Well, as chairman of Arm Holdings, it's my pleasure to welcome you to our half year results for 2015. Just before I hand over to Simon and Ian, let me just restate our recent board changes. As many of you will know, last year, Tim Score announced his desire to retire from full-time work, although he seems to be filling himself up with non-exec work as we speak. Indeed, left recently. Having previously announced that Chris Kennedy would be joining and taking over as our new CFO, we were able to further announce today that he'll be starting in a few weeks. In fact, on the 1st of September will be his formal start date.
Therefore, on this occasion, joining Simon to present these results will be Ian Thornton, who I think all, if not definitely most of you, will know very well and needs no introduction from me. Thanks again for coming, and I'll hand over to Simon.
Thanks, Stuart. Morning, everyone. Thank you for joining us for our first half results in Q2 2015. Before we start, we just dwell on the usual customary statements. I assume you're all deeply familiar with this. Anyone who isn't, we can give you a bit of a teach-in later, but I'm assuming we can take that as read. Let me talk about the business. Q2 has been a very successful quarter for Arm. We've delivered a strong business performance and made great progress towards our strategic goals. We signed a record number of licenses in the quarter, 54 licenses with 31 companies. That is a record which, in a Q2, is in itself quite unusual. Our licensees shipped 3.4 billion chips into a wide range of end products, and we saw strong growth in all our key sectors.
We also launched some major initiatives for us, thinking about how technology is used for the future, looking for ways to stimulate innovation and stimulate people's interest in the technology sector. We launched programs with UNICEF and with the BBC, and I'll come back to those later. We've increased our investment in the business. We've been hiring strongly over the last few years because we believe there's a great opportunity ahead of us. This quarter really demonstrates the benefit of that investment. The strong licensing, the uptake of our latest technology really does reinforce that investment that we've made in the business to date, and we're continuing to invest, hiring about 127 people during Q2 and making two acquisitions as well. The net of all of this was 15% top-line growth in revenue terms, in dollar terms, and 34% growth in EPS.
Let's now look at some of the licensing. It's normal in any quarter for about 25%-30% of our licensing to be with companies looking to develop products for mobile devices. That norm was continued in Q2 as well. What was particularly key was that during Q2, we licensed all our next-generation technology. Our R&D teams back at Arm are working on the next generation, higher performance processors optimized for power efficiency. During Q2, we engaged with lead partners who want to take that technology before we've even finished building it, work with us in a lead partner capacity. We work very closely to deliver the technology that they want to make their product successful. Again, this bears the fruits of our investment in R&D, the creation of new products.
We've teased a little bit with code names in the press release, Artemis, Greeve, Kite. These are our product code names for our next generation, and we have more to come later on as well. That licensing, the mixture of new products, helped our backlog grow. We also licensed mature products as well. That leads to turns revenue. The combination of all of that led to one of our strongest ever quarters for licensing revenue. Although we haven't given details of the new processor products, what we did launch during the quarter was an IoT subsystem and a new radio product called Cordio, and I'll come back to that a bit later on. A lot of the licensing that we do in any one quarter is with repeat customers. It's with people we've developed long-term business relationships with.
We work closely with them to understand their roadmap needs. We feed that into our product development to create our next-generation devices. What was particularly interesting in Q2 was there are also a number of new customers. In any one quarter, certainly of late, we do license companies who are coming to Arm for the first time, and there were 12 during Q2. Over the last 12 months, in fact, there's been 40 customers who have licensed Arm technology for the first time. Some of them are new companies, startups, a lot of those focusing in the IoT space. Some of them are companies who've been building chips for many years, but only for the first time need to put a processor alongside their analog technology. There are also OEM customers who are coming to Arm and licensing it from us directly.
Companies who want to understand more about the technology that's going into their product to be more involved in the specification and the development of the chips that are going into their products. That is leading them to come and engage directly with us. The licensing, as I said, the number was strong. 12 new customers now takes the total number of licensees to over 400 and nearly 1,300 licensees that we've signed in total in the lifetime of the company. All of this is covering a wide range of technology. It isn't just about mobile. Mobile, very important sector for us, but we're seeing the adoption of our technology in many different areas of digital electronics. As I said, mobile, very important to us.
We're always looking for ways in which we can improve our products to help our customers and our customer's customer deliver an even better mobile experience to consumers. It's actually quite incredible to look back over the last five years and think about the journey that smart mobile devices have taken. Now we take for granted every day now being able to access the internet, get all the data that you want, whenever you want it, on this thing that sits in your pocket the whole time. Let's face it, if you left it at home when you walked out of the house in the morning, you'd go back to get it. It's incredible just how ubiquitous that has become in developed countries.
If you just go back five years and think about the device you were carrying then, I personally was using a BlackBerry back then, fairly basic BlackBerry. It was great for email, but I couldn't stream video to it. The mapping solution in it was very primitive. Wind forward to today, we've got very sophisticated, true mobile computers, high-resolution screens, great processors, high-speed data connectivity. Seems like a very different world from only five years ago. This really is a mobile computer that we are carrying around with us today. In just that five years, we've seen an 80 times improvement in the processing capability that we have at our fingertips. Really quite phenomenal. Against that, batteries have improved barely in that time. They improve very slowly.
Transistors get better, through the combination of design, leveraging technology, that's how we've been able to deliver an 80 times improvement in compute performance. The computers that we have today, these devices we carry around, increasingly, they're being powered by Arm v8-A processors. Processors which have introduced a 64-bit processing capability along with many other features. Right now, we're seeing in the devices that our licensees are providing into the mobile market, multi-core becoming the norm. Many devices based around big.LITTLE technology. That's something that we introduced back in 2011. Let's face it, at the time, there was a fair amount of skepticism about whether this was a good idea or not, but I think we've proven that, and we're seeing big.LITTLE devices all across the performance range. We think that will pervade completely across the performance range in time.
Multi-core is the norm, big.LITTLE, and version 8-A of the Arm architecture is being adopted. A price point supporting end devices in the tens of dollars all the way up to the many hundreds of dollars. Based on the adoption that we've seen, we stand behind this prediction that around 50% of smartphones being shipped at the end of the year will be based on Arm v8-A. Of course, that's good for the long-term trends in our royalty. Alongside that, we see continued uptake of our graphics technology. Again, based on what's happened in the year so far, we stand behind our earlier prediction of between 600 million and 700 million units of Arm-based GPUs. It's interesting to note that about 30% of smartphone SoCs are also based around Arm's physical IP.
We've done a lot of work on our processor optimizations to enable our customers to get the highest performance with the lowest power in the shortest space of time. We're seeing that come through now in some quite significant volumes. What's coming next? I talk to many people who can't really imagine their mobile device getting better. People that think it's going to plateau just like the PC industry did. Well, I don't think that's the case at all. I think ahead of us, we'll see more and more higher and higher data rates into your mobile device. You'll see shorter latencies, and that's going to enable new use cases. Really, the great thing about smartphones are that they are open platforms.
When you put all this hardware together, when you provide these capabilities, developers can then start to exploit them and think up new use cases for how to use this supercomputer that you're carrying around with you. A few years ago, you wouldn't have thought that you might be using this device to unlock your house, control your central heating, order up a pizza, call a taxi. These are use cases that people weren't thinking about when I was walking around with a BlackBerry five years ago. Now, again, it's commonplace. I think when we unleash all of these new hardware capabilities to the developer community, we'll see more and more interesting ways in which mobile devices will continue to change our lives.
Bigger screens, 4K, better cameras, more sensors, enabling your device to interact intelligently with your surroundings and give you the information that you need before you thought you needed it. I think we have a very exciting time ahead of us in the evolution of mobile. It's one, back to my earlier point, that doesn't just rely on Moore's Law. It doesn't just rely on transistors getting better every two or two and a half years. It's all about design. It's about integrating different technologies, optimizing for the use case. That's what has delivered a great user experience so far and will do into the future. Mobile, very important to us. We remain focused on that. We are very pleased with our success in other markets. Embedded intelligence is one that you hear us talk about a lot.
We talk about the Cortex-M series of processors, which really have become, we believe, the most widely adopted architecture in 32-bit microcontrollers. We have about 213 licensees of Cortex-M now. Many companies developing products. Thousands of different product variants addressing different sectors of what is a very diverse end market. In Q2, there were about 1.4 billion microcontrollers based on Arm technology, a growth of 60% year-on-year. That's really quite phenomenal, and we expect a rapid growth to continue. A lot of that is based on Cortex-M and these very tiny chips at very low price points that we're going to see everywhere around us. There are other applications that require more sophistication. They want to run complex operating systems, and they need higher processing performance. We're also seeing Cortex-A adopted in microcontrollers as well.
A great example is the latest chip from Xilinx. They have a product line called Zynq. Zynq integrates an Arm-based processor subsystem with their FPGA fabric, and their latest product has a Cortex-A53. It has a Mali GPU. It is based on our physical IP supporting TSMC's 16FinFET+ process. This is a very sophisticated device. It can run Linux. It can run complex operating systems. One of the use cases for that right now is controlling wind turbines. We are in chips that sell for GBP 0.10 and below. We are in chips that are in your smart mobile device, giving you a supercomputer on the road. We are in chips that are going into wind turbines. The scale of the adoption of the Arm technology is as broad as it has ever been, and we can only see it widening.
The growth of microcontrollers, as I said, it has been great. When you now start connecting those microcontrollers with sensors to the internet, that is what is going to lead to the Internet of Things. Bringing all of that technology together is quite difficult. The need for low-power radios is something we have been researching and looking into for a number of years now and led to an acquisition that we did in Q2 of a company called Sunrise Micro Devices. There, what we have done is take their deep knowledge of building low voltage, low power radios, and created an IP product to enable people to integrate the radios with the rest of their system. We believe in this way, it is going to accelerate people developing Internet of Things devices. We have launched a radio product. We call it Cordio.
Also, during Q2, we launched an IoT subsystem which integrates the processor with the memory subsystem and an interface to the outside world that people can use as a building block to, again, accelerate the development of IoT products. To test all of this out, we built ourselves a little demonstrator, kind of partly for fun, but also to test out that it did all work together. We like to do that before we ship it to our customers. What is shown here is a little Bluetooth LED beacon that we put together. As you see, diameter wise, when you look down on it is about the size of a EUR 0.01 coin. Very small. You could go stick it in the wall. It has got a tiny battery connected to it. Will last a very long time and can communicate with your mobile device when you are in range.
This was an experimentation vehicle. We built this in partnership with TSMC on their 55-nanometer process that they are optimizing for this kind of product. To help roll this out to our customers as well as TSMC, we are working with others such as SMIC in China, Cadence, who are a tools company, to help deliver this to our customers and enable it to be used as easily as possible with as predictable as possible a result. We are starting to license this. In fact, just after the end of Q2, we did our first license for this package, the radio and the IoT subsystem. This time, not with a semiconductor company. It was actually with SK Telecom, a big Korean telecoms operator.
You might think, "Well, why do they care about this?" They're a great example of a company looking at the Internet of Things, thinking about how they're going to deliver products and services to their customers, drive traffic across their network. Is an example of the kind of company that we're working with as our ecosystem expands all the time. Who are thinking about the future of technology, how it's going to be used, who's going to use it, and partnering with us to make that a reality. Networking is another area that's changing. Networking is moving from a world of hardware fixed functions to a more flexible, software-driven world. That is a change that we believe benefits Arm, and it's one that our business model can support in a cost-effective and efficient way. It's an important market to us.
You've heard us talk about it before. We now have 12 partners shipping products. The growth in units, whilst in absolute terms is small, it's grown about 30% year-over-year. An important market for us, one that's growing, and one that we are focused on. The volumes today may be low, but what's important are the design wins and the development of the ecosystem to make it very easy to use an Arm-based SoC in networking equipment. There's lots of work going on on the software stacks that are required here. Lots of great work going on in Linaro, whose membership increases, and there are big names coming to Linaro. I recommend you have a quick look on the website. You can see the names coming there, new companies joining up on a pretty regular basis.
We've had a great collaboration across many of our partners to build a demonstrator around what's called Open NFV. Open because it's open, and Network Function Virtualization, which is one of these new technologies that are going to power the networks of the future and give the flexibility to deliver the performance and deal with the changing needs of the load that's put on the network. Great progress in networking. In servers, again, we believe great progress towards our goal of around a 20% market share in 2020. In Q2, we saw Computex. In Computex, there were many design wins announced for people using Arm-based silicon. Companies like Applied Micro, Cavium, Annapurna Labs are just some examples. Gigabyte, who physically make boards and servers, announced products based on all those three companies' chips.
We've seen products being announced, and we've seen data showing the results of using Arm-based servers. PayPal is a great example. PayPal have taken an HP Moonshot system, one that's got a chip in from TI with a Cortex-A15 and a DSP integrated into one low-power SoC. They've rewritten their fraud analytic algorithms to utilize the fact that you've got this mixture of integer processing and a DSP. Through the combination, they can get a level of efficiency that you just never get in a conventional server. This is enabling them to do analytics at a rate that they just couldn't do previously and with a huge reduction in power consumption. They get about an 85% reduction. The results here are great. We're also seeing the rollout of developer clouds.
We're seeing the rollout of the servers and people really putting them into use. There's an example here of the Barcelona Supercomputing Center, where they have a high-performance computing environment to enable researchers to run these kind of algorithms that study the weather and all sorts of deep data type of applications. Great progress in servers as well. We always said it was going to be a long journey. It's going to be a long journey. That's all about today, what's going on, what's going on in the established market, in the established world. I mentioned our partnership with UNICEF at the beginning, one thing we really believe in is that the technology we're developing shouldn't just be for people in developed countries. We believe that technology should enable opportunities for everyone.
Behind that, we've been working with UNICEF, and we launched a challenge during Q2 back in May with UNICEF and frog, who are a very well-known design company based out in California, to launch a contest to say to people all around the world, using all this low-cost electronics, using easy-to-deploy software, using smartphones, using cloud computers, what can you do with all of this to create wearable devices that were going to solve real-world problems for people in developing countries? We've launched this competition. The uptake so far has been very strong. We've had about 700 people come to the site, download the information, and register their interest in it. It's going to run through this year. Some of our staff are getting involved in mentoring the teams that are coming together.
Come the end of the year, we'll announce a prize, which is like $15,000. We hope to see some great innovation. Already, when you look at some of the ideas that are being posted there, it's just things that you wouldn't think of if you're targeting people who live in cities like London. We're very proud of this work, and we believe it's going to expand the use of Arm technology and, as I say, enable opportunity for everyone. Now, one of the other things we passionately believe in at Arm is getting young people interested in STEM, in sciences, in computer science. This isn't just altruism. We want to hire the people at the end of the day to come into Arm who know what they're doing and have a real interest and a real passion for what we do.
We hire a lot of graduates, and we train them up to become the engineers that we need. We've had about 60 join us in Q3 so far because it's kind of graduate intake season. We want more. We want more in the future. The world's going to need more engineers, and Arm's going to need more engineers. We've been working with the BBC and supporting their new Year of Coding to create a product called the micro:bit. The micro:bit is a very small, I mean, the size here doesn't really show it off. It's about the size of a box of matches. A very small computer that's going to be given away to schoolchildren. In September, there's going to be a million of these given away to schoolchildren around the U.K. It's something that we are, again, really proud to be involved in.
If you go back in time, back to 1985, the BBC was instrumental in creating Arm. They created a program to enable people to learn about computer science. They worked with Acorn. In fact, the reason Acorn existed from where Arm came is because of the BBC Micro. That inspired a generation of engineers. Certainly inspired me to get involved with it. I used to have one of these. Big, bulky, cost me GBP 400, which in 1985 was a lot of money. If you roll that forward to today, that's over GBP 1,100. You contrast it. Right now, we're about to give away 1 million micro:bits. It's got much more compute power. It's tiny. I can remember trying to take my BBC Micro on the back of my bike around to my mate's house, heavy, high probability of accident. This thing is tiny. Kids can play with them.
If it breaks, get another one. We really hope that this is going to inspire the next generation, and maybe 15 years on, another company like Arm is going to come from this. In summary then, Q2 has been a strong quarter for Arm. The first half has really carried the momentum that we ended the year, ended 2014, and we've delivered some solid results and execution on all our strategic markets. We've made progress on a number of fronts, and we've seen the benefits of the multi-year investment in R&D that we've been doing over the last few years. We've seen the licensing of our latest technology to many companies, not just semiconductor companies, but we're seeing a broadening of our customer base.
Along the way, that success has enabled us to increase cash returns and keep investing in the business for the future. At that point, I'm going to hand over to Ian, who's going to give you a bit more color on the numbers.
Thanks.
Yeah.
Thank you, Simon, and good morning, everyone. After 29 quarters sat there in the front row, it's great to finally have the opportunity to stand up and present some of Arm's results to you. Simon has already given you an overview of the royalty and licensing dynamic for the quarter and has also sketched out the opportunity for us over the next few years. I'm going to keep this fairly short because I know that you really want to get onto your questions anyway. What I will do is just give a bit of color around the P&L, about cash for the quarter. Just quickly walk through outlook, and then we've got two dates for your diary. As Simon mentioned, revenues are up 15% to $357 million year on year.
Of course, as you know, through the quarter we did see a weakening of the dollar versus sterling, but we're still benefiting on a year-on-year basis from a stronger dollar. The effective translation rate in Q2 just gone was 156 versus 165 from a year ago. When we translate that $357 million into sterling, we end up with GBP 229 million, which is up 22% year-on-year. As Simon mentioned, we hired 127 new people during the quarter, which is up 230 since the beginning of the year. As a consequence of these extra hirings and this investment in the business, our underlying operating costs were about GBP 104 million.
Because of the change in the exchange rate, when we marked to market at the end of the quarter, some of our longer-term contracts, we ended up with about a four and a half million GBP FX credit. The normalized operating expenses that we've actually ended up reporting is just over GBP 99 million, up 13%. With sterling revenues up 22% and sterling costs up 13%, this has been a very good quarter for profits. Profit before tax up 32%. The effective tax rate in Q2 is a little bit higher than it was in Q1 at 16.4%, but we still think that the effective tax rate for the full year will be around about 16%.
That's ended up delivering a normalized EPS of about GBP 0.073, which is a little lower than in consensus, but that's mainly because most of the analyst notes that we still have in consensus, or analyst models instead of consensus, still have an exchange rate from Q1 and also a tax rate similar to Q1's as well. Q2 was also a very strong quarter for cash, with cash generation of GBP 93 million. Through the quarter, we did, as Simon mentioned, we announced a couple of acquisitions. Two acquisitions, Wicentric and Sunrise Micro Devices, taking to eight the number of acquisitions or teams integrated over the last eight quarters. We also bought back four million shares, taking to 10 million shares we bought back over the last 12 months.
If you recall, this is a limited buyback program with the intended goal of keeping the share count flat over time. Indeed, if you look back from June last year to the end of June this year, indeed the share count is broadly flat. We also have announced that, over the last five years, sorry, we've grown the dividend by about a 25% CAGR. Indeed, the interim dividend that we've just announced as well is consistent with that long-term trajectory. We ended up with cash at the end of the quarter being around about GBP 904 million. That's actually down a little from the GBP 922 million reported at the end of Q1, partly because of the M&A, partly because of the share buyback, but also because we paid the final dividend of 2014, sorry, in this quarter. Just now to outlook.
As I mentioned on the previous slide, the underlying normalized operating costs for Q2 were about GBP 104 million. Most analysts when looking at Arm's costs going forward will usually increase our cost by about GBP 2 million per quarter. As Simon mentioned, Q3 is a little bit ahead of that usual run rate because this is the quarter where we do our graduate intake, and so the head count increases by a little bit more in Q3. We're guiding GBP 106 million-GBP 100 million for our costs in Q3. For revenues for the full year. This is assuming that the macroeconomics don't really change and that there's no further impact on either consumer spending or semiconductor industry confidence.
Looking at processor licensing, the opportunity pipeline for future licenses is robust, and that's encouraging for both license revenue and also for backlog in the second half. Our H2 royalties are generated from chips that are sold in Q2 and Q3 because we recognize our royalty revenue's one quarter in arrears. Standing here on the 22nd of July, we're effectively four sixths the way through that period. There's already some industry data for the first two months, for April and May, and that indicates that Q2 for the industry will be up slightly sequentially, about two and a half % sequentially. Arm should do a little bit better than that because of the increasing penetration of Arm v8 and Mali and chips with higher core counts going into mobile devices, and then our ongoing share gains beyond mobile.
Taking those three things together, we think that for the full year, our U.S. group revenue should be in line with market expectations of about $1.48 million. As I said, two dates for the diary. On the 15th of September, we have our normal analyst and investor day. This is here in London, and as you can see, we have most of the senior team presenting, including, of course, Chris Kennedy, who will be joining us on the 1st of September. We're taking a deeper dive into some of the opportunities that Simon's talked about this morning.
Between the 10th and the 12th of November, Arm hosts its technology conference, which is I think, a unique industry event where thousands of engineers come together in Santa Clara to talk about and to find out about Arm and our partners and our customers' roadmaps and strategies going forwards. In parallel to that, on the afternoon of the 11th of November, we have an investor event where we gather together some of the Arm execs and some of our customers' execs to talk about their plans for their Arm technology going forwards over the next few years. Details of that are on the website as well as on this slide, and you can sign up and register for those events there as well. With that, I'll hand over to Simon, who's going to chair the Q&A.
Thanks, Ian. Not bad for a first go, was it? Okay, we'll go around the room. We have some microphones. If I could, as usual, just ask that we go one question at a time so everybody gets a chance. Off you go.
Thanks. It's Achal from Credit Suisse. Simon, just generally, obviously you benefited a lot recently from the inventory correction we saw last year. Recently, there's some chatter in Asia that we may actually see some more inventory correction, probably in Q2. I'm just trying to see, are you basically seeing that at your customers at this point? Secondly, just taking that along, should we actually expect more of these inventory corrections from time to time because the growth rate in the smartphone industry in general is actually coming down, and we're still getting used to a slightly lower growth rate as opposed to what we saw actually two years back, or 18 months back?
As Ian was saying, in terms of the data points that we've seen, that supports where the market is at for our numbers through the rest of this year. I can remember, it might have been 12 months ago, standing here, and we showed a graph of the industry cycles over the years. Certainly, there are peaks and troughs to that. I think it'd be a brave man who would say that there will never be any inventory correction or oversupply ever again. I'm sure that's probably going to happen at some point. For now, we're in the middle of earnings season, or really quite at the start of that. We're going to be looking at what our licensees say. We're in an early stage of, in fact, collecting royalty reports from our customers right now.
As we sit here now, there are no data points to really change our outlook for the rest of the year. Just going to move along now.
Thank you. It's Andrew Dunn, RBC. There's a line in the statement which talks about a subscription license being signed with a major Chinese OEM. Are you able to say anything more about that, the type of end markets that customer might be addressing? Maybe perhaps more broadly, just a little update on the importance of China to Arm. Thanks, Ian.
If we had been in a position to say more, we would have done. Generally, as I said in the presentation, we're seeing OEMs take more of an interest in the technology that's going into their products and really wanting to optimize that to deliver the best experience to consumers. That was an example. China's an important market to us. The Chinese government's investing significantly in developing its technology sector. There's a lot of money going into semiconductors, manufacturing, and design. We have a very strong team on the ground working with the companies who are emerging to support what they want to do. It's an important market to us. It represents a reasonable portion of our revenues, and it's one that we're expecting to grow over time.
As part of our recruitment and the evolution of our organization, we're making sure that we're best positioned to work with the market as it develops. Carry on there.
Thanks. It's Kai from Merrill Lynch. I had a question on the relative performance versus the semi industry. I think in this presentation, you did not have the chart in there. Just wondering because in mobile, it seems v8 adoption is still fairly limited to the high end. I think low and mid-range, it will come eventually. I'm just wondering, from a directional perspective, how do you think or how should we think about your relative performance against the semi industry, bearing in mind their demand? I'm just wondering about for the next couple of quarters, how we should think about that. Can it increase or go up further from the last couple of quarters? Is it going to stabilize at a higher level or maybe slow somewhat? Thanks.
Well, our royalty revenues in Q2 were up 31% year-on-year. The semiconductor industry over the relevant period, which is Q1 on Q1.
Was up in dollar terms 4%. That would optically look like a significant outperformance versus the industry. To a certain extent, we have to bear in mind that last year we had easy comps because of the inventory correction. Going back over two years, our royalty revenue growth over the last two years has been about 16%-17%, and the industry's been sort of 3%-4%. You're still seeing that sort of around about a 15% outperformance versus the industry. I think certainly over the next few periods, we should continue to see a similar sort of level of outperformance because as well as gaining share beyond mobile, within mobile, we're still very much in the early days of Arm v8 penetration. Of course, that typically delivers a higher royalty revenue per chip.
Also Mali, our graphics technology, is gaining share that adds a higher royalty as well. As we see more chips with higher core counts, typically if you have an octa-core, you need two Arm processors, and that will typically deliver a slightly higher royalty as well. The combination of those things means that even within mobile devices, we should be able to get our royalty revenue continue to increase. I think that outperformance should continue for a while. The crystal ball gets cloudy in the outer years as to when does the growth from enterprise networking and servers that start to contribute in the future. François.
Yes, thanks. I've got a few questions. In terms of networking, I think you say that your revenues are up 30% year-on-year.
In units.
In units. Right. Do you have any number on the market share compared to last year, which was around 10%, just to compare? It's gone up then.
We haven't done quarterly updates on that. We think that 10% probably trends towards about 15% through this year.
Okay, 15% increase. Okay. In terms of backlog, it's good to see that Tim Score has not left any surprise for the next year.
I know where he lives.
I know where he is as well. It looks like the backlog was actually up quarter-on-quarter. Shall we say nearly for sure that Q1 was a trough, basically, and that given the pipeline you've given, it should kind of at least be flat for Q3 and Q4?
Well, I stand behind what we said before. Licensing is lumpy.
Yeah.
There are quarters where people sign long-term subscription deals. There are quarters where people are signing shorter term deals for mature technology. When you're talking about the number of licenses in a quarter that we are, that can move around all over the place in a quarter. A lot of our license revenue comes from backlog. That's in a sort of 40%-60% range. Q2 was consistent with that. We think, looking out, if you stand back from it, you're going to see a smoother progression. That sort of 5%-10% growth in licensing over time. Backlog obviously needs to grow fundamentally to support that. We think the opportunities are there to make that happen. On a quarter-to-quarter basis, you'll see lumpiness around bookings, and depending on the mix of new technology, old technology, either more or less of that comes into turns revenue.
That is going to be a dynamic thing. It has been in the past, and it will be in the future. As I mentioned in my remarks as well, when we look at the pipeline for this year, the license deals we have visibility on for Q3 and Q4 are encouraging both for license revenue and for backlog this year as well.
Okay. The last question is about v8. There are two new words in the press release. One is called Artemis, and the other one is called Kite. Does that mean that v8 is actually going to more or different end markets than the one we know about? Is that the plan?
Well, it might do. We didn't say what those products were. We've talked about Artemis before. If you go back in time, when we introduced v7 processors, that started in mobile. We produced v7 processors for other markets over time. Up until we formally give a product name to Artemis and all the rest, we've only got three v8 products out in the market. They're all largely focused towards mobile and the high end. A historian, my historian friend is no longer with us, might look at that and go, "Okay, well, over time Arm's going to create v8 products to address more of the product lineup," and that wouldn't be a bad assumption to make. Maybe some of those are in the code names.
All right
We'll tell you later.
That carries as well a slightly higher royalty rate?
Well, yes. v8 commands a higher royalty rate.
Just checking.
Okay.
Thanks.
Okay.
What's your view on Microsoft Windows 10? Do you think that will create opportunities for Arm?
How the market's going to react to Windows 10, I'm not going to make a prediction for. What we've seen is the PC market appears to be slowing. We think that consumers generally are enjoying the computing experience that you get from a smartphone more and more. We think there's a great future in providing more technology to make those devices even better. I won't say always. For a long time to come, there's going to be a need for PCs.
Arm processors are used in PCs. There are typically a few Arm processors in each one of them. How the market's going to react to Windows 10, I'm not really in a position to make a call on that.
Thanks. Couple of questions. Firstly, on application processor ASPs. Historically, well, for the past few quarters, you've said that the trend is flat. TSMC in the Q2 said they're expecting their smartphone content in 2016 to go up in the high-end and in the mid-range. I wondered if you could share your observations, whether you agreed or disagreed with TSMC. The second question is more on OEM vertical integration. Not sure if there has been a quarter where five OEMs signed up with Arm. I suppose more OEMs coming in is good for business, I wondered how the traditional Arm customers, i.e., the semi companies, react to this trend, whether they start seeing you as an indirect competitor.
I suppose in mobile, they don't have much of an option, but some of the newer end markets, they might choose to give your competition a shot just to make sure that they can offer something different with what the OEMs can build in-house with Arm. Thanks.
Okay. To your first of two questions. In terms of TSMC, remember that they manufacture wafers containing other people's designs, and if more people choose to manufacture with them or someone else, that's about market share shifts in the foundry space. In terms of the content, as we've explained a number of times, we think there is a trend towards more processors, more graphics processing, more integration in mobile, and therefore more Arm content in mobile devices, whether they're high-end devices or low-end devices. That's the trend that we're looking at, where a licensee of ours gets their chip manufactured is secondary to that, and it doesn't really affect the outcome from our perspective. In terms of OEMs, I think to your question about whether that changes the competitive dynamic for us, I don't think it really does.
The beauty of the Arm architecture is that our business model means that lots of people can get access to it. Lots of people can innovate around what we do. There is a big ecosystem around Arm, writing code, providing other products, other technology, software that relies on the Arm architecture. It's that that people look to optimize in different ways. Whether you're a merchant semiconductor company or an OEM, your approach to that is going to be different. We see this as a great overall trend because the more people innovating, the better the outcome at the end of the day. There's obviously some consolidation going on in semiconductors. This broadening of our market is good for us in that perspective. Keeping the number of people innovating high, I think is the best thing for the product at the end of the day.
That has benefited Arm greatly and is enabled by our business model. Just there. Yep. Sorry.
Hi there. It's Eoin from Liberum. Just a question. You've maintained your full year outlook, your full year US dollar revenue. I'm just trying to correlate that to your key customers. If I take the consensus numbers for, say, customers or partners, TSMC numbers have crept down over the last, MediaTek numbers have crept down, Samsung numbers have probably come down, and Apple after last night's numbers will probably come down a bit. I'm just trying to square off, you're maintaining your full year outlook, but your key customers are kind of cutting or reducing their outlook. I was wondering, what's the difference here? Did you start the year with a lot of upside in your royalty numbers?
When looking at the second half, as I mentioned earlier, we're already to certainly four sixths of the way through H2 on the royalty side. We've got data already regarding some of our customers and what they're going to be paying us in terms of their royalties. Because of things like increasing penetration of v8, increasing penetration of Mali graphics, then we're actually getting a higher royalty from some of the chips that our customers are selling. I can see a route to getting to the royalty expectations and consensus. Consensus is basically looking for a 20% year-on-year increase in royalties for the second half versus the first half.
On the licensing side, again, because we've got good visibility of the licenses that we're likely to sign in the second half, we can then make some assumptions about which one of those deals are going to be made based on our past experience. The combination of those gives us confidence in license revenue and backlog for the second half. Adding all those up, we can see the $1.48 million is indeed an achievable number.
Thanks. Then just a brief follow-up on the consolidation. If two companies, Arm customers, are merging this year, could that lead to a delay in them signing a new license agreement as they try to work out their roadmap? Could that licensing deal slip into next year?
We haven't typically seen that as a factor when companies merge. This is not a new phenomenon. It's been something that's been going on for a long time. Typically, when we see big companies merge, you see more usage of Arm technology as companies look to consolidate costs and align on a single roadmap going forwards. I look at the consolidation that's going on and view the positives, and you might say, "Well, you would," but when you look at the complexity of the chips people are building, you need scale. When you look at the cost of manufacturing, you need scale. Having big companies being able to make the investments in the R&D that's going to be required to keep technology moving on, this is a positive thing for the industry. We expect that's going to translate into a positive result for Arm.
Thanks.
Yeah. It's Gareth Jones from UBS. A couple if I could. The OpEx going up in the course when you talked about graduate recruitment. I just wonder which areas you're pointing the graduates or just generally the hiring to. What are the new areas that you're kind of investing in for the longer term? That's the first question. Second question is just to follow up on an earlier one. Can you just confirm that Artemis, Greeve, and Kite, for like product, the sort of royalty rate pickup that we should expect from that? The last one is just on the server market and sorry. Actually, I'll pause on that.
You remember that one question?
Okay. I'll pause on those two.
Your first one was about hiring. I mean, Q3, we recruit a lot of graduates. We look for people who we think have the right technical skills and the aptitude to be flexible as they come into Arm because what they're going to be doing in the future is going to be very different from what they're going to be doing today. We have a long track record of recruiting graduates, moving them around the organization, and growing our own talent from within. At the same time, we hire people mid-career, we hire people senior career as well, to make sure we've got a mixture of homegrown talent and a deep and broad gene pool. When you look at our headcount profile, about 70% of our employees are technical engineers developing our products, and we maintain a high proportion of investment in R&D.
That's what these graduates are doing in the main. They are coming into our R&D teams, learning about microprocessor design, learning about graphics design, learning about silicon implementation, learning about software which are all the things that we do. Where these people end up working is simplistically, you can look at our revenue breakdown and go, "Well, that's roughly where the people are going." As I said, we are looking for flexibility in the people that we hire so that we can change the company in response to the needs of the outside market. That investment in R&D has been very important for us, because the company really exists from the quality of the products that we produce at the end of the day, and you've seen that come through here.
In those products, to your second question, their v8 architecture, we have worked hard to establish a higher royalty rate for v8. On different markets, support different royalty structures based on the dynamics of the selling price and the profitability of the end markets. Overall, the trend is towards higher royalty rates for v8 products.
Just gentleman in the front there, sorry.
Yeah. It's Lee Simpson, Stifel. Just three quick ones from me, actually. I think I saw GBP 10.6 million Linaro related charges. Just wondered what they were going forward in the out quarters. Second to that, a lot of the hires that you've taken on have been U.K. and U.S. I just wondered if that was indicative of a new architecture coming 2016, i.e., it's not India that you're hiring in at this point. Maybe thirdly, on PIPD, you talked about the footprint increasing, I think 30% attach in smartphones. Yet the backlog split is down noticeably in this quarter and mainly Q1 Q. I just wondered if there's an order book gap that we saw Q2 perhaps related to node migration at big customers. Thanks.
First question, I think was the Linaro charge-
Linaro, yeah
Which is I think it's a three-year subscription fee to fund Linaro and to keep that going. The main companies that are partners at Linaro pay to fund and keep it going. If you look back over two, three years ago, you'll see a similar charge. In terms of where the people are going, I'd say that bias towards U.K., U.S. probably reflects where we have more of our engineers working on processor development. Sorry, I'm just now trying to remember your question. Your third question was on physical IP licensing.
Third was PIPD. Yeah.
We've had a lot of success developing our POP IP, which our customers use to optimize their processes. There were a number of new licenses signed in Q2. There are a relatively small number of semiconductor manufacturers, foundries. As they develop new variants of their process technology, we typically engage with them in a license to develop physical IP that supports that process. Over time, the development of the technology as it matures can actually take a couple of years. The licensing and that and the flow to royalty is slightly more complex than it is, and slightly less predictable than it is with our processor IP. I wouldn't read too much into any of the trends there.
Great. Thanks. Here we go. Just on that one.
Thank you. Jerome Ramel from Barclays. A quick question on effects. You reported on an effective effects rate of 156
If you look at the average over the quarter and what everybody else was expecting, actually, there is a gap. Can you remind us why is it so different? Where did we get so wrong in our modeling?
Well, you have to bear in mind that in the quarter, 57% of our license revenue came out of backlog, and revenue out of backlog comes out of backlog at the same exchange rate it went into backlog. Deals that were signed, say, a year ago, would've been going in at 160, 170, and so they're coming out at that higher exchange rate. That is not something that you're going to be able to model particularly easily. Clearly that then gets combined with whatever deals are done during the quarter at the moment those deals are done. Obviously, the exchange rate moves around in the quarter. Again, that's something that's going to be very hard for you to model.
Also, as we receive royalty reports, they're recognized at the rate on the day. Again, if the rate moves around during the quarter, if a big report turns up on a day when the exchange rate's bad, nothing we can do about that. If it's good, hey ho.
Okay. In that context, since you have visibility into your backlog and licensing revenue over the next couple of quarters, is it going to be the same sort of rate that will be coming out of the backlog in the licensing index? Basically, the disconnect versus the actual rate for the next couple of quarters, is it going to remain as it was this quarter, or is it going to be different?
It can be, because obviously, if the rate stayed at, I think last I looked, it was about 155. If that was a stable rate going forward, we would still be taking revenue out of backlog that had been signed multiple quarters away. There is usually a lag effect between where the exchange rate is today and where it was. In periods where the exchange rate is higher, we'll be recognizing revenue at lower exchange rates, therefore, our exchange rate will be slightly better than the average for the quarter. Right now, where the exchange rate is, generally speaking, lower than it was 1 year ago, then it will be a little bit higher on average, going forward than the exchange rate, that if you just go and have a look at the Bloomberg and what it's telling you today.
Given we're not currency speculators, what the exchange rate's going to do over the rest of the quarter, I have no idea. Whether there is a disconnect between that and the average, impossible to call right now.
Okay. If we came to the other side of the room.
Hi. Johnny Pate from BTIG. Talking about consolidation, scale, and opportunities, Brian Krzanich, if that's how you pronounce it, Intel CEO, at their results presentation, called out the relationship between Arm and Altera in his strategy commentary. In fact, it was the first thing he said. He said Intel could substantially enhance Altera's Arm-based business. Are you entering a new phase of relationship with Intel? What can you tell us a bit more about that in terms of the consolidation, the opportunity, and the scale?
Well, in terms of what Intel and Altera do when they get together, again, I've got no control over that. Intel have licensed a lot of technology from us over the years and shipped actually quite a few chips, more than a lot of people expect based on Arm processors. Altera have developed a strong roadmap of technology over the last few years. They've been a licensee of ours for a long time and have some great products based on our technology. What they have said is that they're going to continue to ship those. Obviously, I hope that they do. Where that goes, hard to say. Your sort of follow-up question generally on consolidation?
Yeah, more broadly. Intel, clearly opportunities might develop with the rest of the management.
Yeah. We'll have to see. The kind of landscape for FPGA hasn't really changed with Intel's acquisition of Altera. That is a market that's dominated by two players, traditionally, Altera and Xilinx, both of whom are licensees of ours. I know I mentioned the Xilinx Zynq product there. They've been shipping FPGAs with embedded Arm subsystems for a number of years. That's been a successful business for them, and we expect that to continue. Just move further back down the room.
Okay. Sorry. Robert Sanders from Deutsche Bank. Just a question on Mali. It looks like you've licensed most of your addressable market in mobile now. How do you look at your Mali units in 2016 in terms of growth? Do you think that the growth rate could slow from your relatively decent number in 2015?
I think that's going to come down to the overall growth rates in mobile into the next couple of years. As you say, we've a very strong licensing base, most widely licensed GPU out there. We are developing new products. Of course, we've got a roadmap there. We're investing in that roadmap. We expect as long as we execute, as long as we deliver, that our customers are going to come back and license the next technology from us. Obviously, GPUs are less broadly applicable than CPUs are. Many of these microcontrollers aren't going to have a GPU in them any time. We are seeing more and more things having more and more sophisticated graphical displays, the market for graphics does broaden over time, and that's going to help us grow units as well.
It's about the dynamics of that market and making sure we develop the right products to take advantage of those dynamics.
Hi. Sandeep Deshpande from J.P. Morgan Cazenove. My first question would be regarding your royalty growth in the quarter. You had about 31% $ royalty revenue growth in the quarter. What percentage of that growth would you accrue to the move to the v8 architecture, which is increasing the royalty per device? The reason I'm asking is, given this ongoing inventory correction, the units might go down, but that increase you see associated with the royalty per device will probably remain right through this year into next year, potentially. I'm trying to understand that. The second question I have is on licensing. When we started the smartphone cycle, 2009, 2010, there were maybe 15-20 producers of baseband chips and application processors. All of them contributed to your licensing revenues, which were very strong through that 2010-2014 period.
Today, That industry has consolidated very dramatically towards 3 players in both those segments, basebands and application processors, maybe 5, if you take some of the stragglers who are still in the market. Does that not impact your future licensing in that market?
I'll take the second part of that first, and maybe Ian wants to take the first part. We have now 400 licensees of our technology. We've just announced many new customers in this quarter. What's interesting, if you look at how long it took to acquire the customers, what you see is it took about 10 years to get the first 100 customers. It took about five years to get the next 100. About another five years to get the next 100. Took three years to get the last 100. Despite the maturing of some markets, because of what we've done with our technology roadmap, we've been able to increase the number of end markets where Arm can be used. That's why new customers are coming to the Arm partnership. That's why we're seeing design wins. That's why we're seeing a strong licensing business.
Through that investment, we expect to see continued growth in royalties. I know we love mobile market, but we have been working hard to make sure that we are diversified in where the technology is used. I believe that is that strategy that's been in place for many years, is working well for us. Do you want to talk about that?
Yeah. In terms of the Arm v8, overall in our royalties revenues are up 31% year-over-year. The units are up 26%. Very strong growth in the number of units as well as the dollar values. If I look at the mix of Arm v8, bearing in mind that most of these Arm v8-based chips were sold into Q1, or devices that were sold in Q1, which is pre the launch of most of the premium smartphones. If I look at the mix of Android, 80% of the chips going into Android devices were quad-core A53, so the lower end of the market. Normally, a new technology like this, you would expect to come in at the very high end and then gradually over years, move into the mid-range and then move into the lower cost entry-level devices.
Apart from a very few number of high-end devices, most of the volume seems to be actually coming out to the low end and mid-range smartphones. I think that actually from Arm v8, there is more mix to come through the year as we start to see more smartphones being sold in Q2 that are going into more developed markets with higher priced smartphones with higher priced application processors in them. I'll let you do the math to exactly the question that you're asking, which is what proportion of that 31% versus the 26% is Arm v8. I think the answer to the question is, it's not as much as you might think.
Go there.
Hi. Jag here from Arete Research. Just a question from your licensees. There's quite a lot of inventory on their balance sheets at the moment, guys like MediaTek, Qualcomm, Broadcom. When do you guys get paid your royalties? Is it when it gets shipped by those licensees, or is it when it gets recognized on their balance sheet? Then secondly, with Cortex-A72 now entering the market, I'm just wondering, have you guys done any benchmarks of your FinFET chips versus Core M from Intel? Maybe how does that flow through to seeing Arm in bigger screen notebooks? Thanks.
In terms of when we get paid a royalty, it's typically when our customer sells their chip. When they get paid, we get paid. That's simplistically how the model works. There's a lot of complexity around that, but simplistically, that's how it works. In terms of performance, yeah, we believe Cortex-A72 in a multi-core configuration is a very high-performance processor. We've worked with partners such as TSMC on advanced FinFET implementations. We believe there's no reason why that can't be used in a variety of devices, including bigger screen clamshell-like things. Now, the only thing that stops that is whether any one of our licensees sees that as an opportunity that they want to go and explore. Is that market big enough? Is it growing? Is there an opportunity for differentiation? Is there a software solution?
There are lots of factors at play, many of which we aren't in control of. For our part, we think we've done the work on enabling processor technology that is of that class of performance. That can be used in lots of places. The networking applications, the servers, these are all high-performance computing environments. If one of our licensees wants to go and pursue that market, then fantastic. But that's not something that we are particularly pushing ourselves right now. Gareth.
Thanks. Gareth Jenkins from UBS. My third question was on the server market. I just wanted, looking back from where you set the 20% target and where you are going to, do you feel Intel's response has been perhaps more aggressive, or they have been able to fend off your partner, or will be able to fend off your partners better than you thought previously? Are you still comfortable with the software ecosystem development, et cetera, to get you there? I am just thinking that we talked about the lack of custom design chips from Intel. They are doing that more and more. They are kind of responding with lower priced, lower performance chips effectively. Thanks.
Yeah. I think our view on Arm's opportunity in this market hasn't changed. We think our business model enables many people to innovate and create different solutions. That many people addressing a market is going to give a variety of solutions that is broader than any one company can do on their own. I think if competitors in that space do build different solutions, actually I think that validates the strategy here. Ultimately, it is going to provide more competition into the end market, and that is going to make servers better. This is all goodness from our perspective. As I said, there are some great progress points. In just, I think it was last week, Cavium came out with a whole raft of press releases about progress that they are having some big cloud companies deploying ThunderX.
This is all progress, and the software ecosystem is moving forwards as the market develops. It is all very chicken and egg. Nobody wants to write the software unless there is the hardware. Nobody wants to build the hardware unless there is a software and so on. It moves together kind of in tandem. Linaro, where we have continued our commitment to Linaro, is a place where a lot of that work is happening.
One more.
just want to come down to the front here and better make this the last question, just being conscious of time.
Quickly, you said that you are comfortable with Q4 target of 50% of smartphone being 64-bit. You already ship 150 million in Q2, which was Q1, which is not far away from being already half of the smartphone market, assuming that most of your v8 is handset. Don't you think you're a little bit conservative on this target? Second question, we see more and more octa-core. We heard about the deca-core. What is the critical limit of what smartphone application processor guys are going to do? Is it 8 core, 10 core, or do you think going forward we could see more cores?
Hard to answer that last question. There is a theoretical limit for how much parallelism you can ever get out of an application. I think that we'll continue to see core counts go up as your phone is doing more and more in parallel. Processing what's coming in from the sensors is work that can be parallelized. Having multiple processes there when you're doing something that's really compute intensive gives you a lot of power. I think for the time being, we'll probably see core counts increase, and it's going to come down to this use model which, as I was saying during the presentation, is something that is created by not us, and that's really the beauty of it. The innovation that can go on once the platform is deployed kind of pulls through more technology over time.
We'll have to see how that plays out, but I think it's going to be very, very interesting to watch. Your first part of your question was about whether we're being conservative on Q4 shipments. Let's hope you're right. Looking at it right now, given what are some uncertainties out there in the market, I think standing right now behind the 50% exit run rate feels about right for us.
Of the 150 million, we have chips going to tablets, chips going to TV, there's TV sticks as well, as well as enterprise networking and servers. My best guess would be of where those 150 are going, about 130 into smartphones and obviously to Simon says, we'll have to see where those other chips get designed in over the rest of the year and what consumers actually end up buying to find out what the actual proportion is going to be in Q4.
Great. Well, thank you very much for joining us today, we'll see you on the road.