Good morning, everyone. Thanks for joining us today, and welcome to Arm's Q2 and H1 2013 results. Before we start, I'd just like to direct you at the customary cautionary statements, which I'm assuming you're all very familiar with, and we're going to take that as read. What I'm going to do today is talk about some of the progress we're making towards our strategic goals. Tim's going to talk through the detail behind the numbers. Let me, before we start, just give you some of the highlights. Q2 was a very strong quarter for Arm, driven by very strong licensing result. We licensed 25 processor licenses, seven Mali licenses, and this is really driven by leaders in the semiconductor industry, committing to Arm into their long range roadmaps. Royalties also grew strongly, 24% year-over-year growth.
That's significantly outperforming the industry, which only grew by a couple of percent, in the same time period. What that leads to is total revenue of $264 million. That enables us to continue investing in Arm's roadmap in hiring people to drive our R&D, whilst at the same time delivering profits and cash, and being able to increase our dividend return to our shareholders. We're very pleased with the way that Q2 has developed. What I'm going to do now is look at some of the key markets and the progress that we've made, starting by looking at smartphones. At the beginning of the year, we were forecasting a very strong growth for the smartphone market, anticipating about 1 billion units, in additional sales through the year.
It would seem from the industry statistics gathered so far that we're pretty much on track to deliver against that. Smartphones themselves, of course, have not just one type of device. We're seeing a tiering of smartphones. We have premium super phones, we have mid-range devices, and we have low-cost entry-level phones. You need to look at each of these markets separately when considering the smartphone space. At the high end, we've seen in the quarter the first big.LITTLE devices actually on sale, and delivering great performance and low power. Recently, we've just seen Samsung announce a second device based on big.LITTLE, the Exynos 5420, also incorporating Mali-T628 in a 6-core configuration. Another device based around big.LITTLE technology. The mid-range we see as a particularly interesting space as there's high growth potential for that market.
We see that growing to over half a billion units out in 2017. At Computex in June, we launched a family of products specifically targeted at that market. We launched the Cortex-A12, for which we have four licenses now. The Mali-T622 and the Mali-V500, which is a video accelerator. Together, those processors form a very power-efficient processing subsystem, which we think is ideally suited for mid-range phones. At the low end, in entry-level phones, we're seeing greater adoption of Cortex-A and Mali, which together are providing high processing performance, giving a great user experience, even though the device itself might be less sophisticated than some other products. Smartphones performing very well. Now, what we're seeing across the range is an increasing attach rate of Mali. In the entry level, in the mid-range, in the high end, all we're seeing devices containing Mali as well.
That's led to very strong unit shipments of our Mali graphics processor through the year. Far, in 2013, unit shipments have already outpaced all of 2012. We're on for very strong growth, through this year. The smartphone market itself, as I said, is developing in quite interesting ways. It's easy to think that everybody on the planet has a smartphone, but actually, when you look at global penetration of smartphones, it's actually quite low. This is industry data that we've taken for 2012. You see in developed regions such as North America and Europe, smartphone penetration around 50%. In other countries, in other regions of the world, much lower.
The portfolio of technology that Arm is producing allows devices to be built at different performance points, different price points, to allow smartphone penetration globally to increase by having the right product for the right economics of a given market. We're particularly excited by growth in the entry-level phones, the low-cost devices. Although many people say these are a bad thing, actually, we see these as a very good thing because it enables many more people, billions of people, to get access to smartphones, which they just wouldn't be able to if the industry only produced very high-cost premium phones. Smartphone penetration, as I say, across the world, quite low. If you look at Russia's got a population of over 140 million people, and penetration there is only about 9%, very low.
When you look forwards to 2017, just taking Russia as the example, we see smartphone penetration going up to 70%. That's about another 85 million people getting access to smartphones. When you consider that the average monthly income there is about $800, clearly, few people are going to spend $600, $700 on a smartphone. Penetration there is only going to increase through the availability of low-cost devices, we are designing products specifically to address those expanding markets. If you look at 2017 as a whole, we see growing smartphone usage in all areas of the world. In very large populous countries, we're seeing significant increase in use of the technology. China and Brazil, we're seeing roughly a doubling of smartphone usage. In India, you're seeing a factor of 10 increase from 4% today to about 40% out in 2017.
Even if you factor out increased population in that time period, is about another half a billion people using smartphones. As I said, that wouldn't happen without the availability of low-cost devices. We wouldn't have access to this additional market without progress in that technology space. To understand that a bit further about what's going on behind it, we need to look at some of the silicon content and the devices themselves. Here we have that bar graph showing 2017 breakdown of devices from premium phones, mid-range, and entry-level, compared to today in 2012, you see significant growth there. When you look at the devices at the entry level, we're expecting phones at below $200, probably significantly below $200, containing multi-core processors, Mali GPUs, and built on silicon, potentially using our physical IP as well.
In the mid-range, where the price goes up, anything up to about GBP 400, we would start to see big.LITTLE processors, adding GPU compute. There are numerous algorithms which can benefit from GPU compute. If you take, for example, taking a photo of a group of people and you want to change the color or the contrast of it. You can do that today, but when you do that using a GPU compute, those algorithms are very well suited for running on a GPU. You can do it more quickly and use less of the phone's battery along the way. GPU compute, important technology, and we'd expect to see that in the mid-range phones. In the high end, we would expect to see processors based on version eight of the Arm architecture that adds support for 64-bit.
That's going to enable more memory in devices, more sophisticated algorithms to run. Again, GPU compute probably in a greater core configuration. Across the range, the silicon can be built using Arm's physical IP. A lot of Arm content and obviously different Arm content in different devices across the tiers here. When you look at that silicon content, you'll see different application processors designed specifically for these different market tiers and different numbers of companionships based on the amount of connectivity, the number of sensors, and so on, that will be found in the end device. What that leads to is a different royalty opportunity for Arm in each of these different tiers, but a significant one. What we've done here is compare against the royalty opportunity of a basic voice-only phone.
You can see we have 4 times the royalty opportunity there in the entry level all the way up to 20 times in your super phones. Again, the market wouldn't grow in this way without the availability of these lower cost devices. When you run those numbers through, we expect to see compound annual growth rates of about 20% for handsets themselves. That leads to a silicon value, a CAGR, of about more than 10%, leading to a royalty CAGR for Arm in this period for handsets or smartphones in the 15%-25% range. What we're doing is designing products specifically to target these markets. I mentioned Cortex-A12 and the T62 specifically targeted at mid-range. Cortex-A7, we designed that for entry-level phones, and the adoption there has been quite strong. Cortex-A50 series we designed for the higher-end phones.
We're developing a portfolio to address these growing markets as they grow in different ways. Now, smartphones aren't the only device that's evolved quite a lot over time. When we look at mobile computing, we see a lot of interesting changes in the dynamics of the market. Computing used to be something that you did sat at your desk in front of a keyboard and plugged into the wall. Of course, computing now is something you do or can do whilst on the move. Driven significantly by tablets, which really have revolutionized the way that people access the internet and access technology as a whole. Tablets really are dominant in this mobile computing space. We've seen lots and lots of designs around Arm processors in tablets. Again, price points are changing, prices are coming down, and low-cost devices are being produced in this sub-GBP 100 category.
I bought one myself recently in Fry's in California, $100 tablet. It's very good, great for reading on, for example. These different tiers, these different prices are again going to enable access to billions of people across the world. Again, we're seeing good opportunity for our Mali graphics processor in tablets. Today, we see about a 25% attach rate of Mali graphics in tablets, we expect that to grow. Overall, we're expecting our market share in mobile computing, which we categorize as the sum of tablets and netbooks and laptops, to grow to over 50% through this year. It isn't just about tablets for us. We are starting to see clamshell form factor devices based on Arm.
Whilst we've talked about it a lot, it's interesting to note that the Samsung Chromebook is still the number one selling laptop on Amazon in the U.S., that's been there for over 260 days. It's proving to be a very popular device based on Arm, great performance, very low power, long battery life. Arm is more than just about mobile. We spend a lot of our time developing products for mobile devices, that's given us a lot of expertise in low power design, low power is useful across the entire spectrum of embedded processors. If we look at embedded, we see great progress in the use of Arm technology in this market. The Cortex-M series has proven very popular amongst companies wanting to produce very low cost, very low power microcontrollers.
We added nine new licenses of Cortex-M in Q2, taking the total up to around 180. Five of those nine new licenses were with companies who'd never licensed Arm technology before. The microcontroller space is very active right now, new companies are coming into the space because there's a vast range of end applications that can benefit from embedded intelligence. Anything from washing machines, obviously cars, dishwashers, energy control, embedded in light bulbs. There's just a gazillion, it seems, different end applications for embedded technology. The chips are so low power, so small, so low cost, that they can be used in many different applications with very little impact to the end price of whatever it's getting designed into. A particularly interesting area right now is in wearable electronics, even ingestible electronics. Freescale have produced a microcontroller that's about two millimeters on a side. It's tiny.
That's been driven by probably U.S. medical industry looking at embedded chips in something that you swallow for monitoring and data collection, whatever. Interesting application, again, driven by very low power, very small die size. This space is really ripe for explosive growth. If we look at our 180 licenses that we have, what we see today is only about 50 are shipping and contributing to royalty. But cumulatively, that's added up to 4.7 billion Cortex-M based chips so far. The other 130 licenses we've signed, most of those are being designed into [end market] products right now, we would expect to contribute more to royalty over time. We see huge potential for growth in the microcontroller space based off the continued success of the Cortex-M family, which again, we're continuing to invest in. The other end of the spectrum, are servers.
Q2 was a very busy quarter for Arm and our partners' progress towards delivering on this promise of again, lowering the power and disrupting the data center. Particularly, Calxeda and Applied Micro have been very active in Q2, talking about the design wins they've had and that's great to see the progress there. We also saw AMD announce two products, their Seattle roadmap which is an eight-core and a 16-core Cortex-A57 based devices, two chips there. They are very bullish about the prospect for Arm in the data center. This is a quote from Andrew Feldman, who's a general manager over at AMD. A long history in developing servers, knows what he's talking about. He's seeing great prospects there for the products that they've developed. We get asked a lot about progress on software for servers.
A lot of the infrastructure, a lot of the software that's running on the data centers right now is based on open source, that's obviously contributed to by many different companies. It's been great to see Oracle just recently announce their support of the Java SE for both 32 and 64-bit Arm architectures, specifically targeting enterprise and in fact embedded as well. Great progress on both the hardware side and the software side in servers. Connecting servers to sensors, to smartphones, to tablets, is billions of GBP worth of networking equipment. We've seen strong design wins over the last little period. We anticipate opening a lot of the next generation of network infrastructure to be based on low power on technology. We've had our first royalties in Q2 from base stations.
We've seen LSI, who've been very active in this space, with win sockets at two of the three largest base station vendors in the world based on their Cortex-A15 devices. Again, software is key. Linaro, which is an activity we set up to create a community around developing Linux infrastructure, Linux software for Arm. Linaro has formed a special interest group for networking called LNG, Linaro Networking Group. In the quarter, we saw both Cisco and Nokia Siemens Network join Linaro Networking Group. What that's going to do is provide the key software building blocks, but done in a very industry efficient way. No one company has got to foot the bill for all of this. It's shared amongst everyone, and the results are shared amongst everyone. A great partnership way of solving the problem of software migration. Networking we see as a big opportunity.
Looking out at 2017 again, we see a chip TAM of 700 million devices. The revenue implications of that are a GBP 17 billion silicon TAM. We would expect to win about a 20%-30% market share out in that time frame. This is a big silicon market, we do expect growth there as these design wins start to come to fruition and the royalties start to flow from that. All of the markets I've been talking about benefit from high performance and low power. All the emphasis that Arm has had in the last 22, nearly 23 years now, on mobile and low power is paying off and enabling us to target all of these different markets. Everything can benefit from lower power, everything smaller die size, and lower cost. That heritage in mobile has given us a lot of expertise.
When we look at how we compare, we believe that we compare very well. Obviously, with the size of these markets, there is competition. When you look at how we stack up against some of our competitors, picking different markets in turn, I think we're in a very strong position. If you look at entry-level phones which today are being based, designed around Cortex-A57, really Intel don't have a product that's suitable for that market. Clover Trail+ is the smallest thing they have, but is enormous in comparison to Cortex-A57. Sorry, Cortex-A7. Cortex-A7 delivers roughly the same performance, but at considerably less power, and crucially, in a really small die size. The benefit of that die size is that everything else that you need to put into effectively a single chip application processor with modem can be integrated onto one device at low cost.
The modem, the connectivity, the GPU, the memory can all be integrated onto one SoC. That, built on a foundry process, can sell for as low as GBP 5, and our customers are doing that today. These devices are shipping today. If we look in the mid-range, which again, devices are shipping today based on Arm, based on Cortex-A9. Cortex-A9 outperforms Clover Trail+. Again, it's lower power, it's considerably smaller. Again, it can be built into an SoC with everything else you need, and built on a foundry process, and sell for around about GBP 10. Very low cost, very low power, great performance. Shipping now. In the premium end, as I mentioned right at the beginning, we're seeing big.LITTLE devices shipping now. That adds significant performance in comparison to Clover Trail+, but at the same time allows power to be lower because of the big.LITTLE architecture.
Again, relative size-wise, the cluster configuration is still very small. When you build that on a foundry process, you can build this whole SoC and sell it for GBP 15. We can deliver the performance, we have very low power, and we can hit these very low price points that are required for these large volume markets. The same extends up to servers. In server size, it's about total cost of ownership, and how much it costs you to acquire the equipment in the first place, and how much it costs you to run it based on the electricity the servers themselves are going to use, and all the air conditioning that you need to power your data center.
Today we have 32-bit Arm processors based on version seven of our architecture, shipping today in scale-out server applications that are lowering power, lowering that total cost of ownership. Again, built on a foundry process, our licensees can sell chips at this sub-GBP 100 point, which if you're building servers, is a very low cost. All of those examples are based on processors which have been designed and are shipping today. We of course have a next generation of technology coming through. Talked about Cortex-A12. Last year we announced the Cortex-A50 series, which has Cortex-A57, Cortex-A53. That's a big.LITTLE 64-bit configuration. As that technology delivers into these markets, again, we'll see more performance. We're taking that to the next level, whilst at the same time maintaining the focus on low power, low cost so our partners can produce these very aggressively priced SoCs to address these very large markets.
Before I hand over to Tim, let me just summarize the quarter. It's been a very successful quarter for us, characterized by major semiconductor companies making long-term commitments to Arm technology. That licensing has delivered great financial results, allowing us to continue to invest in our R&D while returning cash to our shareholders and increasing our dividend. With that, I'm going to hand over to Tim.
Thanks, Simon. Morning, everyone. Quick run through the numbers, a little bit of color, talk a little bit about guidance. Simon has already gone through the headlines, I won't dwell on them in too much detail. A 24% overall revenue growth in the quarter. The very strong licensing that we've seen in the last three-plus years continues, up 32%. Lots of new licensees, lots of existing licensees, either upgrading in the vertical or broadening their use of Arm into other end markets. Royalties up 24%. We were up 33% in Q1, you probably recall that we called that probably a slightly inflated level based on the inventory correction a year ago. A 24% and 22% outperformance versus the industry at the top end of our normal experience. That's driven normalized PBT growth and earnings growth that you can see there, 30 and 37% respectively.
Encouragingly, perhaps what you'd expect with the margin expansion that we've seen, GBP 96 million in Q2. That's the highest net cash generation we've seen in an individual quarter, and now just over GBP 600 million of net cash at the end of June. We've announced an increase in the interim dividend this morning of 26%. Looking a bit more at licensing and royalty. You will remember, no doubt, that my guidance for licensing in April was GBP 75 million ±. At GBP 88 million, we were at the top end of that. It's probably an understatement. Very positive. 25 licenses. As you've heard us say before, we typically build this installed license base at about 100 licenses a year. Completely consistent.
For the second quarter running the take or the contribution from order backlog into the licensed revenue at the top end of the normal range, which is typically 40%-60%, at 60%. Quite a high drawdown from backlog, given the recognition of engineering milestones. As you can see at the bottom there, the backlog in itself still up more than 10% sequentially. Obviously a very strong bookings quarter, which has more than replenished the backlog, notwithstanding a 60% contribution into revenue this quarter. Physical IP licensing, also strong, up 23% year-over-year. Five POP IP licenses signed, which again is encouraging for future contribution from physical IP royalties. Of course, the importance of all this licensing activity and the accelerated build of the installed base is what it means for our royalties going forward.
You can see from the chart in the bottom right, which you've seen before, that this licensing activity, 450 licenses in the last 3 and a half years, barely moving the dial really on royalties. There's a lot of royalties in the can yet to be recognized. In this particular quarter, GBP 119 million, as I say, up 24% year-on-year. 22% outperformance in the quarter. Mostly market share gains. Cortex-A is now 17% of total shipments versus 8% a year ago. Simon talked about the accelerating trajectory of Mali shipments also contributing. In the bottom right, you can see the historical outperformance. We've often talked in these meetings about a sort of 10%-15% outperformance.
I think as was noted in the last analyst presentation by one of the analysts, that that has actually expanded in recent times to be really at the 15%-20% end and slightly north, as we can see this quarter. Encouraging trajectory. Similarly in physical IP, smaller numbers, but a good trend, up 17% in headline level and up 20% in the quarter, excluding audit catch-up royalties from this year and last. Good trends. Looking at costs and tax, just to help with the models. Normalized OpEx in Q2, GBP 78 million. That's pretty much in line with guidance last time and consensus, as you'd expect. Expected to grow a little bit in Q3. We continue in investment mode. We're investing in R&D. We're investing in the back office infrastructure of the company to make sure that this growth opportunity is supported by a resilient infrastructure.
In GBP 79 million, GBP 81 million, broadly in line with current consensus for Q3, I think it was about GBP 80 million. You can see that in the first half of the year, operating margin is up about 4 points year-on-year over the six months. Most of you would have noticed, I'm sure, from the release this morning, there is a fairly significant litigation-related cost sitting in the IFRS numbers. Just a little bit of color on that. Most of you will be aware of patent trolls, non-practicing entities, building portfolios of patents, asserting them around the industry. This is actually in the normal course of business. This is going on kind of all the time. Arm is rarely, and not in this case, party to the litigation, but does provide indemnities under certain conditions to our licensees.
Usually these things close by settlement or cross-licensing, and typically the contribution from Arm and the impact on Arm is really under the hood and goes through our numbers, and you wouldn't really notice it because it's from time to time small to single digit millions ongoing cost of running an IP business and protecting your technology and protecting your ecosystem. In this particular situation where patents were asserted against a number of our licensees towards the end of the second quarter, all of those licensees some together, some individually settled. Arm had a contribution in total of GBP 42 million, which was a combination of some indemnity payments to certain licensees and us buying a license to the whole patent portfolio of this non-practicing entity, which protects the Arm ecosystem. This matter is now closed. That's a full and final settlement.
As I say, these things come up from time to time. I think about five years ago, we had a case related to Nazomi that stuck its head above the reporting parapet. About five years before that, we had one with an outfit called picoTurbo. These will happen from time to time. We don't see anything in view of this magnitude. We should all be aware that the patent troll exists. It is something that we have to manage on an ongoing basis with our partners and with our ecosystem. No change to tax really. You remember the Q1 rate, just under 17%, was kind of artificially low because we recognized the benefit of the U.S. R&D tax credit in Q1, the 2012 tax credit, because the legislation was enacted in January and not as normal in December.
This quarter, more normal, just over 20% but still guiding the full year tax to be just under 20%, which is pretty much where the analyst community are on tax rate. In summary, looking into the second half The backlog's obviously in good shape, record level. The opportunity pipeline for licensing, which obviously we have fairly good visibility into on a three to six months basis, is looking healthy based on our product portfolio, and the new markets that, as Simon explained, our licensees are taking us into. That looks good. License revenue expectations, 25 licenses a quarter, lumpy business. The baseline goes up based on this backlog, so you're probably not that surprised to hear me say that GBP 75 ± moves to GBP 80 ± as I think a realistic base. Could it start with a seven? Yes.
Could it be north of 80? Yes. This looks like a sensible way, because this is trying to look forward in a way, multiple quarters. I'm not just trying to forecast one quarter. That looks to be the right base, I think, to be thinking of at this stage. Looking into royalties, obviously some mixed messages going on around the industry for the second half. Our data suggests that in Q2, which is the relevant period for our Q3 royalties, there is a small sequential increase in overall industry revenues, which is obviously the context for our Q3 royalties. As I said before, OpEx in the range GBP 79, GBP 81 milliion, just growing up, continuing to increase slightly as we keep investing in our people.
Full year, we've had a strong first half, we expect the full year revenues to be at least in line with the current expectations, which are just under GBP 1,080 at GBP 1.08 billion. I think with that, we'll open it to the floor. Thank you.
Ladies and gentlemen, if you have any questions on the phone lines, please press star and one, and wait for your name to be announced. If you wish to cancel the request, please press the hash key.
Questions. First, on the outlook for the second half. Clearly, Tim, you are guiding about GBP 80 million plus or minus, I guess, although you are GBP 88 million in Q2 on licensing. Also for the full year for PD royalties, I think the street is at GBP 520 million. Do you think that it's appropriate to have still GBP 80 million per quarter in the second half, or can that be slightly better than that? On the royalty front, obviously we see some mixed messages from the premium smartphone market, et cetera, sometimes very strong growth in the low priced smartphone market. How do you feel about those numbers? Effectively, would you expect the GBP 1,080 million number to move up? I guess that's my question.
The second question is on the benchmarks. You've given us some interesting data there of A7, A9, big.LITTLE versus Clover Trail+. I guess what people are wondering is where you think you will be versus Silvermont. My final question is on the announcement made by Samsung yesterday of the Exynos 5420 using your graphics solution, which I think is a big surprise to everybody since they launched the first big.LITTLE chip with Imagination only less than six months ago. I'm just trying to understand why you think Samsung has done that. What are the benefits of switching so quickly? Thank you.
We take them in order. I'll do the first one.
Yep.
You do the next one. The appropriate split of responsibility. Yeah. As I said, current consensus going through the results is just under GBP 1,080. You know as well as I do that if I say that licensing is going to be GBP 80 plus or minus, consensus is going to be a little bit north of GBP 80 in both quarters. Clearly, we feel comfortable with that. I think the royalty outlook is much harder to call. The consensus is currently plus GBP 12 million, plus GBP 13 million in Q3, and plus GBP 17 million in Q4. Reasonably chunky uplifts. In the last two years, our Q4 royalties have been up GBP 17 million in each quarter. Not unprecedented, but clearly we'll have to see whether some of this noise around actually manifests itself in lower sales.
What I would expect to happen is the overall consensus to increase broadly by our outperformance on Q2, maybe with some rebalancing between licensing and royalty for Q3 and Q4. That's what I would expect.
Okay.
On your question about Silvermont. You got to remember, there's actually very little data actually out there to say what Silvermont is going to do. From what we've heard, from what we've seen, from the analysis we've done, we're pretty confident that today's solutions around big.LITTLE are going to continue to give Arm a performance and a power efficiency lead. We'll have to see what happens when it comes out. Of course, I think the last few weeks have shown that you have to be very cautious about how much benchmark data you use and where it's come from and how it's been created. A big health warning around benchmarks. What ultimately matters at the end of the day is the user experience. How good is the phone going to sell? It is the combination of lots of parts in there. We think our technology's in good shape.
Which kind of leads quite nicely onto the question about Samsung's choice of Mali in the 5420. You said that it comes as a surprise. That hasn't really come as much of a surprise to us, I actually thought we'd been kind of signaling that for a while. Obviously, you need to ask them about specific reasons for why they choose different devices. They produce a lot of silicon in the year. They're constantly updating their devices. The graphics processors don't have as much stickiness as the CPUs because of the way the code's written.
It is a relatively easy thing to switch from one architecture to another, and back again, of course. The onus is on us to make sure we've got the best technology available to our customers at the time at which they're going to make that decision. It appears for this next generation device from Samsung, we did.
Hi, it's Suman from Redburn. Thanks very much for taking the question. It's essentially three questions. One is to do with the license beat essentially in the Q2. I was just wondering whether that was driven by licensing from the Greater China customer base. If that is so, is that a growing trend or do we expect that to happen going further as well? The second one is to do with the PIPD division. I think congratulations for the first time you've hit profitability on that one. I'm just wondering if there is a sort of operating profit margin guidance or something of that sort. Looking forward, what should we consider should be the sustainable margins coming out of this division? Finally, the third part is on the smartphone chip growth.
Maybe I'm completely mistaken on this. I was wondering, is there a sort of slowdown you're expecting in terms of overall chip growth? I remember 30% CAGR being talked about. If I look through the slide, it's now 20% on Arm's addressable market chip growth. Maybe I'm mistaken on this. Just a bit of color on that would be really
Yeah. Okay. No, the licensing beat wasn't really China. We drew out China in the release because we had our first kind of subscription license there. As you know, subscription license is recognized ratably over time. That doesn't have much impact on short-term revenue. Having said that, China has become, over the last few years, an increasingly important part of our overall licensing activity. The general beat, it was a combination of the terms of licenses yielding revenue in the quarter. Obviously, reasonably material drawdowns based on engineering milestones being hit on the newer technology, releasing revenue into the P&L. Kind of business as usual. 60% contribution from backlogs at the upper end and 25 licenses is probably on average, slightly at the upper end as well. PIPD. Quick question on PIPD. There's two reasons why we have a physical IP business.
One is to drive revenues and profitability in its own right, and the other is to enable us strategically to create better implementations of Arm CPUs. The two are both important, although we're more focused on the strategic benefits of having the business than driving it for profit in its own right. Because obviously, that would lead you to drive the business in very different directions. What we've seen is continued uptake of our POP IP. That's the product line we created specifically to enable our licensees to get better, higher performance, lower power implementations of their Arm CPUs. The licensing from that has been a great success. We've got nearly 50 licensees now, or sorry, licenses of the POPs. We're starting to see the impact of that, and design wins generally for advanced technology flowing through to the royalties.
We do want to drive this as a profitable business. At the same time, the real focus is on the strategic benefits there. We don't have any formal guidance for how we see that growing. As the design wins continue, I would hope that that would go in a positive direction. Your last question was about smartphone growth. I don't think we've changed the numbers there. Gareth.
Thanks. One follow-up on the drivers of license. I just wondered if you could talk about the products driving. Is it 64-bit? Is it Mali-V500? What's the kind of key driver of that? Secondly, I just wondered if you could talk to the royalty rates that you expect for Mali-V500 as you ramp into next year with 4K coming on tablets potentially. Finally, just on the attach rates, you helpfully gave the attach rates for Mali on tablets. I just wondered if you could give a sense in the lower-end smartphone market, what your attach rates for Mali are. Are they higher or lower than the broad range of smartphones? Thank you.
In terms of the technologies driving the licensing, I think it was pretty broad, really. We're seeing Cortex-A, R, and M licensing, and a bit of subscription as well, and Mali. It's been fairly across the board there. There's not one area that particularly stands out. Your second question, I forgot because there were three in there.
The royalty.
The royalties there. We expect to see some incremental royalty on top of the royalty that we get for Mali. You might think of that as another percent-ish on the overall royalty. Your third question was?
Was on attach rates for Mali in smartphones.
Okay. Yeah, is the attach rate higher in the lower end than others? I think a bit early to say on that. I think right now fairly consistent across the entire tiers of the smartphone market. Pradeep.
Just one question, again, on the Silvermont. The supply chain seems to be indicating at this point that Intel is being very aggressive on the pricing of their processors in the fourth quarter. Whenever they're going to come, I guess, some point in the fourth quarter. Would you comment on, you've talked about these 3 tiers. I guess Intel is initially going to play in the tablet market rather than in the smartphone market, given the lack of an integrated product. That where your ASP is at this point, will you be impacted by a very aggressive ASP from Intel associated with their initial Silvermont product?
Well, hard to predict there. Obviously, I don't know what Intel is going to do in terms of pricing, and that price competition is more of an issue for our licensees than it is for us directly. I think in terms of uptake of those products or not, it's going to come down to the handset makers or the tablet makers looking at what they can get from Intel, looking at what they can get from the Arm partnership, and weighing out multiple factors. Software compatibility, performance, power, cost, all of this is going to come into play. Particularly that software ecosystem and the amount of applications written for and optimized around Arm, I think is going to be a big factor in our favor for a long time to come. We'll see how that pans out.
What is your ASP? A couple of quarters ago, I remember it being like GBP 11.5. Where is it today?
For a tablet?
For an application processor using Cortex-A9 or Cortex-A15, wherever all these
Well, I had in the chart there some different price points for 5, 10, GBP 15 for the different tiers. Beyond that's the kind of broad generalization of the pricing that we see in those tiers, and exactly the spot price is what our customers said. Francois?
Thank you. It's Francois from Morgan Stanley. I've got a few questions. Maybe the first one is I'm trying to understand the guidance. I'm trying to read between the lines. Basically as Didier was saying, you feel more confident about licensing, and maybe you think that royalties should be a bit lower. Is that because you see some inventory correction in the high end of the smartphone market, or is it because you think that some people in the sell side got ahead of themselves in terms of numbers, and maybe the 2014 royalties are a bit too high as well? That's my first question.
All we're saying right now is that, if you look at the consensus of royalties in the second half, it's plus GBP 13 million in Q3 and plus GBP 17 million in Q4. Those are seasonal shapes that we've seen before. There are some mixed messages around the industry. It's hard to put your hand up and say that's in the bag right now, because it isn't. We're not making any comment on the longer-term royalty trajectories. We're not making any comment on 2014 royalties. As you can see from the charts, our performance in royalties versus the industry as a whole has been on an improving trend, and the gap has been expanding. We see nothing fundamental that changes that dynamic.
We're just pointing to the fact that the sequential increases that are in the market for the second half, which are, as I say, pretty close to normal seasonality are not in the bag as of July 24.
Okay, thank you. The second question is about the licensing, which is, as you explained at the beginning, moving more and more to Asia and maybe a tiny bit to mainland China. I think for us to be difficult because, I grew up with TI, Qualcomm, and these type of companies, and now it's maybe names that I'm not as familiar with today. Of course, we know about MediaTek, Spreadtrum, maybe a new big one is RDA. What I'd like to understand is, as we are moving from Western chip players to Asian type players, how do you see this having an impact on the pricing? I think we all believe that if you have more customers in Asia, then it could have more pricing pressure down the road for all the industry.
I think there's been a gradual shift in our business towards Asia generally over about the last decade. This is not a new phenomena. It's been going on for a while. You'll note that in that time, our license revenue on a quarterly basis has gone up. These two things, if you correlate them, would not suggest that we're under increasing price pressure because of that shift to Asia markets.
Mm-hmm. Okay. Another question would be, I'm sorry, about Intel, and I hope I'm not bugging you too much on this, but I've been very surprised myself to see the new MacBook Air from Intel, which is not even Silvermont, it's a big chip as well. Of course, it's much more expensive than yours, but still getting a very good battery life for the MacBook Air. It's 13 hours. If you basically compare on the same battery size basis as the iPad, basically it got 11 hours browsing time versus nine hours for the iPad. I'm really surprised because my belief was like, well, Arm is so much better, Intel can't catch up. Where do you think the progress have been made, and where do you think you can basically leapfrog them again going forward?
The chip that's in the MacBook Air is a very different chip to the kind that you put in a phone or in a tablet, and it's built on a different kind of manufacturing process to that you'd use for a phone or a tablet. MacBook Air, great product. I've got to say, my own experience of using one isn't quite what the claimed battery life statistics are, but it is still very good at the end of the day. I think the technologies that we've been delivering on, big.LITTLE, very energy efficient graphics, now with video accelerators as well. The approach we're taking to system design, the approach our partners take to SoC-level power control is going to enable Arm-based gaming solutions that ideally target tablets and phones.
Okay.
Gentleman behind you.
Thank you.
Jerome BNP Paribas. Can you update us on big.LITTLE and Cortex-A15 adoption beyond Samsung and potentially MediaTek at the end of this year? When do you expect meaningful unit shipments of Cortex-A15 big.LITTLE?
There are a number of customers who have licensed the Cortex-A15 and Cortex-A7 big.LITTLE combination. Exactly when they're going to ship is hard to say, but I would expect to see that volumes will grow through next year. Gentleman next to you.
Thanks. Morning. It's Nick James from Numis. A couple of questions. One was on the benchmarks. You've given us a health warning on benchmarks, then you've drawn attention to your own benchmarks. I guess I'm a little bit surprised you've used big.LITTLE there. There's one implementation of big.LITTLE on the market, which there have been reported overheating issues. There have been criticisms of the size of the chip. If you could just address those in the context of the benchmarks that you set out there. Then the second question was on the exceptional from patent litigation, we had a big investment last year, which was basically stopping there being a patent troll for the MIT's patent to protect Arm. We've got this unexpected thing come through in Q2. I know you don't foresee more of these, I don't know that you foresaw this one.
What is the assurance that we're not going to see more and more of this type of big patent litigation type of cash outflows?
Okay. I'll do big.LITTLE and then the other. In terms of big.LITTLE adoption, as I said, there's a number of customers who are designing products there. Yeah, you do have to take a health warning with benchmarks. You do have to look at what is being compared. Obviously, what we've shown on the chart there is a synthesis of what we think is important when looking at comparing one chip against another. There's a lot of data behind that that we haven't bored you with here today. In terms of big.LITTLE implementations, what we provide to our customers is a very flexible set of components that can be built in big clusters, smaller clusters, mix and match. Depending on the level of performance that one of our licensees want to produce, they can use that technology in however they see fit.
You're going to get some larger implementations. You're going to get some smaller implementations. It's all about what we provide is flexibility and choice in how our customers put their chip together, and how they best think they're going to target the market, and that's what Samsung have done.
Those benchmarks reflect shipping Arm chips because that's kind of what you said, correct?
Yes.
So which-
They're real chips shipping. They're real phones. Samsung Galaxy S4. There's a slide on in the appendix with details.
Thank you.
Questions at the back.
Yeah, on patents. As I said in the remarks there, the existence of non-practicing entities and patent trolls and the reality that sometimes in this space, operating companies fail, and investors want to monetize what's left in the form of patents. This is the reality of the business. It's been going on for a long time. Clearly our long-term interest is to protect Arm and the Arm ecosystem from the disruption, irrespective of the merit of these claims regarding infringement or not. They require time and money to deal with. They usually end up, as I say, in settlements that you don't notice because the numbers are so small. From time to time, you will get a patent portfolio that requires more to close down.
As I say, this is the first one that we've seen, and that's kind of material to Arm to draw out in this way for many years. Typically, when these things do emerge, you get quite a long sight on them because actually being a patent troll is not all easy money. It takes a long time to get the money. Usually we have good visibility if there's anything in the pipeline, right? When we say we can't see anything material in view, that's based on what we can see now. Something can emerge, and it could end up in a quick settlement that the Arm ecosystem, the Arm partners, and Arm consider to be in the best interest of our business and for those of us that are public, our shareholders.
I can't guarantee that the landscape is clean, but I'm not expecting to be talking about this type of thing in the foreseeable future.
Can I just follow up on one of the other exceptional charges for Linaro, which I think was GBP seven million in this quarter, it was GBP seven million last year. As I understand it's a not-for-profit organization, but it's not for profit, i.e., funded by Arm.
Why isn't this an expense of the Arm business, given it seems to be ongoing?
Well, it is an expense of the Arm business, but the reason we draw it out separately is because it only occurs every two years. Last time it was two years ago, in fact, not one year ago. Every two years, based on the current membership and the current, if you like, price list, Arm will be spending
GBP 7 million over 2 years. Given that that gets paid in one lump and recognized in one lump, it is obviously a cost of ours. It's charged to the P&L, but we draw it out separately because it doesn't appear in all the other quarters. We'd sit here explaining why next quarter it wasn't there. It's just easier to have it transparent.
Great. Thank you.
Simon?
Thanks so much. Simon. I mean, structurally, how big is this opportunity for you in terms of machine to machine? Most of us have heard Ericsson and a lot of the telco infrastructure guys talking about this opportunity more broadly, whether it's GBP 30 billion, GBP 50 billion, I guess no one really knows, but it's certainly a big opportunity. Structurally, given that there's still 130 licensees, I think you called out that will start to ship. What's the structural runway to growth in this market? And then more importantly, going forward, is that still going to be a very low price point or eventually is there a chance that these guys require higher value technology from you and your partners?
In terms of that market size, as you say, it's GBP billions, GBP tens of billions. Exactly how big, a bit hard to say. We can think up new applications for the Internet of Things all day long, and hopefully they're going to come to fruition. I think initially you're going to see quite closed systems where some data is collected, it's processed. Eventually, I would expect you're going to see much more open systems where data collected from different places can be made accessible to different people to process in different ways. I think that's where this starts to become very interesting. I think those devices that the end node where the sensor is going to be a very low-cost device forever. In fact, I think the tens of billions of units relies on it being a very low-cost device.
What that's going to do though is create an awful lot of data that's going to need aggregating, maybe processing locally, along the way, and then up into the cloud. I think it's going to drive local aggregation points, I think it's going to drive network infrastructure, it's going to drive the expansion of the cloud, which are all, of course, higher cost ASP devices. That's how I think this is going to pan out. Of course, we are all crystal ball gazing here slightly. When you look at the kind of applications people are already starting to build, it's not a huge leap of faith to see some of this coming together.
The gentleman next to Simon.
Achal Sultania from Credit Suisse. Just to follow up on the Mali business. Obviously we've seen very strong volume growth. I'm just trying to understand what part of that growth has come from the smartphone market and what part is coming from the tablet market. Also just trying to get an understanding of, obviously we're seeing higher core GPUs being implemented in some of the, for example, the Exynos chip. What does that mean for your royalty attach rates for Mali specifically going forward?
In terms of growth, Mali has pretty good market share in smartphones, in tablets. We're in about half of all Android tablets. We're in, I believe, about 20% of Android smartphones, and about 70% of digital TVs. These markets all have different growth trajectories, but we have pretty good established market share there. I'm sorry, the second half of your question was?
It's about the implementation of higher core graphics in your GPU. What does that mean? Actually, as you move towards higher core GPUs, what does that mean for your royalty attach rates for Mali?
Okay. In terms of the royalty from those chips, graphics cores tend to be fairly large in comparison with CPUs. We design ours to be as small as possible, graphics processing does take a lot of silicon area. If you put a large number of cores down, the silicon device tends to get larger, and therefore the silicon die costs more, and so that would drive higher royalties into Arm.
Does that mean that there's a potential for royalty rates to increase?
There is a potential for royalty rates to increase with more use of cores, yeah.
In percentage terms.
In percentage terms, yeah. We're going to have to wrap up fairly shortly, so maybe we can make this the last question.
How much time have we got? I think unfortunately we've got one more question.
It's Janardan Menon from Liberum Capital. A few short questions. One is going to your licensing trajectory. You've always said that licensing is a precursor of royalties to come. Historically it has been true. Today, given that you're getting so many licenses being signed in markets like China by a large number of new entrants all trying to get into the same market, the similar market, which is tablets and smartphones to a large extent. Does that mean that the royalty rate, there's a potential for royalty rates to increase?
At the end of the day, from a power consumption point of view. The last question is, the industry is quite dependent on Qualcomm for LTE modems, and increasingly so. They are the de facto supplier. To what extent do you see the fact that they want an alternative, and Intel provides that alternative because they are one of the leading suppliers of 3G modems today, especially to Samsung and other players. Would that edge market share towards them just because someone like Samsung wants the LTE modem, someone apart from Qualcomm, Intel is the best alternative?
I do China.
Okay.
I don't expect actually to have a material difference in the conversion of China licensees to royalty from the rest of the field, really. As Simon said, we have been licensing companies in China now for a long time, and actually a number of the leading players are now reasonably significant contributors to royalty. By the same token, there are one or two who don't make it or get bought or disappear in the same way as-- When we show our charts of number of licensees and that, remember, those are licensees who are still in the game. They're not licensees that we've ever licensed. These are licensees who are expected to generate royalties. Okay? Our experience so far in China is that the conversion rate has been pretty good. As I say, we've now got a number of significant contributors to royalty from China.
Your question on big.LITTLE and Cortex-A15. When we're looking at the performance graph of those benchmarks, that is A15 powered up. I would expect to see devices which just use A15 and don't use the big.LITTLE combination. There are going to be some applications that don't need to either switch like that or don't have the benefits of varying work rates. The thing about phones and tablets is sometimes you need a lot of performance, sometimes you don't. There are other applications, like networking, for example, where it's about constant throughput. You don't have that variability, so big.LITTLE doesn't help you there. Which is why we've designed all our processors with power efficiency in mind. It's just there are degrees. I think A15 specs up well on its own, and I think it would be used on its own, independent of big.LITTLE. Your last question was about LTE.
I think generally, people building things or anyone wants choice of suppliers. More people providing LTE is not a bad thing. Intel's solution does actually have Arm technology in it. That's certainly not the end of the world for us if they do start to ship some volume. Okay. Well, I think we better wrap it up there, and thank you all for joining us today.