounts recoverable under contract. In the past, that's been because of revenue booked that is net milestones but hasn't been delivered. Is that the reason why it's gone up quite a lot?
Yeah. That is revenue booked where we haven't yet hit an invoicing milestone. That is a consequence. To start with, there is nothing unusual about that movement. It is to do with the timing of the payment versus the timing of the delivery of the IP. Typically, say, under a subscription contract, people can be taking IP. We're recognizing the revenue, but actually, we haven't invoiced them for next month's subscription. I wouldn't read anything into that. For me, it's within the normal tolerances is what we'd say.
Okay.
Any other questions?
Hi, Doug Smith from Agency Partners. One part of the business you didn't talk about very much in your presentation is physical IP. Obviously, it has a different customer base. What trends are you seeing here? Maybe an update. Also, I noticed that the royalties there never seem to get bigger than the licenses. Is that something that's permanently the case, or is it somehow a different structure of how that business works?
I think when royalties get bigger than licensing is a timing thing. I remember thinking about 15 years ago that there would come a point in Arm's processor business where royalties are so much larger than licensing. The longer that remains the same, the better, really. In the physical IP business, we didn't go into it today, but the way that works, we've got to do a lot of R&D on advanced process technology to create the libraries then that chip designers download from our website and build chips and take out and buy wafers. Again, there's a long disconnect between the R&D that you do that drives revenue recognition on the licensing side and the royalties that come. What we've seen in that is that is a slightly different customer set. It's all part of the same supply chain.
In the foundry space, there's been a lot of consolidation there. There's only a very small number of people innovating on the leading edge. We're working with them all on next generation physical IP for the next generation process technologies. Meanwhile, the volume and the royalties is coming from the shipments of wafers with 28, 20 FinFET technology on. The adoption of those process nodes is very strong. We're seeing good performance on the royalty side of the physical IP business. The two things are unrelated. It's about developing the products. The volume comes in the future. Our engineers in that part of the business are very busy working on this next generation of processes, and that is quite cyclical. Moore's law cycles come around every 18 to 24 to maybe a few more months.
processes get introduced on that kind of clock tick, and that leads to some kind of cyclicality on the licensing side.
Actually, one quick question. I'm looking very far ahead. Is there ever going to be an Armv9 architecture?
Well, if you look back in history, you can see the three, four, five, six, seven, eight. One might assume that that's coming down the road, but periodically we do change our naming conventions. We have R&D work going on, obviously looking at everything I was talking about today. The next generation of computing devices will have different performance requirements. Look at the way the architecture's evolved over the years. We've delivered more performance. We've delivered greater efficiency to take advantage of what can be manufactured in a cost-effective way. These two things are kind of related, and there's a lot of work going on thinking about where that goes architecturally, how we roll that up into architecture products, and then into CPUs, et cetera. TBD, that's the fun bit for the future. A lot of work going on in that right now. Any other questions for Chris?
Darren?
I actually got loads, but I'll catch you afterwards. I just wondered whether you could talk about your market share in 32-bit microcontrollers. I think last year you said 65%, not of the total market, but of just 32-bit. Is that still a fair assumption? It's probably gone up a bit, but can you talk about your market share there and the feeling that it blends to that higher rate over time? Second question, unrelated. Do you have any customers that are over 10% of sales for the full year of last year?
I'll deal with that one because it's easy. If we do, we have to declare it, and we didn't. In terms of share in 32-bit micros, I'm afraid I don't have the number in my head.
It's higher.
Yeah, I'm sure it's higher given the volumes there. That's a number off the top of my head I don't have. Ian might look it up momentarily. If not, we can follow up with you on that.
Okay. Well, if there are no other questions, thank you all for joining us today, and we will see you either on the road or at our Q1 results in April. Thanks very much.
Thank you. That does conclude your conference call today. Thank you for participating. You may now disconnect.