Ladies and gentlemen, there are more than just us involved in this. We need to crack on, even though there are a few still signing in as they come in. It's my great pleasure as the Chairman of Arm Holdings to welcome you all here to our Investor Day. I'd like to start by welcoming Chris Kennedy to the team of more familiar presenters to you. Chris joined us as our CFO on the 1st of September, as I'm sure most, if not all of you know, but has very much been getting involved before that and coming up to speed, but also meeting a lot of the people both relevant to Arm, and is very much in the chair.
Today, we're going to be outlining increased investments in a number of areas, one of which building on established success and things which are going well in the interest of accelerating progress, and particularly three areas, particularly increased opportunities in mobile, accelerating our share gains in network and server, and also creating new revenue stream opportunities in Internet of Things. Of course, all of which then importantly leading to strengthening our returns. A lot to get through, therefore, without any further delay, let me hand over to someone who needs no introduction, Simon, who's going to kick us off.
Thanks, Stuart, good morning, everyone. Thank you for joining us today. Before we start, let me just direct you to the usual customary statements about the forward-looking statements we might make today. We're going to take that as read. Then to the agenda. As Stuart said, what we're going to cover this morning is how we plan to grow this business. We see many exciting opportunities ahead of us, and through this morning, you're going to hear me talking about mobile, what's going on in mobile, why we view this as an exciting business. Again, it's the backdrop of a change in growth rates per unit. We still see this as an exciting business, still see this as an area where there's going to be lots of innovation. I'm going to talk about some of the trends there.
After me, Pete's going to be up talking about what we've been doing in the enterprise infrastructure space, and by that, we mean networking and servers, and what we're going to be doing to accelerate our market share gains in this very exciting area for us. After that, Mike, our CTO, is going to be talking to you about what's going on in the Internet of Things. This is a space we've spent a lot of time talking about through our presentations during results and at other Analyst Days. We're going to bring you up to speed with, again, what we've been doing, the opportunities as we see ahead of us, and opportunities for us to grow revenue streams beyond the microprocessor within the microcontroller. Lots of talk today about the end market.
Of course, whatever we do here only matters as long as it does translate into money. It is great to have Chris on board at last with us and to help translate all of this good effort that we're doing into pounds and pennies. He'll be wrapping it all together with a finance view of how this opportunity translates into money. Later on, we'll be doing a Q&A. We'll have microphones going around the room, and I'm sure there's lots of questions that you want to ask. Let's make a start on the real content. I'm going to talk about what we're doing holistically to address the long-term growth opportunities that we see, and then specifically talk a little bit about mobile. Arm's going to be 25 years old this year, and our business has changed a lot over that time.
When we started out, we knew that this concept of embedding intelligence into chips and into end devices was going to be a good idea. If I'm honest, we didn't really know where that was going to take off, but we could see that as an overall trend, embedding a microprocessor into a chip surrounded by a whole load of other functionality by many semiconductor companies was going to lead to fascinating new consumer products and change the way that computing was going to happen. The technology that we had at the time when the company was formed was a very small, very power efficient, but very powerful microprocessor. We set about building a licensing business to make that a pervasive standard and build an ecosystem so that software could be written for that processor. We got our first real success with this technology in mobile.
As mobile phones became digital mobile phones, as GSM and as other digital standards took off, we were in a very fortunate position. We were at the right place at the right time, and we developed the right technology to enable a powerful 32-bit microprocessor to be put into a cell phone at very low cost. Of course, the units took off, and that helped grow Arm's business. As we grew, what we saw was that the companies that we worked with started using our technology in other markets. That led us then to invest in technologies to allow the Arm architecture to be used in much more powerful applications and also in much smaller applications, these tiny microcontrollers that we're going to find everywhere as the Internet of Things develops.
We've grown up with the technology, we've grown down with the technology, and we're in a position now where as computing changes, computing is becoming pervasive everywhere. We see an opportunity for everywhere that that computing is happening, for that to happen utilizing Arm technology. That's the opportunity that we see ahead of us, and we are investing in our business to really take advantage of that. We're going to talk today about what we're doing to address some of these markets particularly. Before we look to the future, let's step backwards for a moment. If we go back to 2004, Arm's partners, our licensees, shipped 1.3 billion chips, and we were pretty proud of that at the time. I can remember us celebrating, actually in a pub not too far from here, the billionth unit on an annualized basis.
When we look at the breakdown of those units in 2004, the vast majority of them were going into mobile phones. As I said, this growth of digital cellular had taken off. In that year, there was about 700 million phones shipped by various OEMs around the world, and virtually every one of them had an Arm microprocessor in it. That was a great thing. We were starting to see shipments, though, in other markets. Our licensees who had success with Arm in mobile were starting to use Arm processors to address other markets. We were starting to see unit growth outside of mobile. As the chart shows here, you can see the breakdown of that in 2004.
About 20% of the units being shipped at that time was in what we classified as enterprise, and that was largely hard disk drive, 7% in the home, some TVs, home Wi-Fi routers, that sort of thing. About 5% were embedded. We were starting to see the first microcontrollers built using an Arm7, which was a product we hadn't designed for microcontrollers. All of these other uses for Arm were really being derived from products that we had designed for mobile. That unit breakdown painted a picture of a company that was very mobile-centric. Indeed, the opportunity around mobile looked very exciting ahead of us, and we were doing a lot of work to make sure that we maintained our lead in the mobile space.
If you look at the licensing that was going on at the time, that paints quite a different view of the company. We were, of course, getting a lot of repeat business from our licensees who were focused on mobile. We were starting to see more and more companies coming to Arm, seeing the benefits of the technology that we had, and seeing opportunities to embed a microprocessor in chips that were targeting other markets. We were starting to work with companies who were focused outside of mobile. As you can see there, the breakdown of the licensing really shows a different view of what might happen in the future. Microcontroller at that time was a particularly interesting space for us. In 2004, which is where this data's from, that is when we launched the first processor that we designed specifically for microcontroller, Cortex-M3.
We saw huge adoption of that, and that was driving the embedded side of the licensing. The units were showing one picture, the licensing was showing another. If we fast-forward back to 2014, our last full year, what that change in the shift of licensing, change in our emphasis has resulted in is diversification of the shipments of product into various end markets. You can see here now that less than half the chips being shipped by our licensees are in mobile. We've seen big growth in embedded. On a proportional basis, actually a reduction in home and in enterprise. In absolute terms, that's very different. If we think about enterprise, back then, as I said, that was largely disk drive. Now we're seeing the adoption of Arm technology in base stations, in switching. Pete's going to talk about this later on.
Last year, that 16% of our volume amounts to 2 billion chips. That's obviously a lot more than the entirety of the chips being shipped in 2004. The diversification in the licensing has led to a diversification in the end market and rapid growth in volume of Arm processors in mobile, but also outside mobile. We look today at how the licensing is going, we look at the market that customers are adopting Arm technology to go and address, there's still a lot of repeat business of an ongoing roadmap of technology for mobile. We're seeing growing licensing in other markets, which again, when we think about the future, points to more units being shipped in markets outside of mobile. In 2004, we had about 140 licenses. We've now got over 400.
We're working with companies really targeting a vast array of end products in many different end markets. This diversification has come from investments that we've made in technology. What we first started seeing was people take the things that we designed for mobile and start to use them in other markets. As we achieved success in these other markets, we started to diversify our product line so that we could have greater success in these end markets. This is a philosophy that we've followed now for many years. The net result of that is that we now have a broad market penetration in many different end markets just beyond mobile. We have a very high share in smartphones, in mobile devices that I'm sure you're all carrying around with you. Over 95% market share in smartphones.
We've continued to gain share in some of those early markets that we were successful in. Hard disk drives, for example. TVs have become digital TVs, and a vast majority of those are based on Arm technology. Of course, microcontrollers have grown rapidly. The Cortex-M series has been very successful in the transition from 8-bit microcontrollers to much more powerful 32-bit microcontrollers. We've grown in the markets where we had that initial success, but we've also seen Arm become the technology of choice in new markets. Tablets have appeared, wearable devices have appeared, our licensees who are looking to target that space are looking at the technology that we have and seeing all the attributes of small die size, which equates to cost, very low power, great performance, as the key attributes that they need to address these emerging markets.
This relates to very broad market penetration, but it's only happened because we've invested in the right technology. I think over the last 25 years, we've invested in that right technology at the right time to capture a growing market opportunity. That's what we've been doing, we're going to continue to address the future in this way. We're going to expand our product portfolio with the right processes for the right markets, hitting that trade-off between performance, power consumption, features, die size. We're going to broaden our portfolio to make sure that we can address that market with the right technology. It isn't just about processes.
We're introducing new technology through organic investment, and we're complementing that with acquisitions that we're doing to bring the right technology into Arm, to complete the picture of what our customers need to go and address the market and to generate new revenue streams for us. Vitally, though, it isn't just about what we're doing, what our licensees are doing. We pride ourselves on a very broad ecosystem of partners. People with companies that we work with, and there are thousands of them, who are building their own businesses, utilizing technology that we create, and in support of the Arm architecture. There is a very wide ecosystem of companies who are all doing work, building, creating profitable businesses in their own right, that really do amplify what we do and what we do amplifies what they do. That ecosystem is really, really important.
It's important for the markets that we're in, and it's really important for the markets that we're growing into. This approach, the right technology, supplementing it through our investment, building that ecosystem, has enabled a sustainable growth for Arm over a very long period. We've been able to create a more valuable technology that's allowed us to generate more royalty on a per microprocessor business. It's enabled us to create more technology that our customers are incorporating into each of their chips. It's enabled the Arm partnership to accelerate the gain of market share in these new markets. It's created greater stickiness for our product. We want our licensees, our partnerships, to use Arm technology over and over. Everything that we've been doing is creating this environment where it's an easy choice to use Arm.
All of that has led to us being able to increase the amount of royalty that we get on a per-chip basis. It's enabled us to grow market share, and it's enabled us to create new revenue streams from new technologies and new markets. This philosophy has created sustainable growth for Arm. I see this really continuing into the future. It's what we've done for many, many years now, and it's what we intend to continue doing. A lot of our engineering is focused on the roadmap of products that we have defined today. We work closely with our customers to identify their needs, where the market is going. We take that information to drive our product roadmap, and Pete's team, he'll be up later on, are then charged with executing on that roadmap, delivering the product.
We also have a team of engineers who are kind of slightly outside of that day-to-day effort of delivery. They're thinking about longer term, where does technology go? Under Mike Muller, our CTO, we have what we call our R&D team. It equates to about 10% of our engineering headcount. They are focused on kind of outside the main day-to-day, thinking about the long term of technology and running experiments to see what things we want to incorporate into our product portfolio, where we might want to build relationships, how we might want to work with universities. It's looking outside just the heads-down day-to-day work. That effort's been really important for us. Over the years, it's created architectural features that we've built into our main processor roadmap that's enabled some of this market share gain that I've been talking about.
It led to, ultimately, some of the embedded technology that Mike's going to be talking about later on. With our own R&D, we've complemented that with acquisitions. We've brought in technologies such as our display IP, another building block that many of our customers need to put inside their chip to allow a display containing graphics and video to be put together and out actually onto a screen. That key technology came from an acquisition. We'll build that up, and we're now starting to see the benefits of that in licensing with royalties to come in the future. Long-term R&D, acquisitions are the way that we have delivered this growing technology portfolio. We've done some larger acquisitions. Our physical IP business came from, in fact, the largest acquisition that we've done. We're looking into the future and thinking about the complexities of building very advanced semiconductor devices.
It was clear that we were going to need to enable our licensees to get a more optimized solution in a more predictable and shorter time to market way. That is how we've driven our physical IP business. This combination of organic investment and acquisitions has allowed us to grow. It's enabled us to create now a much larger range of products that we're out licensing. In the early days, back in 2004, in fact, beyond then, earlier than that we had the ARM7 processor. We had some of the basic building blocks that our partners needed to put a chip together around that, and we've grown it. We now have a very broad portfolio of technology driven by the needs of the end market. Where is the end market going?
What do our licensees need to put their chips together to complement their own R&D with technology that we can provide? That's led to a very broad portfolio of technology that are enabling some of the world's most sophisticated chips to be built. We add a few new processors every year. We've got some more coming that we're going to be announcing the details of later on this year. In 2014, when we look at how many different processors did we license, it was about 20. We've gone from the early days of a processor, another processor, to a portfolio now of about 20 different processors that we're licensing concurrently. The product line evolution, though, isn't just about performance. Everybody wants more and more performance, lower and lower power.
As we've evolved this product portfolio and as we've thought about the needs of some of these other markets, it isn't just about performance. It's about having the right features. It's about making sure that the processors are as small as they can be, and it's about making sure that the systems that our licensees are building around our processors are as efficient as they possibly can be. We've built our portfolio to address now a very wide range of markets. The breadth of technology that we have enables many different end markets to be addressed. We've enabled technology for the embedded market, these microcontrollers for real-time applications. When you hit the brakes on your car, you want it to stop and not think about it first. We have technology that helps make that a reality.
We've acquired technologies to help grow the overall offering that we have in any one market to help accelerate time to market for our licensees and generate recurring revenue streams for us. A great example is the radio technology that we recently acquired from a company called Sunrise Micro Devices. We've complemented that with software to put the Bluetooth stack together and can provide a much more complete solution now for somebody wanting to build a chip targeting IoT. The breadth is as wide today as it ever has been. We're going to continue in this way. We see many exciting opportunities ahead of us, and we're going to continue to invest in our core roadmap, and we're going to continue to look for opportunities to bring teams and technologies in that we can grow.
We're going to grow organically, and we're going to continue to grow through acquisition and, crucially, keep building that ecosystem. As we look at some of the new markets, there are players that we've never worked with before that we need to go and engage with to create the complete software offering to enable some of these new markets to really be addressed fully. You'll hear about some of the details of how that ecosystem is expanding in the later presentations. Through this way then, through this philosophy that we have to investment, we see that we're going to continue to grow a sustainable business with growing revenue per chip and recurring revenue streams. Now, a lot of the investment that we make is in our engineering headcount.
Pete is always coming in the office saying, "I need to hire more people because of this opportunity that we have." Sorry to single you there. We are investing in the headcount to grow the opportunity. When we look at the engineering headcount devoted to just mobile product, this has been going up and up over the last few years as the market has grown. It's not just in engineering. We are growing the company as a whole. The supporting infrastructure that we need to make the business efficient and effective is an area that we're investing in as well. Mobile remains a very important market for us. It's the major driver of our business. The products that we create, we're able to morph into other markets as well.
As we've seen this opportunity grow outside of mobile, we're investing in engineering teams specifically to go and target that market. In parallel with the headcount growth focused just on mobile, we're growing the number of engineers who are working on products outside of mobile. And in fact, the number of people that we have working on that, we expect to grow by about a factor of three in comparison to where we were back in 2012. In mobile, outside of mobile, we see significant opportunities for growth. You're going to hear about the way we view the enterprise infrastructure market and IoT in some detail later on. For the time being, I just want to focus for the next few slides on what's going on in mobile. As I said, mobile remains a very important market for us.
The proportion of the units that our licensees ship into mobile is still very, very high. The royalty revenue from that, very, very high. It's an important market, and one which is in a phase of transition. I want to spend a little bit of time talking about what's going on in the mobile market, what it means for our business, and how in a world where growth is slower, how we still see a great opportunity for growth and an exciting opportunity to provide more and more value into mobile devices, which we're all carrying around, are going to become even more sophisticated. The first point I want to make is about the market size. Today, we have about an 85% market share of mobile devices. You may be thinking, "Arm used to call this about 95% market share. What's happened?
Has it gone down?" It hasn't gone down. What we've seen is as the capability of our product has expanded, there is an opportunity for us to target more than just smartphones. We've seen obviously the tablet market grow. We're seeing the performance of Arm technology and the chips that our licensees are building around it really achieve very, very high levels of performance. Therefore, we believe that we should consider the entire mobile computing market to include laptop devices within the addressable market as we see it. When we think about that redefined market, then we consider that we have a greater than 85% market share. Mobile smartphones is still a very high proportion of that. One of the key things that's getting a lot of commentary at the moment is how the growth rates are changing.
The growth rate over the last few years has been very, very high, 40% CAGR over the last five years. I mean, that's a growth rate that very few markets go through, and they certainly don't sustain it forever. We're going into a phase where growth rate is going to slow. It's not going to fall off a cliff. It's going to slow. We believe over the next five years, we're looking at about a 7% CAGR. This isn't just our data. This is market data taking into account what we see, what we hear talking to our partners working in this space, and what the world analyst community who go to stare at markets believe how the market is going to shift.
Given this lower growth, and again, that's higher than many markets you might come across, but given that lower growth, why are we still excited by this market? Well, first of all, it's a huge market for silicon. Even in that lower growth end unit world, the amount of silicon in devices is going up. The sophistication of those silicon devices is going up. The market analysis that we see predicts that the value of all the silicon sold into mobile devices goes from about $15 billion-$25 billion by 2020. That's a lot more silicon. All of those silicon devices, we believe there's an opportunity for an Arm processor and an opportunity to create a royalty for Arm. Whilst the growth may be slower, there's an opportunity for us to have an increased content, an increase in the content per device.
A lot of that assumption is built around the introduction of version 8 of the Arm architecture. That is something we have been talking about for a few years now. Version 8A, which is the variant that is specifically targeted at application processors, is seeing very rapid adoption by chip companies looking to target this space. When we think about last year, version 8 was in about 10% of the application processors that were sold last year. We think come the end of this year, it will be in about 50% of smartphones that are being sold and up to about 90% by the end of next year. The other factor that is going on, as well as the adoption of version 8 of the architecture, is the increased core count in smartphones. Many devices now are being designed around an 8-core device, an octa-core processor.
We think of the version 8 architecture application processors that are being sold by the end of 2016, about half of those will be based on octa-core implementation. That drives more royalty for Arm. That is something that is going to grow. About 20% of the v8 processors sold in Q1 this year were octa-core. We think that is going to grow steadily over the next couple of years. More Arm processing in each of these devices. It is not just the CPU. With continued attach rates of our Mali graphics, of our video technology, of our display technology, there are other ways in which our royalties can grow beyond the processor.
Although the end unit growth rate may be lower, the amount of Arm content per device, we believe is going to continue to grow, and that is going to lead to increasing growth in Arm's royalties from mobile. Our goal is absolutely to maintain our share here, to grow our share as best we can, to maintain our share in smartphones and tablets, and grow in clamshell devices. That is a new opportunity that is afforded to us by virtue of the performance and the energy efficiency of the technologies that we are delivering. A lot of the improvements you have seen in your mobile device over the last few years has been in the kind of conventional parameters. It is speed, it is pixel size of the screen, it is connectivity. You are going to continue to see that.
I mean, even just in the last week, we saw Sony, I think it was, launch a phone with a 4K display on it. Screen size is going to continue to go up. As we get to 5G, connectivity is going to go up. One of the things that may be less obvious in your device, when you go into a phone store and look at it, is the increasing amount of sensor technology that is being built into the handset. That is going to continue to grow. You are going to see your handset interacting with other devices, information being shared across devices. The interface to your device changing really quite fundamentally. I mean, if you think about the early implementations of voice control, it is pretty clunky.
I mean, in my car, I have to pretend to put on an American accent at times to make a call, depending on who it is. That's going to get a lot better. It's going to get a lot better because of some of the sensors, the way the multiple microphones and noise cancellation, and it's going to get a lot better because of the processing in the device itself. Your handset is going to start to communicate with other things, your car, your home, your office space. Fundamentally, you need more processing and more sensors to make all of that a reality. That is more content that will be in our pockets over the next few years. The device is going to improve. It's going to take on what appears to be artificial intelligence. We'll talk about that in a moment.
It's fundamentally driven by version 8 of the architecture. We have designed this to be the computing platform for the next generation of smart mobile devices. One of the things about version 8 is that we've introduced features to make it easier to build these multi-core devices, clusters of multi-core processors. We've made it easier to write software that runs across these devices. We've made it easier to partition that software from other pieces of code that you might be running. As we become more and more dependent on the smartphone that we carry around with us, security, confidentiality of the materials that's on the phone become more and more important. Again, we've added features within version 8 of the architecture to make security better, to make the device more trustworthy.
Security and trust go hand in hand in a really important attribute, not just the phone, but pretty much everything that we're going to do. Again, you'll hear a bit more about that later on. Version 8 has been architected with the future in mind, thinking about some of these new use cases, thinking about the performance that you're going to need, and designing the architecture so we can scale up and scale down across those performance needs and add those key features. One of the other things that I think is really exciting about the mobile phone space, the mobile device, is how competition is driving innovation. We're seeing still many OEMs in the market, and they all want to be number 1. They're competing with each other, and they're driving the need for more sophistication in the silicon devices.
One thing we hear regularly from our partners is that they need to provide more sophisticated chips every year, and therefore, we need to provide a platform of IP that becomes more and more sophisticated every year to match that cadence of product introduction. When you look at our product roadmap, as shown on the slide here, lots of part numbers, lots of names, but you can see what we're doing for mobile is introducing a new family of technology on an annual basis. This is specifically to meet the needs of our customers to allow them to meet the needs of their customers. More technology is being introduced frequently, which is why Pete needs more engineers every five minutes than I ask him. It's driving an opportunity for increased licensing in our business as well. We have many new products coming.
You're seeing many code names here. I'm sure people are jotting these down and wondering what they are. We'll tease you about that later on. Obviously, there'll be formal product launches at the right time. I often tell people that microprocessors are useless without the software that runs on them. I spoke about how we've developed our ecosystem to really ensure that our technology is usable, not just by our licensees, the people building the chips, but by people building systems around the chips. As well as working with foundries on manufacturing, with EDA companies on design tools, we spend a lot of time collaborating with our partners to create the right software environment to run on Arm. A lot of the software that's running on Arm platforms is based on open source.
We saw a need to collaborate within our industry of partners on the optimization of open source for Arm. Five years ago, we created an organization called Linaro. At the time, that had just a handful of companies, about five or six companies that participated. We brought about 20 engineers together to form the heart of an organization that would optimize the Linux kernel on behalf of everybody who wanted to run that on an Arm processor. Over the years, that has become a really successful endeavor. We now have about 30 different companies. We have about 250 engineers from across the partnership working together to optimize different parts of the open-source stack focused on different end markets. We have programs running in Linaro focused on networking, on enterprise, on mobile, on portals, on home consumer electronics. Many different programs are running within Linaro.
This has been very successful for us, and it's enabled more optimized software available for anyone then who wants to go and address the market. I believe has accelerated innovation. It's to the point where now Linaro is actually the third largest contributor to the Linux kernel, and that really is quite phenomenal. Just in conclusion then, we're going to continue to invest in this business. We're going to continue to invest in mobile. It's our most important market. A lot of what we do in mobile flows down into other markets. The technology that we create forms the basis for products that we produce for other end markets. As we continue to invest, as we grow our portfolio, that enables us to provide more value to our licensees, and we get paid for that value that we generate and that we provide.
It enables our partners to spend their R&D on the things that really do create differentiation for their end product. We're going to add technology organically, and we're going to continue to add technology through M&A, and then investing in the teams and the technology we bring in, and in parallel, investing in that ecosystem to make sure that anyone contemplating building an Arm-based system has everything that they need, whether it's from us or from one of our partners. Mobile, really important. What I want to do now is switch gears and start talking about some of the other markets. Later on, you're going to hear about IoT from Mike. Next, we're going to spend a bit of time talking about the opportunity that we have in enterprise infrastructure. By that, we mean networking, and we mean servers.
I mentioned Pete a few times. He's the guy with all the engineers. His day job is making sure that we have the right roadmap of technology and that we have engineering teams focused on delivering it. He's going to describe what we're doing to address these markets. First of all, I want to show a short video, which is going to show what's going on, why this market is changing, and therefore, why we see it as a really exciting place for us to be. Let's play the video. Thank you.
Today, we are increasingly reliant on our connection to the cloud. It allows us to exploit the value of the sensor-driven devices many of us depend on. Our phones, our tablets, cars, home appliances, and wearables, all producing and consuming more and more data that enriches our lives and makes us more productive. Between these devices and the cloud lies a vast and complex network. All around us, devices capture and transmit a constant stream of information to the cloud for processing. At the same time, massive amounts of data, such as video and audio streaming, flows back to the endpoints for consumption. This churning bi-directional traffic can quickly clog vital data arteries. By creating a more intelligent, flexible cloud that allows cloud applications to extend out into the network closer to the endpoints, total network congestion reduces and application performance greatly improves.
For example, sensor data can be analyzed and aggregated to reduce traffic going upstream. Content servers can be pushed closer to consumers, improving aggregate bandwidth and responsiveness. In the intelligent, flexible cloud, these new services can be adjusted dynamically between nodes based on the evolving application mix. Flexibility also means the network can scale more easily as bandwidth grows, making an intelligent, flexible cloud the solution to address future devices and services.
Good morning. For those of you who are listening to that on the phone, the video really helped, I'm sure. What I'm going to talk to you about today is some of the work that we've been doing in enterprise infrastructure. We've been working in that area for quite some time. We're very excited with the progress, I'm going to tell you where we are and then how we're going to invest to accelerate gains in that space. For me, I think we're now at the tipping point. We're now at the tipping point where we have silicon partners out there in both networking and servers. We have the software ecosystem ready. We have big demands from OEMs, from cloud vendors, from operators for Arm technologies, and now is the right time to invest a small amount, a small and limited amount, to accelerate our gains.
I am going to cover exactly what we are going to do in that area. Why are we excited by this space? It is a very, very large space for us. As Simon covered, the mobile TAM for us is about $25 billion. If you sum the enterprise networking and the server TAM, it is $38 billion. Even more exciting is we have a small share in that. In mobile, we have 85% of a $25 billion market. Here we have a very small share in a $ 38 billion market, the potential is extremely large. What we covered on the video was the concept of intelligent, flexible cloud. I think a few of you have been through that before. What that basically means is we are moving from a world where we have big data centers, a fairly dumb pipe out to gateway devices or terminals.
With the intelligent, flexible cloud, intelligence is being spread across the network. We are getting more compute, we are getting storage, we are getting acceleration spread throughout the network. The networking and server worlds are colliding, which is very nice for us. It is very nice for our partners because what the OEMs who are deploying this technology want is they want the same architecture across all points. They want an architecture that would scale from a lamppost all the way up to the DataCenter. In terms of efficiency of development, it is also very nice because what it means is I can develop technologies which are going into networking, which are also applicable into servers and vice versa. We have server technologies which we are developing, which we are selling to enterprise customers. It gives us a larger customer base to deploy the new technologies into.
No one supplier can provide all of these devices. It is impossible. There are just too many devices, too much optimization needed throughout the network. The only way that customers, that OEMs, that operators can access solutions in this space is through something like the Arm partnership. A partnership whereby we are developing the base technology, putting it out through all our silicon partners. They are deploying it, they are customizing it, they are innovating on that, and then they are delivering it to the end users. It has to come from a wide range of partners. To win in these markets, we need three things. We need a scalable portfolio of technology based on a single architecture, a portfolio of technology which can scale across all points. As Simon covered, we already have technology which we are deploying into mobile. That is also being deployed into these markets.
In addition, we are investing to develop point solutions for both networking and servers. We also need a rich software ecosystem. The ecosystem in networking and servers is very large. It is very diverse. It is being ported to Arm. We need to keep momentum going in that. The critical point is we need a set of end OEM, a set of operators, a set of cloud suppliers who are pulling on this technology. The reason they are pulling on this technology is because they want software-compatible range of solutions. They want a range of suppliers who are competing with each other, a range of suppliers who are delivering innovative solutions, a range of suppliers whom they can actually use to trade off in terms of pricing flexibility. They want a wide range of suppliers, but at the same time, they want an identical software architecture.
Otherwise, they have to redo their software every time they change a supplier. This is the standard part of the Arm business model. This is why we've been successful in mobile. This is why we've been successful in our existing markets. This is why we're going to be successful in enterprise. Let me take you through networking first, and then I'll take you through servers. In networking, every major networking chip vendor has licensed Arm technology. You can see a range of logos here, you can see a range of designs here. We have 30 platforms shipping. This gives the Arm architecture one of the widest deployed bases in the industry. I would spend a few minutes just looking at the range of partners we have up there. The result then is that the OEMs can select from a range of vendors.
They can have competition, they can have differentiation, they can have innovation in this space. They still need to retain their software investment. What we've been doing is we've been working with things like, the Open Platform for Network Function Virtualization. We've been putting together new ecosystems, largely based on open source. The great thing about the networking space is a lot of it is moving to open source. My engineers, customer engineers, can work in that space. The collaborations in that area, things like the Open Platform for NFV are all based on Arm technology. We have a large partnership, large number of partners working in that space, all contributing back to the same community. Simon mentioned Linaro. Linaro is a key part of that, through our LNG group in Linaro.
A lot of this work is pulled together, Linaro is now the number 3 contributor to open source software. We have talked and we have updated you on our progress in the networking market for some time. I think last year we said we're around about 10% of the networking market. That's actually fairly concentrated. One of the big areas it's concentrated in just now is wireless access. This is base stations, this is remote radio heads, and in this particular area, in this particular subsegment of the market, this year, we're going to be about 30%. 30% of the shipments into this space are based on Arm designs already. By 2020, we see that moving to 75%. For us, this is a standard way of deploying.
We find a subsegment of the market, we deploy technology into that, we get the ecosystem working, we build very good relationships with partners, and then we move to the next subsegment. In wireless access, 75% by 2020. Simon talked about my engineers. Thanks very much. They're lovely. I'm going to take you through some of the investments we're making in this space. Clearly, we're going to be investing in cutting-edge SoC technology. In networking, that means we are designing processors in conjunction with lead partners, some of the 30 partners I showed you earlier. We are developing network-specific processors. Processors which are designed purely for networking. We're also taking existing deployments, existing software, existing workloads, running analysis on that in our labs, and feeding those results back into current designs and future designs and future generations of our technology.
We're also investing in next generation software adoption. We are leading open source developments in those places. We've had a lot of requests from our partners to invest more in this area, to work very closely with them, and that's what we're doing. A key part of getting to volume, getting to market is enabling end OEMs, end operators to deploy to market. There, we're working very closely with end operators and end OEMs at an engineering level. The three areas of investment. If I take you through some fairly detailed examples here. What you can see is we have three separate architectures here for base stations, customer premises, equipment, and Cloud RAN. You can see one size definitely does not fit all. There are very different SoC architectures here, very different mixes of CPUs, different interconnect, network processors, accelerators. They are very different.
They are very customized. You need to do that. We need to do that to get the throughput in the network. We need to do that to get the power efficiency in the network. We need to do that to give the differentiation to the end OEM. Three very separate architectures, all based on Arm technology, some based on generic CPUs, some based on very optimized CPUs. My engineers, my teams are working to produce customized solutions in that space. It's not just processors. We're rolling out new interconnect in these areas as well. What that gives the OEMs is a range of devices from highly accelerated at one end all the way to massively multicore. They get a very wide range of highly differentiated solutions from the silicon partner.
Of course, the problem then is you then need to leverage your software investment across different semiconductors. There's no point in working on one particular solution and being unable to move. As the solutions are more and more customized, it can be difficult to move. What we've been doing through Linaro is coming up with concepts like OpenDataPlane. What OpenDataPlane allows is it allows the OEMs, it allows the vendors to actually develop their software at a very high abstraction layer and take out a lot of the acceleration complexity. When they run it on the device, it moves either from software to hardware. This is a key facet of deploying optimized solutions. We have to be able to deploy them, get them adopted, and still allow OEMs to retain their software investment. As I mentioned, it's all very well.
We have the hardware out there, we have the software out there, we have the OEM pool. In some cases, we need to get that kick-started. We are putting engineers with the customers to help get these things deployed, to help address various segments and then build on that. We're going to track this on a regular basis. I think you're used to the so-called SMARTIE chart. We're going to produce one just for networking. It's slightly different to our normal charts in that this is based on OEMs. Our normal charts are based on semiconductor partners. This is not. This is based directly on OEMs who are shipping on-based silicon.
If you looked at this last year, if we produced it last year, you would have seen wireless access slightly more orange than green. You would have seen areas like wired access and enterprise would have been all red. What you can see is we are moving OEMs gradually. We're moving OEMs from no Arm design wins to Arm design wins to starting to ship Arm design wins. Clearly, the end result of this is we want them all to go blue, but getting at this point from red to orange to green, I think is a good step forward. The sub-segments that we're going to focus on next, wireless access is doing very well. As I say, 30% market share this year, growing to 75% in 2020.
Wired access, enterprise equipment, and storage and security are the next areas we're focused on. You can see that they will gradually turn the right color. Having made this investment, having gone through this progress, I think last year we talked about gaining a 35% share of a GBP 18 billion market. This year, with the investments we're making, we're upping that. We're increasing that to 45%. By 2020, we expect the Arm partnership to be addressing or taking 45% of the enterprise networking market. That's networking. Very similar pattern in servers. The servers are deploying. We have server momentum building. There are six chip vendors in production already this year. I'll cover the details of those in a minute. We have at least four more expected in 2016. A lot of silicon is out there now. A lot of silicon is being produced.
Actually, some of it is outside in the toy. Very similar pattern in terms of software ecosystem. The first commercial OS was deployed in 2014. There's a large number of additional OSes which are at the developer stage, which we are working with people to get to commercial deployment. Again, one of the particularly nice things about this market is all the major companies in this space, and OEMs, all the major cloud vendors are evaluating Arm-based solutions. Now the solutions are out there. People can evaluate them. People are researching in the lab. People are starting to deploy them. Very happy with our progress in this space. Here's some examples of the silicon partners that are already shipping. You can see on the left-hand side, the partners that are shipping already. Six different forms of silicon.
On the right-hand side, you can see solutions that are on the horizon. Solutions that have been announced, there are more in the works, but these are the ones that have been announced already. You can see it's a very diverse range. We have companies in the U.S., we have companies in Europe, we have companies in China. We have merchant semiconductor vendors, we have OEM teams. A very broad base of silicon partners. The Arm silicon server ecosystem is very, very healthy. What that ends up in is devices, is racks for people to buy. Again, you can see some of these things outside. I think you can see from these pictures, again, one size doesn't fit all. There's a lot of competition. There's innovation in this space as well, not just on the base silicon, but in terms of solutions.
Some of these solutions are optimized for specific markets. There are storage-optimized solutions, there are networking-optimized solutions, there are compute-optimized solutions. If I give you some examples of the deployments that we have ongoing, we have a range. We have the standard cloud compute vendors, such as OVH and Data Center. They're deploying Arm servers into the cloud, you can go and access Arm servers online right now. Folks internally like PayPal using solutions for things like data analytics. This is purely internal. One of the things I think is quite interesting is we have multiple HPCs, so high performance computing deployments as well. We have them across a range of geographies. There's the U.S. folks, there's Sandia, there's the Barcelona Supercomputing Center.
One of the things you can take out of this, one of the things which I think is a big change over the last year is we're now addressing the entire market. The Arm partnership is now addressing the entire server market, from low-end servers all the way up to high-performance computing. In fact, I was talking to a memory vendor the other day, they said their most challenging demand for memory bandwidth, memory performance, are coming from Arm-based products. Again, we're doing investments in this space. Again, we are developing technology purely for server, purely for enterprise. Luckily, it spreads across networking as well. Some of the products we're developing go into networking, some of the products we're developing go into servers, but there are some very high-end processors coming which are just purely for servers.
Again, we're working on optimizing the software. There are thousands of packages out there already for Arm, tens of thousands of packages out there for the Arm servers. We're optimizing those. We're working with partners on those. We are getting the deployments to volume. Again, it's great having the technology there, it's great having the ecosystem there, but we want commercial deployment. We're investing directly with end users and cloud vendors to get those proof points out there to make sure that we're deploying things and then feed that back. In terms of development, clearly we have the developments we do internally. We have all the processor developments. We have the interconnect development. We also work very closely with foundry. Some of our server partners are implementing right now on 16 FinFET.
Some of them are targeting 10 nanometer, some of them are targeting seven nanometer. What we're doing with our physical IP is, again, we are producing physical IP, which is targeted purely for servers. Very different constraints, very different requirements in the server space to a mobile space. We're customizing everything from processor technology, interconnect technology, and physical IP technology, just for servers. We're working very hard with the developer community. This is not just an Arm thing. This is an effort across the entire partnership. The great news is we are leveraging, clearly, resources at our partners out there in the developer communities as well. We're helping all those developer communities. We're helping them by providing them with access to ARMv8-A hardware. 64-bit hardware is out there. We're providing it direct.
You can also go online and get it from the cloud vendors. We're conducting proof of concept with software partners, we're working upstream in the software community to create Arm versions of those platforms. All the work everyone is doing is aligned, it is going back into the open source community. What this is helping do is expand developer communities on Arm. We're leveraging the Arm partnership. We are leveraging the developer community. Now, as I said, initial deployments. We have initial deployments in storage, we have initial deployments in cloud, we have initial deployments in HPC. That's great. Very happy with how that's going, very happy with how that's rolling out. We are investing further. No part of the market is now restricted to us.
We are going after, the Arm partnership is going after every single bit of that GBP 20 billion TAM. Now again, we are going to be communicating our progress in this space to you. We'll break it down in terms of how we're doing in high-performance computing. Right now, I think I covered it. The deployments are going on at national labs, research institutes. We have four with advanced proof of concept in HPC. Not a very high volume market, but a very high value market. Cloud deployments, lots of evaluations going on at different cloud companies. Some are rolling out, some are in the process of rolling out. Enterprise, we have a lot of enterprises starting to evaluate the technology internally. Now, last time we talked to you, we said we were shooting for 20% of the server market by 2020. Today, we're increasing that.
With the successes we've had in this space, with the customers that are rolling our product, with the ecosystem investments that we're making, we're going for 25% target share in 2020. That's 25% of that GBP 20 billion market. Hopefully I've covered how we're investing to accelerate our share gains. We're very happy with how we're doing in networking, very happy with how we're doing with server. Now is the time to do small, focused investments to accelerate our gains. A lot of focus on the technology we're rolling out, a lot of focus on the software ecosystem, working directly with end OEM to get those solutions into market. Thank you for listening. We have a 15-minute
Good afternoon. Good afternoon. I'm in the wrong country. Good morning. I'm Mike Muller, the CTO. I appreciate that I am between you and the money, I will try and keep this relatively short. I was told that as CTO, I was allowed to lose some of you some of the time, but I wasn't to be too techy. We'll see how I go with that. I'd like to actually start with a little bit of history. Three of you in the room are old enough to remember the BBC Micro, which was an education project to teach children in the U.K. how to program. That was in a company that started the first Arm development, we created the first Arm silicon in 1985. Five years after that, Arm was spun out as a separate company in 1990.
We're now coming up to celebrate in November our 25th anniversary. As part of that, it's nice that we've actually gone full circle back to the BBC and the launch of the BBC micro:bit, which you'll see outside, which is effectively the BBC Micro for the modern generation. If you look at that BBC micro:bit, and compare the processor in that, a Cortex-M0, with the first ever Arm that we developed, ARM1, 30 years ago. I find it quite interesting that they are about the same design team, six man-years to do the original Arm, about 11 man-years to do Cortex-M0. They have about the same number of transistors, 25,000 versus 48,000. The technologies that they were built on are very different. 20 nanometers for Cortex-M0 compared to the 3,000 nanometers or 3 micron as we knew it at the time.
If you look at that's a difference of about 150 times on the linear dimension. Chips are, of course, two-dimensional, so that's about a 20,000 times reduction in area. You could fit 20,000 Cortex-M0s on the original chip that we designed. That's kind of Moore's Law. That 20,000 scaling factor is what has either turned a GBP 10 chip into a GBP 0.10 chip or allowed us to have all of the complexity that we have in a mobile phone today. We thought it'd be nice in my research group to do a 30, 25-year anniversary edition Cortex-M0 and see what would happen if we built a Cortex-M0 on effectively a 2 micron or 2,000 nanometer process to take us back in time to what technology used to be like. We produced this anniversary edition, Cortex-M0, and here's a wafer.
Now, of course, normally on a silicon wafer these days, they're a little bigger than this and you'd have 10,000 Cortex-M0s on it. Because this is old style technology, we only get 12 Cortex-M0s on the chip. There's one thing that actually really makes this anniversary edition for me really exciting, and that's that you can do this with it. Okay. It's not a silicon chip. This is actually printed on plastic. Because of that, you can eat it, you can wrap it around a fizzy drink container, you can spool it out on vast printing machines. You can start to do all kinds of things that I think open up new opportunities for IoT. Now, this is a research project. We partnered with a company in Cambridge called PragmatIC, who do plastic printing technology.
The chips that they're doing for commercial deployment today have a few hundred transistors on it and are really small. We pushed them to the absolute limit to make a chip with nearly 100,000 transistors on it. This is some years away from being a commercial reality. As I say, we're at the hundreds rather than the tens of thousands of transistors, but we can make it in the lab. I think this points the way to how IoT will actually enable us to put electronics in, around, on, and underneath all kinds of devices that we've never thought of before. For me, that's an interesting insight into what the world will look like. Being an engineering thing, it's got a roadmap. It's got a roadmap like Moore's Law. If you look on the left, it's where we've been doing our prototype development.
The number of transistors you can fit on a chip is actually doubling in the same way as it does in transistors. The imprint technology is based on all of the technology that's used to make CDs. If you then follow that through and go from CD manufacturing to DVD manufacturing to Blu-ray manufacturing, you start to get down from the 2 micron, 2,000 nanometers that we're at with this today, through to tens of nanometers in the future. I predict that around 2020, I'll be able to show you a chip of the complexity of a Cortex-A5 or that chip that I've just shown you now about this size, the size of a small chip today, which you really could put on, in, and around your clothing and not even notice it's there. I think there is an interesting road to future development.
What has changed in this kind of 25, 35 years of development from the original BBC machine through to the micro:bit? Well, I think what's happened is, apart from Moore's Law driving either reduced costs or increased complexity, we've actually gone from a world that had standalone devices to devices that are actually connected to the cloud and interact with that. That's brought us great opportunities for whole new product ranges and whole new services. What it's also brought for us is the whole threat of people attacking those systems. Once you no longer just have your machine in the cupboard and it's actually out there and interconnected, you're under attack in many different ways from many different people, whether it's malicious, mindless, or really after your money. That, I think for me, is one of the most significant changes that's happened to our industry.
The whole issue of security is becoming to the forefront. Last conference I went to, a slightly different demographic of the audience, the Arm Engineering Conference. Some of us are a little more statistic than others. It's a fairly normal distribution of engineers. We thought, well, we need to find a way of actually socializing them, shall I say, getting them to talk to each other, interact. We commissioned the creation of an app for the conference. That allowed you to vote on panel sessions you've been to make posts. By making posts, you win prizes. Then there's a leaderboard, and there was a nice interesting competition between people to how could you make meaningful posts and get people to like it. If they like you got even more points, you went up the leaderboard.
It was actually a good way of actually creating some social engineering in that community. After the first day, it had gone very successfully. I was sitting at dinner with two of my colleagues from the research group who specialize in security. I turned to them and I said, "How secure is this app?" They kind of perked up and they went, "Well, we don't know because we didn't think you'd approve if we had a look. If you're asking us to have a look, we will." This was 10:00 P.M., and they disappeared.
I met them back for breakfast, and they said, "Well, it took us two and a half hours, and then we got bored." I said, "Well, what did you find out?" I'd like to, in the next few slides, try and explain what they found out, and then I'll say why I think that's relevant. If you have an app running on your phone, and that is the same as an embedded IoT device somewhere built into this building talking to a service somewhere. It's really the same problem, whether it's a phone and an app or, as I'll come back to, a light switch talking to a building management system. The first thing you want to do is to stop people eavesdropping. Those of you who've connected to the local Wi-Fi network here might have done that.
You want to stop people being able to listen to what you're doing. The way you do that is you encrypt the communication between the server and the phone, and you use a certificate to say who you are. The server has a certificate to say it's authenticated, and you check that with a certificate authority. It works. The channel is encrypted. You can't do anything. What the guys did, they said, "Well, we've got this phone and we've downloaded an app to it. Why don't we take our laptop, and on our laptop, we'll run an emulation of the phone." It's Android. You can just go and get an emulator. "We'll download the app and run it on our PC." You're now running the app actually on a PC emulating the phone.
Because I'm emulating all the phone, I can intercept the secured communication and put this eavesdropping device in, and that's called the man-in-the-middle attack. You've probably heard what it is. We're going to try and explain how you now execute a man-in-the-middle attack. You still have the server with a certificate going to the certificate authority. I can't hear what's going on. On the PC, you say, "I'm going to change the certificate authority, make it mine. I'm then going to intercept the certificate and put in my certificate, which my certificate authority will authenticate." All of a sudden, I can now see what the communication is between the server and the app running on the emulator. I can see all the traffic. How did that happen? It's quite simple to fix. It's called certificate pinning.
The problem is the app trusted the phone. It said, "I have a certificate. Phone, please authenticate that this certificate is correct." What they should have done is say, "Please check that this certificate is correct with this particular certificate authority that I trust." You'd have to choose one. You'd have to make sure it was available in all the regions that your app was being deployed. You'd have to do a little bit more work. If you don't do that little bit more work, you're susceptible to a man-in-the-middle attack. Now my guys can see what's going on. They can look at the messages. They can't do anything. The next thing is, well, I actually want to start trying to create messages and do things. I need to log in. The app developers got this right.
Even though we could crack some of the passwords, we couldn't actually penetrate this system because the whole thing is protected by a shared key. There's a secret that the server knows and the app knows, and we don't know what that shared app key is. We can't actually do anything about it. Scratch their heads. This is after about 20 minutes. They went back and looked at, what is the application we've downloaded? We got it from the Android App Store. The application they downloaded contained two files. One of them is the program for the app, and the other one was a file with a magic key in it, which was the shared app key. Lo and behold, the shared app key in this separate file was in plain text.
Shared app key, you're off to the races because now they could actually log in and start doing things. The question is, why was the shared app key in plain text in the file? Well, secure provisioning is hard. You're building a company. You're dealing with 10s of thousands of clients. Each client you've got to give a separate key to. If we're in the same building, in different conferences, my app won't talk to your app. They need to have different shared keys for each one. You need those keys to be generated by your support staff. You haven't really put in a secure infrastructure to manage these keys, distribute them, upload them onto the App Store. It's just easier to make the key a simple thing that you actually use in the company. Secure provisioning is hard.
We now cracked the app, but you still got to do bit about what does it take to be a hacker. You've worked out that you want to get to the top of the leaderboard, the way you get to the leaderboard is to make a post. Well, now we can make posts and win points and repeat, but it's a little unsophisticated because I'm just going to spam all these posts against myself, and people are going to look at that and go, "Well, it's really rather crude. All you're doing is publishing the same rubbish, rubbish and winning points." They thought, "I wonder what? Do you think you can delete a message? I've never seen a delete message go across the channel because in the app you can't delete a message, but you probably can.
I guess that that's just going to be called delete.message. Let's try delete.message and see if it works." Lo and behold, you can delete a message. You can post, win points, delete the message, repeat, and nobody knows what you're doing. Now, being hackers, they're also a little crude because they and some are people desperately posting useful material to get to the top of the leaderboard, and they just, you know, game. At the last minute, guys that did it pushed themselves to the top, took me with it. I didn't even register onto the app, and suddenly I'm second on the leaderboard. Why am I telling you this? Because this was a reputable company deployed in 10s of thousands of clients running conferences every day. They got most of it right. Okay? Pretty good security. They knew what they were doing.
As in many systems, they had just been a little careless in a few places because it was hard. The reason I am telling you this is because we need to make this work for sub-GBP 1 devices built by people who have. As I said, that phone running an app is really the same as the light switch in this building controlling this room connected to a building management system. If you cannot make those systems scale and make them trustworthy, we will not get the deployment. If I own 1 million light switches, each one of them was deployed at GBP 1, not really high value, I am not going to put that much effort into the security of it. Where is my return on that GBP 1 microcontroller into the GBP 10 light switch?
If I control 1 million light switches in the City of London, I decide to turn them all on, I would probably crash the power supply. You can do serious damage from a GBP 1 microcontroller if it is deployed at scale, I find a way to take that over. We need to solve that problem. The good news is that solves the connectivity, solves that provisioning in a secure way and allows you to manage those devices. I would like to give one example of a partnership that we recently announced with IBM and explain why are they engaging with us. IBM has a massive business at scale in deploying big data applications, whether it be to manage your city or your business.
It is a large, complex business, they need to attach to a large number of devices in those smart city deployments. What our platform offers them is the ability to integrate the server-side software we have developed into their cloud offering, one, then that can communicate with all of the variety of clients, the 10s of thousands of different SoCs built into hundreds of, or millions of different products on the left-hand side. They do not need to worry about the diversity of these IoT clients on the left-hand side, because they are all abstracted in one clean way by the software that we have installed in their servers, in their cloud, on the right-hand side. We offer them the ability to get access to diversity of IoT clients, while then having to only do a single integration into their cloud.
That is one of the main value propositions that we have. We are dealing with the complexity and the diversity of the Arm community that we understand. We are taking that embedded world and packaging it in a way that the web and cloud developers understand and offering a solution. Of course, the way we do that is through our traditional semiconductor partners who actually have to build all of those chips. They embed our client software like that. We work with the device ecosystem players who actually turn the chip into the light switch. We work with the cloud providers who can then actually offer a service that is then deployed to either OEMs, cities, or individual users. The mbed platform started some years ago.
We've developed a good community from that. In the latest incarnation with all of our web technology, we've actually now got significant engagements with most of the semiconductor guys who are developing all of those embedded controllers, good partnerships on both the cloud and of course, the ecosystem side, and real deployments by OEM. I'd like to just look at a couple of end deployments on the professional end of the spectrum. As I said at the beginning, I think there's lots of IoT development is going to be done on the kind of maker, Kickstarter end of the project, but a lot of the money will actually come from the commercial, industrial, professional end deployments. It's something that will apply to both ends of the market. Just going to take a couple of the professional deployments that we're engaged with.
The first one is GE Lighting. GE Lighting are deploying smart lighting, their LightGrid solution into 20 cities in America and Latin America. The picture on the right is actually of a deployment in San Diego. The way this works is Arm technology in the client is actually in each of the light bulbs in each of the lampposts. They have local area networking that allows the lampposts to form an ad hoc network. The lamppost finds the next nearest lamppost, which then talks to the next lamppost, which talks to the next one. They form a grid that has redundancy, so one light bulb fails, you can probably manage to talk to the next light bulb beyond that. Then just a few of those lampposts are actually connected to the wide network.
Their cloud infrastructure talks to one or two of those lampposts connected to the wide infrastructure, and then the rest forms an ad hoc network to allow you to manage the lighting in the smart city, to make it safer, to save energy, and do all of those good things that you can do with smart lighting. It's a significant market segment. By 2025, there'll be over 90 cities that have seriously deployed smart city and smart lighting infrastructure. Out of the predicted 5.3 billion devices that are going to be deployed there, 4.5 billion of them have yet to be shipped and will be shipped in the next five years. Significant market. Here, perhaps, is a slightly different one. SK Telecom partnered with us to do smart fish farms. It's a big market, okay?
It actually turns out to be a big market, and it's a similar problem. You have a fish farm in a fairly rural area dotted around the coast or tidal lake way, and you have to monitor water quality, pH, temperature. You need lots of little sensors deployed throughout the fish farm. You need them to form an ad hoc network and talk to each other. You then connect to SK Telecom's backhaul over 3G to their cloud, and they've then provisioned in their cloud, mbed Device Server, to allow a third party to then write a smart fish farm app. You then access, as a fish farmer, that app that allows you to monitor and control multiple farms within your region. Again, precision farming is going to be a significant business as global population grows to 10 billion.
Food production goes up by about 70%, there's over a $5 billion market in actually deploying infrastructure into those markets. IoT tends to be a fairly fragmented thing. People talk about all sorts of kind of wacky things, including a lot of the kind of personal wearable devices. That is a big part of the market, but there's also a real kind of commercial industrial end of the market as well. If you just look at the products we have, mbed Device Server deployed in the cloud, mbed Client software deployed on the chip, the chip itself, which is what Arm traditionally is engaged in, an ecosystem that goes with it. To explain how we make our money, well, we are licensing the software in mbed Device Server in a recurring revenue per use, per transaction, per device fashion.
We are providing the device software for free. It's free, it's open source, it's about engaging in a design community. As you say, that's where you go from the thousands to the tens of thousands of different developers. You need to make that easy and open and accessible. It's free software to get wide deployment. There is the classic hardware IP business we have today with hardware licensing and royalty, and the ecosystem it's built on, which has optional membership fees. Another way to look at this is why do those partners engage with us? Well, the chip guys engage with us for our hardware IP to build chips, the software IP to make it easy for them to build those devices and get access to developers. They can go to developers and say, "Here's my chip. It's easy to program.
It's ready to go." The cloud partners want access to devices. In that IBM example, it's about how a single integration with mbed Device Server allows them access to this very wide range of devices. It allows them to have trustable data. You need to know that the data you're getting from your fish farm is actually accurate and reliable and not spoofed by somebody else. It allows you to manage those devices. The OEM partner who's buying the ultimate service has the trusted platform. If you want to become a deeper partner with us in the ecosystem by paying membership fees, you get early access. You get access to source code for binary components. You can actually deploy products quickly. Let's just quickly look at one example. Here's a little tiny Bluetooth beacon with maybe a 10-year battery life.
What does it take to build that? Well, you have to build the SoC, the chip at the bottom. That's our traditional business. You're going to be using a Cortex-M processor, of course, license fees and royalties. You're going to be building it using our physical IP, generating incremental license fees and royalties. We've recently launched a sub-one volt radio to actually make it easy for people to build those local area networking with things like Bluetooth, where we made recently two acquisitions with Sunrise and Wicentric. That brings us, again, incremental licensing and royalty. Recently, we made an acquisition of Sansa, and part of their technology is actually a technology that goes into those chips to build trust into the very bottom of the stack from the hardware up. That's the hardware. On top of that, you have to put your real-time operating system.
There are hundreds, probably thousands of them out there in the world, very fragmented. We looked at that and said, industry needs consolidation. One OS is better than a thousand for developers. It needs to be architected as secure from day one. Not a question of adding security to the solution. It's actually architecting it from the ground up. There is a low layer called Microvisor that actually provides control for how you manage secure assets, how you stop some of the mistakes that happened with the example I gave at the beginning by architecting it from the bottom up. In some senses, it's prescriptive, but if you do it this way, it will be secure. Challenge with a lot of the web development is there's 101 ways of doing a bit, two of which are secure, the rest which aren't.
If you don't really know what you're doing, you aren't going to pick the right combination. Make it easy for those developers. Build on top of that, a classic set of device drivers, an Mbed OS operating system, and all the application code that goes on top of that. Again, the acquisitions we've done from Sansa and Offspark have added critical security components into that protocol stack to make this something that can deliver a trusted platform. I think if you do that, we can actually achieve the kind of extraordinary numbers that you see being published for IoT that are back to those CAGRs of the mobile industry. Each one of those segments isn't the scale necessarily of the mobile industry. Each one of them is growing significantly, and each one of them becomes in itself a really interesting business for our customer participation.
Full circle, BBC micro:bit. You'll see that outside if you want to play with one. Why are we involved in it? Two reasons. The kind of sordid commercial side, it's built on Arm Mbed technology . We're getting Arm Mbed technology out there to the millions of new developers that we hope to create because we need more software-savvy schoolchildren to become web developers, mobile developers. When they grow up, it's probably they're going to be IoT developers rather than mobile phone developers. We want to make sure that for the long term, we're seeding that community. We're getting our technology deployed at scale really quickly, and I think it's a really exciting program.
In summary, if you look at what we're trying to do for IoT, from a hardware perspective, we are bringing additional components to actually generate incremental licenses and royalties in the hardware side with Cortex-M physical IP and the 4G radio. On the server side, we're actually fueling some of the demand for the servers and networking infrastructure that Pete talked about. I mean, a lot of the IoT transactions that are going to generate some of that growth in servers are ideally suited to Arm low power solutions built out at scale. That is the classic market that will drive Arm server adoption. We have the free software for the Mbed Client that actually drives adoption of the solution, and then licensed software into the cloud and OEM partners, which is where the revenue stream comes from.
I think put that together, we actually have an exciting platform that allows us to move Arm forward in generating incremental revenue streams in the future.
All right. Thanks, Mike. Great to be here finally. It's been a little while coming, and I'm very pleased to be here. One of the things that first attracted me to Arm was the opportunities that it has, both in the smartphone market, but also in other products and services. It's particularly pleasing for me to be here as part of the team presenting that sort of next leg of the growth story. Now, I'm a finance man. I've spent 30 years in commercial finance roles around the world. First and foremost, what growth means to me and technology means to me is money and turnover and returns for shareholders. There's actually a more personal connection for me as well.
Not many people in the room will know, but I actually studied electronic engineering at Cambridge University in the early 80s, probably just a couple of years after Mike Muller. I went in a different direction. As I've gone through my induction to Arm, I have felt that sort of inner geek being awakened in me. As I said, I'm really proud to be a part of the management team on this next leg of the journey. Simon talked you through all of the investment that has put Arm in the very enviable position it is today in smartphones and also the other technologies. Pete and Mike have talked about the future growth opportunities. What I want to do in the next 25 minutes is sort of pull all that together and, as Simon said, talk you through the financial implications.
Before I do that, I know that some or even many of you in the room will be interested in the short term as well as the long term. What I wanted to do was give you just a quick update on current trading. As we stand today, we've had a good proportion of the royalty reports in for Q3, and we've got a fairly good line of sight on the licensing deals that we can close. Off the back of that, what I can say is that we're comfortable with where Q3 consensus revenues is. That's subject to the usual caveat around licensing income. We're comfortable both in USD terms and subject to what happens to the exchange rate in the next three weeks, we're also comfortable in GBP terms. We'll update Q4 in the normal way when we announce the Q3 results.
Looking further forward, I know there are some concerns around China's slowdown, what that means for the end market. Arm is not immune to that end market. I hope what you've heard here today demonstrates that the strategy of increasing our market share, increasing our royalty content ship, and developing new recurring revenue streams puts Arm in a very good position to outperform the general market. Having dealt with the short term, what I'm going to do is I'm going to do this in three sections. I'm going to take a look back at the track record of investment over the last 10 years because it's the successes of that investment that gives us the confidence to continue to invest for the future. I'm then going to describe the impact on the P&L of the initiatives that Mike and Pete talked about.
Then I'm going to finish with taking you through our approach to developing a long-term view of the financials for Arm and how we balance the needs of maintaining a strong, resilient balance sheet with the need to reinvest in the business to give it future growth and to give cash returns to shareholders. Arm's investment strategy is really clear. We'll continue to invest in new, more advanced processes, that enables our partners to produce ever more capable chips. Because we're adding more value, we can charge a higher royalty per chip, as we are doing with version 8. We'll continue to introduce new technologies to create new revenue streams. Some of those revenue streams come from existing partners, where we're giving them the opportunity to integrate more Arm technology in their chips.
You've had the example of Mali, where mobile phone chip manufacturers can integrate a Mali GPU with a CPU in their application processes, then more recently, what Mike talked about, where partners have the opportunity to integrate Cordio radio IP into a microcontroller in an Internet of Things device. We'll continue to invest in the ecosystem to accelerate market share gain and to increase market share penetration. That involves making sure that the software is not only available, but it's optimized for Arm technology. What you've heard here this morning is simply a continuation of that strategy that worked so well in the past. Let's look back. What I see on this chart is a fantastic growth story. This is a company that's grown its turnover from GBP 153 million in 2004 to nearly GBP 800 million last year.
At the same time, it's expanding operating margins from 30%-50%, it's done that through a continual commitment to invest in technology to create future revenue streams. Let me give you an example. Arm started work on version 8 in 2008. Very opportune time, the time of the financial crisis, when the end markets were in turmoil, licensing and royalty revenue was flatlining at best. Arm chose to continue to invest at that time. OpEx went up as a result. There was margin compression. Look what happened afterwards. In 2009, the first architecture license was signed. In 2011, the first processor license was signed. Thereafter, you had a 30% compound growth in license revenue over the next four years. Now, today, we're starting to see the royalty benefit from that investment.
As you heard Simon say, by the end of this year, half of all the smartphones shipped in Q4 of this year will be based on Arm's Armv8 technology. Moving away from GPUs. Sorry, CPU. Arm's also created new revenue streams from existing partners. Here, as an example, we've got the Mali Graphics. They typically start with an acquisition. In the case of Mali, it was in 2006 with the acquisition of Falanx. At the time, the engineering team in that company was 30. Today, there are more than 500 engineers working on Mali Graphics. It's a great example of how Arm has scaled things internally. There's been a couple of bolt-on acquisitions to give video and display technologies as well.
That continued investment over many years and using the Arm ecosystem to leverage that investment means that today, Mali graphics processors are the number one shipped processors in smartphones, in tablets, and in digital TVs. We can also create new revenue streams from new partners. The investment in physical IP is a good example of this. As Pete mentioned, and Simon as well, it's really difficult to build chips based on advanced multi-core processors using advanced manufacturing techniques. Arm's investment in physical IP and providing that physical IP alongside the CPU IP means that we've lowered the barriers to entry for people who want to build Arm-based CPUs. What that's meant is that they've been able to, by optimizing the physical design, they've been able to produce very low power chips with high compute.
That's given us clearly a boost for our licensed processor revenues because we're introducing more companies into the ecosystem. It's also given us a standalone revenue stream as the leading supplier of physical IP in the market. Again, in the previous three presentations, you heard about the ecosystem. Here are some examples of where Arm has led the industry in creating open-source software, which is optimized for Arm technology and helps drive adoption, which helps drive market share gains and helps drive increased market share. That's a look back at the past. Now let's look at where we're focusing the investment today and how we're going to develop that going forward. Smartphones are still important. Even with the handset growth slowing into single digits, we expect our royalties from smartphones to more than double by 2020. We'll continue to invest in advanced processes.
We'll continue to provide the graphics, the video, the display technology alongside that. The goal here is very simple. It's to maintain our high market share and to increase the IP content and the value that we're giving so that we can increase the royalty per chip . We're increasingly confident in network and servers. We've got very competitive hardware. We've got a roadmap to improve that and our pace on competitors. Here, the object is to invest further in the ecosystem to expand and optimize the software available. As you heard Pete mention, the goal here is to accelerate and increase our share gain. We've raised our 2020 targets on market share to 45% in networking and around 25% in servers. What that means in real money is that that's $3 billion of additional silicon by 2020, which will be Arm royalty bearing.
Finally, you'll have seen from Mike's presentation, we've got many opportunities in Internet of Things. It's a market that's ideally suited for Arm low power technology. In this market, we want to maximize royalty. We'll do it in three ways. We want to grow the overall market. We want to maximize our share within that market. As Mike described, one of the best ways to do that is by creating trusted IoT devices and hence our investment in security. There's also an opportunity for us to create brand new recurring revenue streams from the likes of Mbed Device Server for enterprises who want to manage large networks securely. Enough of the words. Let's move on to the numbers, because I'm sure some of you are itching to hear about those.
Firstly, I just want to say, putting 5-year numbers out is as much an art as a science. What we wanted to do was give you a flavor of our ambition, give you something to base your models on. The numbers I'm going to take you through on the next few slides are the incremental cost and the incremental revenue from the stuff that you've heard Mike and Pete talk about today. What you've heard about as Simon talked about should already be there. As you will have read in the RNS, by 2020, we expect these investments to be delivering GBP 200 million of additional revenue. Looking at how that builds, in 2016, you'll get the full annualized revenue from the recent acquisitions of Sunrise Micro and of Sansa.
You'll also get increased numbers of licensing wins in servers and networks because of the investment we're making in the software. As we move into the middle years, you'll start to see incremental royalties in networking and servers from the OEM wins that, again, the software will accelerate in the early years, and you'll start to see some of the revenue that Mike talked about building up in Internet of Things. By 2020, you'll have all of the revenue streams contributing. Because in large part, these are increased royalties and licensing from an increased share of a known market or new recurring revenue stream from IoT, we expect these revenues to continue to grow beyond 2020. If we turn to the cost. First thing to say is everything you've heard here today is in our Q3 OPEX guidance of GBP 106 million to GBP 108 million.
The investment in CPUs, in graphics, in physical IP, they're all business as usual. All of the investment in mobile is already in the guidance. We'll formally guide Q4 costs when we announce Q3, but you can expect an additional GBP 3 million to GBP 5 million over and above the normal year-on-year run rate there. Regarding the investment going forward, most of that GBP 40 million is people related. It's more engineers with a few business development guys to help organize the ecosystem. We'll be hiring throughout 2016, so you'll see the cost ramp up. In 2017, you'll see the full annualized cost because by then we anticipate we'll have most of the extra resource we need. From 2017 through to 2020, you can expect those costs to grow much more modestly. As with the existing business, there is some cost now to generate revenues later.
You will see a small EPS dilution in 2016. It's more than offset by 2017. Because the revenues are growing much more rapidly than the cost from then on, you'll see profits rising to 2020 and beyond. What I've shown here are three charts of high-level numbers. I hope from the previous presentations you realize that an awful lot of detailed planning has gone into this. For every initiative that contributes to these high-level numbers, we've got a detailed plan, we've got milestones, we've got targets, and we'll measure ourselves against those targets. We'll continue to invest in a very disciplined way. I want to conclude by taking you through the approach to the five-year planning process.
All businesses need a view of the longer term, and particularly Arm, where the investment today is going to create licensing revenue one to three years afterwards, and then royalties one to three years after that. The planning process has been a key part of Arm's business cycle for many years, and there's an experienced team running it. The process we've got in place for this year is in flight. It will finish at the end of the calendar year. That's when the board will be setting the 2016 budget and the revenue target. If I take you through how we build it up, it starts with the addressable market. You saw on the previous slide, we're very focused on where is the money, where should we be going, and what technologies do we need to address those markets.
A key driver of Arm's success over time is a very deep and collaborative understanding of its partners' product roadmap. That understanding has been built up over many years through a network of account managers around the globe. From those roadmaps, we know what technology the partners need and what they're likely to take from Arm. We know the licenses we've got signed, so we know the royalty rate. We need to take a view on market share shifts, what the size of the end market is, what the ASPs are. The team have been doing this for quite a number of years, so we have a reasonable degree of confidence in the royalty plan that's built up. Definitely the trajectory. Clearly, there are always ups and downs from year to year, but we can be reasonably confident in the trajectory and the royalties.
In the same way we can build up the licensing plan. Again, we know the technologies. Again, we need to sense check and bottom up. Again, the pace of consolidation in the business and the possibility of new entrants. Another really important outcome of this planning process is the ability to identify technologies that our partners will need in the future to meet their product roadmap. It gives us the opportunity to identify acquisitions to grow inorganically, to create new revenue streams, and by going into new technology areas. Once we know what the roadmap is, we know what technology we need to deliver, Pete can work out how many engineers he needs to deliver that technology. Again, there's some marketing and business development heads there as well.
Of course, we need to provide the infrastructure for those engineers to do their job effectively. As with any other business, there are other operating costs associated with running the business. I see management's job, and the CFO, myself in particular, that the job is to control the other costs, make sure that the engineering investment is done efficiently, and it's focused on projects with a reasonable chance of success and a reasonable return. That we can continue to deliver sustainable and growing returns to shareholders. We've got the revenue, we've got the cost, we've got an idea around the opportunities for inorganic growth, and that gives us a view of the cash requirements of the business by 2020. What's our approach to that cash? The first use of cash is to maintain a strong balance sheet.
Our partners need to have confidence that Arm can withstand pretty much anything the world can throw at it. Arm is part of the R&D of some of the world's largest companies. Those companies need to know that Arm is going to be around, deliver on their promises, even in the worst of circumstances. That strong balance sheet also gives us the flexibility and the firepower to take opportunities for inorganic growth as they arise. The second use of the cash is to reinvest in the business to create new growing revenue streams. In the last nine quarters, Arm have made nine acquisitions, all bolt-ons. As part of the planning process, we can see a pipeline of other bolt-on acquisitions, including some of the same sizes you've seen recently and a few larger transactions for future consideration.
Over and above the reinvestment, we see the opportunity to continue to return cash to shareholders. We've got a dividend policy at the moment that delivered a 25% CAGR in dividend per share. I know there's been some discussions around, is there a possibility of future cash returns. Hopefully, you'll see from the process that I've outlined that we have a clear plan to understand the cash requirements. That plan is in flight. Once it's concluded at the end of this calendar year, the board can decide whether it's appropriate to change that. If I were to summarize the last two hours, Arm has a proven track record of investment guided by a clear investment strategy. We will continue to invest in mobile to maintain a high market share and to increase our royalty percentage by adding more value to our partners.
We're increasingly confident in networking and servers, we'll invest in the ecosystem to accelerate and increase our market share gains. There's a great opportunity in IoT, which is a market that's ideally suited to our low power tech. We'll invest to grow the market, we'll invest to maximize our share within it, and we'll invest to create new recurring revenue streams. By 2017, the investments that we've talked about today will be delivering sustainable returns to shareholders that will grow from 2017-2020 and beyond.
On that note, I'll hand back to Simon to chair the Q&A.
Just bear with us a second. We're just going to grab some chairs. Our team will be roaming the room with a microphone, and we'll get going. Okay. As usual in these Q&A, if I could just ask that people ask one question, we'll go around the room, and then we'll reset. Let's start at the front. Thanks.
Sandeep Deshpande from JPMorgan, guys. Just actually a quick question on licensing. You had a stellar run in licensing for the last five years. You've guided now 5%-10%. Is there any change in the licensing environment, which is why you grew 30% CAGR in terms of licensing over the last five years, which was clearly in terms of broadening your market. Are there new opportunities to broaden your market here, or it is now a question of maintaining existing markets and then harvesting from those markets in terms of royalty growth?
We've really always said that we expect licensing to be on a 5%-10% ish range in the long term. We still believe that's the case. With the introduction of ARMv8, with some of the other changes in the market, that did lead to much higher than that sustained licensing growth over the last few years. That has taken our licensing revenue to a very high level, which we believe we can now maintain the growth of in that 5%-10% range, going forward. The introduction of ARMv8 was a big step from a technology point. We're in the process of developing processors which implement version 8 of the architecture. We've licensed three of those now. We have a roadmap as you saw of more technology. We think that's going to sustain us back to the more typical levels of growth.
I mean, a corollary to that just would be on in terms of growth, because you had this very strong licensing growth over the last, I'd say a certain, I think it was about 20% or 22% royalty growth over that same period of time. Do you expect royalty growth now for the next five years because of that very strong licensing growth to accelerate in an environment of slightly lower licensing growth?
In terms of royalties we've spelt out today, part of the success of ARMv8 has been the establishment of a high royalty rate because of the value that ARMv8, and coupled with Cortex-A core and the success of our. See that there is a scope for royalties continuing to grow. We've spelt out in our outgrow the semi industry by about 10 points. We think that is valid. We think that the ARMv8 license. We are seeing the adoption of ARMv8 in other markets revenues in the future that we could further in the longer term. I think in that mid term.
Good afternoon. Atif Malik from Citigroup. [audio distortion] .
I think no is the short answer to that question. If you think about the introduction of v8, if you think about the introduction of new processors, that timeline of creation of an architecture, creation of the processors, licensing, chip design by a customer, shipping it, all of that remains roughly constant there. What you're seeing here is our approach to accelerating growth. What we're doing is we're at a point where the conventional kind of process of architecture product implementation in our product has been running for some time. From the success in that in these new markets, we're seeing the opportunity to accelerate that growth. To get off this growth curve that looks like that, onto one that maybe looks like that through additional investment.
Same is true, think about the IoT business that Mike has been describing is off the back of processor architecture work that was done a long time ago. I said in my presentation 2004 was when we introduced Cortex-A. When you look at it that way. I think what we're describing here is a different set of circumstance.
Thanks. Morning. It's me, James from New Street. I guess I just wanted to ask on the networks part of all the different areas, there was still quite a lot of dots. I guess you've got to convert all those dots to the other reality, which we're only five years away. Where's this certainty coming that you can do quite a lot over the next five years? Working directly with our OEM. Roadmap. Each of those dots, I can point them. Things behind them, right?
Each of those dots we can point to I guess it is an industry where you do quite often delay in terms of adopting new architecture. What extent have you baked in kind of delay? I think it comes back to the earlier point, right? The architectures are already deployed. We already have them out there. We already have the product. The department already has silicon out there. It's not like we're sitting there waiting for the product to be developed. CapEx spending goes through cycles. That's why in Chris's slides there's what we expect in the near term and where we expect it to end. There wasn't much detail of what might happen in between because cycles come and go.
There probably will be cycles through this period. From the technology development we've done, the licensing, the chips that our licensees are creating, we think this market, and then the engagement Pete points about, I think this market is set, that the greater proportion of the product shift around Arm five year, in fact, five years plus, modulated by the way.
Thank you. Back to that Mali chart . I think historically you said that you were taking market share from Power Architecture and from MIPS as well. As you keep on growing market share, where or who implementatively are you going to take market share from?
No, it's a good point. Currently seeing that there's internal development.
Process-based solutions, and then of course there would. Yeah, that's right. Yeah. Ambitions as well. You're basically saying.
Yeah, if we're going to take 20% of the server market, clearly. We will never tell you that there is no competition for Arm. One of the things, though, about networking is it isn't just a case of replacing chips that are built today on one architecture with chips built on the Arm architecture. What Pete described in his presentation was how the network itself is changing from something that connects a client device to a data center to something which is a much more distributed computing platform. The chips within networking themselves are changing. There is a discontinuity going on. There's a shift to a more software-oriented network. That creates then new devices, which we see as an opportunity as well. Some of this is greenfield, and some of it is an upgrade of delivering more performance.
Do you happen to have discussions with the end customer as well? Seem to be a bit lost, the network guys, I don't know, like Vodafone or whatever, who seem a bit lost with the new architecture. Very exciting, but they don't know where they're going to deploy this.
Well, indeed, and we absolutely spend time talking to people at various different points of the supply chain to get that view. There is no point in us looking at a market that's great, and it turns out that actually the people that build the equipment aren't going to adopt it. By developing those relationships and that dialogue, we use that information to hopefully make sure that we're introducing new technology at the right time. Carry on, Daniel.
Thanks. I'm just trying to understand the split of incremental revenue. Would it be fair to say that the GBP 40 million of revenue, bulk of that would be licensing? I guess that would imply that IP team staggered a bit above the 5%-10% growth. As we go into 2020, the split should be more even between licensing and royalties. Is that the right way to think about it?
You've got two factors at play. One is incremental licensing. The other is the annualization of the revenues from the acquisitions we made this year. All of that will play into next year.
2020?
I put the number of the additional silicon that will be royalty-bearing in networks and servers. I think that's a very good start for calculating where all of that GBP 200 million.
Gareth, from New Street Research. Just a couple on the business model. Firstly, you touched on this, Chris, in your presentation earlier. Royalties, licensing model.
The mbed Device Server is a software product. It is recurring revenue, but it isn't quite the same licensing royalties we have. It is driven on size of deployment, number of transactions, number of devices. You could think of it as a royalty, but it has a recurring element because the same device may create subsequent transactions.
Ability of competitivity. You start with a top-down model. Changes in that top-down. How much flexibility is there? I believe that.
One of the graphs that Chris showed was how revenues have fluctuated as a function of the cycles that we've been through, and we kept our investment going. Now, obviously, in Arm, we don't run factories. Our investment is in people. You can change people if you need to. We've only really had to do that once in a material way in the history of Arm, fortunately. We have, as cycles have come and gone, with a view to the future, continued to invest in those cycles and help build trust among customer base. Hopefully we won't need to make adjustments as we go. That is ultimately believable. We think we have good confidence in the investments that we're making. We take the step.
Jerome, BNP Paribas. I got a question I can't help but give you because I already have it. Concerning the octa-core, we have seen indeed a lot of traction, MediaTek, Huawei. On the other hand, each time your clients are using the custom architecture. Now with your new 820 three. They are either dual-core or quad-core. Is there a point saying that while actually custom architecture will be more efficient than using an octa-core CPU than Arm doing so on? What's the rationale for your client eventually to use multi-core?
To be honest, I don't want to get into that. I don't want to discuss what our partners are doing. There is a trade-off in performance and patient. You can look at different benchmarks, and some configurations work better than others, depending on what you're running and what you care about in your device. I don't think there is a right answer. We're definitely seeing a trend to more cores in many parts of the market, to enable more software to be done parallel. The use cases of mobile continue to evolve. I think what we're going to see is different implementations targeting different markets and-
Fundamentally, the use case for multi-core is power consumption. Is it the power consumption is the main reason why it's using?
Not necessarily. Well, it's fundamentally about scaling performance and power. If you have a system with more cores, you have more performance. Scale it back, either in if you've got an ingenious set of cores or if you're implementing big.LITTLE. There are many degrees of freedom. The great thing about our model is that we don't dictate what the answer is. Lots of people building octa-core big.LITTLE configurations. That's great. There are people building other implementations all based around Arm, and as I said, we aren't dictating what the answer must be. We are enabling innovation from lots of different companies, and if it's better, then fantastic.
Andrew, you obviously tempted your smartphone growth to be 7% CAGR five years forward. You also touched on not being in a Chinese situation. How much of that growth coming from those developing markets? How much of it is deep?
There is a mixture. That data is really the market data, predicted market view. How mobile is going to grow over the next few years. We don't have a fundamentally different view on geographically from that. What we are seeing is a lot of innovation going on to stimulate upgrade, and clearly China is a big market, but there are other markets which are growing. Other countries where there is very relatively low penetration today, but All those devices are highly likely to contain Arm processors, multi-core graphics, video, et cetera, with a broad range of technology to enable different price points in different markets.
It's Lee from Jefferies. Maybe just a question to Mike, if I could just Find it out for it. It looks like with Tango you've got a software license. I just wondered if you could go after more bolts on the potential transform, to look for a bigger software license business going ahead now. Or are you looking to transform that base of something that
I think as you look at the business going forward, it is all about developing the server-side software, the sales that go into that. Personally, data analytics is not somewhere where I'm interested. Our strategy is to enable the cloud providers. If you look at most of the places that they're extracting value, it is around big data management and analytics, and I'm very happy for them to play in that. It's more how we enable to get the data to them rather than to try and then compete with them in terms of doing analytics with that data. I'm interested in the little data and making it available, but not actually playing with the data.
Being in lookout for FMC implementations in Tangos?
Well, as I said, we have an investment in PragmatIC. We've got Propping as a medium term investment to see where that goes. It's an interesting adjunct to the business.
Thanks. Johannes Schaller from Deutsche Bank. When you talk about, say, doubling in mobile world by 2020, that's roughly a 15% unit growth, 7%. Can you help us understand that delta a little bit better? It looks like the A is pretty fully penetrated the end of next year. On the octa-core side, outside of the custom cores that Apple or Qualcomm, it looks like you're probably relatively well penetrated end of next year as well in Mali, outside of Apple and Adreno due to be well penetrated also. How do you keep outperforming? Right, we should be thinking about higher Mali royalty is something that.
Well, if need be, we can have the team work with you on some of the details of the mathematics behind that. We can talk you through the forecast where it's function of more technology per device growing over the next few years. A lot of the licensing has been done. A lot of chips have started to come out. Devices are going to come through to market, that's going to grow over the next couple of years. We're feeling quite resilient.
Might be repriced. If we look at octa-core and we say you have 50, where do you think you can get to by 2020? Do you think same on Mali, really?
I'd be very surprised if it's 100% opportunity. There are choices of technology, as I was saying earlier. What we look to do is to create a platform and then to have people, our licensees, make choices about what technologies they put together. It's a competitive space. There's a lot of innovation going on. Rash for me to say we're going to achieve 100%, but I think over time we get an increased-.
Keeping item. Where do you think you come out at the end of this year in terms of octa-core and everything?
Let me take that under review. Thank you.
Okay. Sorry. I just had a quick follow on the doubling in smartphone royalty. Kai from [audio distortion] , by the way. My question is, how much have you stress tested that for sort of pricing in the It seems to happen in the PC industry once that unit growth slows, say, sub 10%, there is increased competition from the FMC vendors. I think we already saw that. Is that the narrow range or do you think that at current pricing, current mix that's going to remain a tailwind that can drive you onto that sort of multiple?
Yeah. No, we absolutely build different views on pricing into that model as we put it together. We do end up with a range and at each point it's going to look different. One of the dynamics that's going on in smartphone is there are merchant semiconductor companies supplying the market. There are OEMs themselves building chips. That changes over time as well. We can't just look at the merchant pricing to get a complete view of it is changing. Insight into that because we work with everybody in the space, but it's something that. To your question, yes, we stress test that with a few of our.
Thanks. This is Matt Rinsey from Canaccord. Steve, two questions, one clarification and one question. TAM numbers that you're forecasting share for and not unit numbers? If that's so, what does that mean about your unit share?
Yes, those are TAM numbers in terms of billions of GBP. For the second portion was?
I guess what does that then imply about your assuming an ASP difference.
Yeah, the % is a unit number. Sorry, the %, when we say 25% of servers, we mean 25.
Okay. I would flip the question then, what does that imply about the percentage share of the silicon TAM in dollar terms, assuming an ASP differential by the incumbent?
It is going to be less than that, right? We assume that going into this market. There is an incumbent in the server market. Their pricing is relatively high.
Follow up to that. Obviously, you made some great points, I think, within it, Heather. A key competitor there has made some pretty interesting their 3D XPoint memory, except that your official take on that is that a defensive move on their part. How are your partners reacting to those kinds of things?
I guess the first thing I'd say is one of our partners in the network, one of the companies I showed on there in terms of deploying active general point is that when you look at the data center, a one-size-fits-all approach is not what people want for the future. We have a number of our licensees who are looking at building SoCs, taking the methodology and the approach that they've used in lots of different other markets and applying that same methodology into the data center. If you look at Cavium's chip, for example, high core count cluster of Arm CPUs with various different accelerators for different end markets. It's that combination of general purpose processing and specific acceleration that produces an incredibly efficient end solution. Now, there's lots of ways of doing that. A combination of a standard processor and an FPGA.
People have done that in many different ways conventionally. That's one way of doing it. Building a complete SoC is another way of doing it. Building a standalone chip is another way of doing it. Lots of different ways of approaching the market. I think the net of this is that combination of dedicated hardware with high performance processors gives you a more efficient way of addressing. Not for every application, but for many applications.
On delivering. A question on your opportunity in networking. I suppose who your two incumbents there are, MIPS and PowerPC. MIPS is one of the largest vendors today. If you displace MIPS completely and have 670, it doesn't really move the dial. Is the opportunity here, is most of the market still unsupplied to?
I don't think so. We are displacing MIPS, we are displacing PowerPC, and people are putting more and more. It's not just a crude, we're going after these particular competitors. People are adopting for a purpose. The key point here really is about the way in which the network is changing. We aren't thinking about what happened in the past, who had that, what their financial model was. Frankly, I don't care about that. What we are thinking. I believe that we have the right product lineup to address that. We have the right partners that we're working with on both the semiconductor side and the software to address this new market. As we model that, we see a great opportunity for a new revenue stream.
Moment you think is the right
Dedicated networking. Yes, that's absolutely. They are fairly dominant. Intelligence means the growth of this industry. A lot of functions, devices will become programmable. There's a lot of devices today that don't have a CPU, but will.
Phil Soltani from Credit Suisse. Simon, IoT, obviously you're bit on IoT. What do you think is going to be the biggest threat in terms of competitive point of view? Is it Wi-Fi, microcontroller? Your market share is still around 25%. I presume, given the progress that you've made, why couldn't the market share number be as high as what we are seeing with Mike?
The 25% you're talking about is just the way we've characterized the market so far is a microcontroller. Think about microcontroller as pretty much the 75% very old 8-bit microcontrollers that have been around forever. Code has been written for them. They're shipping in time. What the Arm Cortex-M has done is bring 32-bit processing performance into microcontrollers. Sophisticated software development tools. It allows for code reuse. Because so many of the world's microcontroller companies have adopted Cortex-M, it means there's a lot of software engineering talent suddenly available. If you're building a product, you've got choice of supplier, code reuse, it's all good. That 25% is of that microcontroller market. Those 8-bit products will be shipping forever and ever and ever.
The chance of that going to 95% probably in our lifetime is quite small because there's just a zillion in products that don't need to change, and you don't need to change. IoT, though, is different. IoT, I've showed some numbers, and you can find reports on enormous numbers. Bit hard to say exactly what it's going to be other than it's a big market. We think we're really well placed to have the right silicon devices for that market. With the introduction of things like audio, we can add some of the connectivity to build the mesh networks that Mike was talking about. This software platform that we're building on top, we believe is going to simplify connecting a device to a service. We're not really going to worry too much about the service that's run. That's not our expertise.
We want to take the horizontal technologies that cut across markets. We're going to have competition in that space. A lot of the code that runs on IoT today doesn't exist because let's face it, IoT is in very early stages of infancy, and there are other companies looking at this as a big opportunity. I think the strength of our technology, our business model, our ecosystem will speak.
Follow up on a question earlier. I think you're talking about 7% unit CAGR growth for smartphone tablets, and you're saying that your services SAM would actually increase by about 9%. Think that the ASP is probably going to go up. Trying to understand why would the ASP for application processor for smartphones actually go up, given that we are seeing a big shift towards low end? Or does it actually assume that you're probably going to take some share in-
The reason that we recategorized that market to include it all is because we think there's an opportunity there, and over time, then on the assumption that one of our licensees wants to go build a chip for that market, then yes, we can take some share there and yes, that would obviously help. Probably going to have to make this the last. He had his hand up first.
Andrew from Morgan Stanley. Thinking about mobile from a bigger picture point of view. Comments today about doubling revenue. Sounds as though you're kind of gently nudging up what doesn't get seen. Sounds like you might be targeting. Can we see kind of specific-
We talk about heterogeneous processing. We're thinking about different CPUs, all ARMv8 in this context, but also with the accelerators for specific applications. One way of looking at a video processor is it's a computing engine for a particular task. You can do video decode on a CPU, but your battery won't last very long. It's a genius processor specifically for that. In terms of are we changing our view on royalty process for volume? No, we're not. It would be great if we're in a position where we can dial up different royalties on a year to year basis based on growth, but we can't. We are in contractual. We have, as I said earlier, established high royalty rates with ARMv8. We have potential for royalty from graphic video for display, things like audio over time. Stacking our growth. Well, thank you all for joining today.
We've covered a lot of material, and we'll see you on the road, and we'll see you back for Q3 results. Thanks a lot.