Arm Holdings - Earnings Call - Q2 2011
July 26, 2011
Transcript
Warren East (CEO)
Okay, it's 9:30 A.M., so we will get started. Thank you all very much for coming along to our half-year results presentation. Here's an obligatory notice. What we're going to do, as usual, is I will start with a business update. Tim will go into our results, our financial results, to add some color, and then we'll deal with questions and answers. The high-level structure of this business update, those of you who've heard me talk before will know that we're very enthusiastic about Arm's business model and the unique differentiators in our business model, in our excellent technology, and our ecosystem. I'm going to go over those points to begin with, then look at the business in a bit more detail for the relevant period, and then step back and look at some drivers of growth for the future.
If we look at the business and the business model, this is a great set of results. It's a very strong quarter. It's our fifth consecutive record revenue quarter. It's our eighth consecutive quarter of growth. We're very pleased with the overall picture. I think an important message is that major semiconductor companies are still investing in Arm technology. If you look at the first half of the year, we've sold as many licenses in the first half of the year as we would have been happy selling in a complete year a couple of years ago. This is because existing semiconductor partners are continuing to license new Arm technology. Interestingly, in this current quarter, we're still seeing new companies coming to Arm for the first time. This is in the context of Arm's range of target applications really broadening out.
We're getting into the space where companies who haven't seen the need to use Arm Holdings technology before are starting to license Arm technology. That's great. In terms of volume shipments from our semiconductor partners, we're continuing to outperform the industry. I'll come on and talk on that in a bit more detail. We're particularly pleased with growth in microcontrollers, smart cards, and connectivity. If you look at this whole business model itself, we are continuing to gain a presence in the semiconductor industry with our outsourcing model. This applies beyond the microprocessors, in the physical IP part of our business, and in the graphics part of our business as well. All of that is generating increased profits and very good cash generation. You'll see that in a bit more detail later as well. The business and the business model is in very good shape.
The technology itself is also in good shape. We're continuing to license at the high end of our roadmap, so two additional Cortex-A15 licenses this quarter. The Cortex-A15 is our state-of-the-art microprocessor core that's available at the moment. We've just released Cortex-A15 this last quarter to the lead partners who'd licensed it during the development. Cortex-A15 is now starting to roll out to the semiconductor industry. Cortex-A15 is the first core that's really enabling people to think more with some of the newer products that we have in development. Our physical IP technology is also in great shape. We're very pleased with the processor optimization packages, still selling very well. We sold an additional four processor optimization packages during the period for Cortex-A9 at 40 nm. I'll come back to talk about Mali licensing, where we're seeing Mali being adopted for smartphones and digital TVs as well.
Technology is also in good shape. The ecosystem is an essential part of the Arm business model. We contend that we have an excellent ecosystem. This was a quarter in which various Arm Holdings partners achieved for themselves key milestones. We had Google announcing their Chrome OS for Arm. Microsoft created some stir by demonstrating Windows 8 on Arm, running Office applications, and running Internet Explorer. These big companies in the Arm Holdings ecosystem scored good results during the quarter. We also saw the ecosystem developing around Mali, Mali being used in smartphones and digital TVs, starting to ship in volumes. This is encouraging a developing ecosystem around Mali. You start to get the sort of inertia effect of volumes creating a demand for ecosystem development. That ecosystem development then fuels more design wins, fueling more volume.
We're pleased to see several providers in the graphics space making progress with developing their products around Mali, notably a games engine developer, Unity, who started to develop their engine around Arm. That is great news. I'll come back to talk a little bit more about that in the future. Now let's go to the business in a bit more detail. This is a slide you've seen several times before, but hopefully we can see the progression now. First half licensing revenue, very strong. Q2 licensing revenue, its highest ever. This has enabled us to sell. We've sold 68 licenses during the first half of the year. We're now over 800 licenses in total. When you peel back the detail on that licensing, very healthy licensing of the Cortex products, Cortex A9, six licenses, Cortex-A15, a couple, and some of the new products as well.
At the other end of the processing scale in the microcontroller space, we're also seeing very strong licensing for Cortex-M, 12 licenses in the quarter. We're now over 100 Cortex-M licenses in total, and that bodes very well for microcontroller shipments in future years. Licensing of Mali graphics technology continues. It's solid execution in that part of the business, another five licenses. We could have a step chart for Mali, a bit like the one for processor licenses in total. We're now up to over 50 Mali licenses with those that are added during Q2. We normally talk about the split of licenses and the first intended use that our semiconductor partners have. The picture that has been emerging for the last couple of years or so continues in the quarter that we're reporting today. 20 out of the 29 licenses sold in the quarter were for non-mobile.
That's a very consistent picture with what's been happening over the last several years, and that's very encouraging for our non-mobile shipments as we look forward. Talking about shipments as we look forward then, future royalties, the chart on the bottom right is the sort of extension of the picture that we've been building for some time. A reminder on this slide of some of the timescales involved here: typically, licenses that we sign today, it will be at least four years, more like five, really, for appreciable volumes of consumers going into shops and buying products incorporating that technology, which means our royalties come to fruition.
You can see that in the chart on the bottom right, where if we look at the dark blue color at the top, you can see licenses sold in the period 2006-2011, some 400 licenses now, actually starting to make a difference after about five years. We can see that starting to make a difference. I joked with Ian that maybe this is the last time you see this chart in its current form because I think maybe come February, when we produce it again, we'll have to come up with a fourth color because we'll be doing licenses from 2010 onwards. You will see the 2006-2010 slipping back into the middle as it starts to drive, as those licenses start to drive an appreciable proportion of our total volume. You can see the volume being driven by the cumulative licensing there.
In terms of the chart in the bottom left, and I believe there's a sort of full-size version of that in the appendix, this is a chart we introduced in February. It is basically highlighting the licensing that we are doing to equip the semiconductor companies who service the target markets in which we want to win a significant market share, like 80% or beyond. A reminder of the way the chart's decoded, we highlight with the little dots the number of semiconductor companies who supply semiconductors to those OEMs to effectively take us to about 80% market share. When there's a blue dot, that means they're equipped and shipping. If there's a red dot, it's the other end of the spectrum.
The green and the amber dots are different stages in the process where partners have the technology, but they're not designing with it, have the technology, have designs, but not yet shipping, and so on. There is continued progress there to grow our market share, continued progress in the necessary licensing. In terms of what's actually happening with royalties and the royalty numbers coming through, the first half of the year was very encouraging for us. We saw Arm royalties grow at about 3x the rate of the underlying industry. That is a continuation of the trend shown in the chart on the left-hand side of the screen, where the growth line of Arm's royalties, there's always a gap between that and the underlying industry, indicating our growth in market shares.
On the right of the slide is a little of the detail in second quarter about where all those devices went. We're pleased with the record volumes, up 30% year-on-year, nearly 2 billion units in the quarter. It doesn't seem so very long ago that we were high-fiving 2 billion units in the year as a whole. That is very encouraging for us to see. Mobile continues to be very strong, obviously, but I think the highlight of different application areas is certainly MCUs. On the back of the Cortex-M licensing that has been happening over the last several years, let's say over 100 Cortex-M licenses now, we saw yet another quarter of very strong growth in Arm microcontroller shipments. That is really very encouraging for us.
For those aficionados that follow the detail of the pie chart, this is the first time that non-mobile has actually been over 40%. We've reached a bit of a milestone there in the mobile/non-mobile split. Switching to the physical IP side of our business, this has been a story of continued execution this quarter. The business is in a very healthy state at the moment, licensing and royalty up consistently year-on -ear, 19% and 11% respectively. We're particularly encouraged to see continued execution about selling platforms of physical IP products, absolutely leading-edge technologies, at 20 nm technologies, but also some of our older technologies for those customers who are building products that typically run on much older geometries. We're still selling new physical IP into that sector of the market. Cumulatively, the number of physical IP platforms out there continues to grow, encouraging at all ends of the spectrum.
The processor optimization packages, that's also very encouraging for us. The chart on the bottom right shows the sort of steady execution progress quarter-by-quarter in processor optimization packages. In the quarter that we're talking about, we sold six Cortex-A9s. We sold four Cortex-A9 processor optimization packages. That is a metric which continues to perform well, the effective penetration of processor optimization packages into our Cortex-A9 marketplace. We're pleased with the progress on the pots. That was a quick run around the business. I'll now just briefly touch on drivers for the future. At a high level, our business is subject to several long-term trends. There are long-term trends about the semiconductor industry driving outsourcing. There are long-term trends about energy, and this is driving demand for our low-power technology. One other long-term trend that Arm's particularly exposed to is the increasing presence of the internet.
This was highlighted in a recent report that we're quoting here. This report that Cisco published highlights some really interesting facts. You only have to go back a few years to 2003 to find about one internet connection per 20 people. Around about now, there are almost two internet connections for every person on the planet. As we look forward, that number explodes. In fact, if you look at the numbers, then as we look back over the last seven years, the number of internet connections has grown at a compound annual rate of nearly 60%. As we look forward, we're anticipating the number of internet connections to grow at a compound rate of about 30%.
I think we're nearly in the middle of this huge trend for internet-connected devices as we move from a stage where you had to make a deliberate effort to go and use an internet connection to, "I've got an internet connection on me somewhere," to, "It's just all happening around me, and it's happening without me noticing." That's the world of smart meters, of connected medical devices, machine-to-machine communication, and so on. This is a fantastic opportunity for us because to implement all that connectivity, you need something like an Arm microprocessor to do it, to run the protocols that do it. That's why we're very excited about this long-term trend. It isn't just about smartphones and tablets. It's about a huge range of different products. Just looking back at the last quarter, here are some of the products that bring that sort of vision to reality.
Yes, we've got the high-visibility ones. Things like the Samsung Galaxy S2 phone, a stunning smartphone of the quarter, powered with a dual-core A9 and quad-Mali processor. The benchmarks, the graphics benchmarks for this phone, are absolutely outstanding. That phone is turning heads across the industry. We're very pleased with that. Then there's a range of the typical smartphone, tablets, and digital TV-type devices and a reminder that in our sort of market analysis slides, which you'll find in the appendix, those types of products sum to a total opportunity of about $4 billion in 2015. If you look behind those products, what's driving some of the connectivity, what's driving some of the storage, then you see the middle line, 11 billion units in 2015. The embedded machine-to-machine type world that I was talking about a moment or two ago, that's an even more significant volume by 2015.
Here are some of the pictures of where Arm's being designed in as we speak or products that were released over the last quarter. In terms of long-term trends driving our growth over the next 5 years -10 years or so, I think we're in very, very good shape. Summarizing where we are for Q2, we've got very strong license revenues. What's really encouraging about that is it's a reflection of the design activity that's out there around leading companies continuing to make and making for the first time long-term commitments to the Arm technology. It's in the first half of the year, a record number of licenses by some considerable margin. That's very encouraging. Of course, it is driving very strong license revenue at the moment. We're sitting on top of a backlog at near record levels with a very healthy pipeline of opportunities.
At the other end of the sort of production story, shipments of Arm-based products are going very well. They're continuing to outperform the industry, as I say, about 3x industry growth as we look over the last 12 months. That's very encouraging as we drive Arm into not only the market segments, which are growing structurally and outgrowing the market, but as we grow Arm into new applications. As we do that, of course, we're continuing to invest in new technology. We're growing the number of design engineers we have, which is increasing our capability to design more and more of this technology at the same time. At the same time, the business model is enabling us to make that investment along with increasing our margins, increasing our cash generation. At the bottom half of this slide is a bit of an advert. We've gone all commercial.
This is an advert for our technology conference, which happens in October in Santa Clara. Last year, we introduced an investor event at that technology conference. This year, we're doing the investor event again. There's an email there because Goldman Sachs are helping us with organizing that investor event. There are a few places still available. Do see Ian if you're interested in coming to that event. With that, I'll hand over to Tim. Thank you.
Tim Score (CFO)
Morning, everybody. There are quite a few numbers in the earnings release and lots of detailed numbers in the slide deck that's on the web and actually quite a few numbers in Warren's presentation. The good news is I've got two slides, three including this one. I'm going to focus mostly on outlook and guidance and give some color on that and help you with that if there's any need for that. This first slide of the highlights Warren has largely touched on is clearly this quarter is exceptional for the license performance. $58 million is a record by some margin. Although 29 processor licenses is not a record, it is running at, as Warren said, about twice what we might have considered to be a reasonable quarterly average two or three years ago.
The installed base of licensees, which are, let's face it, in place to generate future royalties going forward, is growing at a faster pace than ever, which is very encouraging. Also encouraging at the end of Q2 is notwithstanding the high billings in the quarter, the backlog sort of flat sequentially, and obviously still at historically high levels, and the opportunity pipeline healthy for the second half. I'll sort of touch on that a little bit more on the next slide when we talk about what that means for looking into the second half. Royalty momentum remains. Warren's touched on it. I draw it out there. Remember, we had a $9 million catch-up in Q2 2010 last year. The headline growth is suppressed by that. The absolute underlying growth is 33%. Still moving in a very strong direction.
Headcount, we are investing, as Warren said, mostly in new technology, mostly in our R&D capacity, developing new technology, up just under 100 people in the first half. We're still in investment mode. We will continue to hire new people through the second half. Mainly, that underlying move is what drives our normalized operating expenses, which were $59.4 million in Q2, a smidgen ahead of where the consensus was, but in the range we guided, guiding to $60 million-$62 million for the third quarter. Notwithstanding that investment, we're reporting today record margins, over 44%. That's somewhere we've not been before. The operating leverage is coming through as the revenues grow. We've announced today an interim dividend of 20%, consistent with our ongoing policy of progressive dividend. Those are the sort of highlights of the numbers. Looking into the second half, very strong licensing looks set to continue.
The Arm addressable market is, as you know, broadening at both ends, at the microcontroller end and at the mobile computing end and beyond. That is bringing quite a number of new licensees to the Arm story, as well as encouraging our existing licensees to re-equip with our latest technology. All of this, as I said, is underpinning share gains out in future in our target end markets. Warren showed the picture of what that means in terms of attracting the key semiconductor players. For the second half, we go in there with a healthy backlog and a healthy pipeline. I'm conscious that I think it's about six months ago I stood up here and said that we could move the range of likely license revenue outcomes from $35 million-$40 million-$40 million-$45 million.
I'm conscious that since we said that, we've done two quarters in the $50 million range. You guys and girls don't actually quite believe me because the consensus for Q3 and Q4 is $47 million and $48 million, so beyond the range already. I think it's fair to say that based on the backlog and based on that pipeline, we would expect to see upside against that current consensus in licensing in the second half. That's encouraging. We also say in the release that if you look at this pipeline of opportunities, the outlook for the backlog itself in the second half is promising as well. Not just the revenues for Q3 and Q4, but also for the exit backlog. Obviously, that's much harder to forecast in detail because it's very dependent on timing of deals and specifically the bigger deals around the end of the year.
The prognosis does look good for licensing and backlog development. I think, as we've said in our release, the prognosis for royalty against the current consensus, which is about $89 million in Q3 and about $102 million in Q4, is less clear based on the industry data that we have today. Our Q3 royalties are based on Q2 shipments. The data that we read about Q2 shipments appears to suggest sort of low single digit, maybe edging to mid-single digit in some forecasts for Q2. That is obviously the context in which we should think about our Q3 royalties.
When we look at Q4, where typically, certainly in the last three years, we've seen a $10 million sequential uptick, part of the normal seasonality, it's less clear to us and less obvious at this stage, given the macro uncertainty and the potential impact on consumer confidence, that we're going to see that type of uplift. Current consensus is for a, bless you, $13 million uplift in Q4, which will be higher than we've ever seen, which at the moment seems to us to be fairly aggressive in the context of the macro environment. I think the net of looking at licensing and looking at royalties is that what we're actually saying today is, notwithstanding the fact that we've outperformed in Q2 by about $15 million, you can essentially bank that and then look at your second half.
We're suggesting that the right thing to do at this stage is to hold it because there looks like there's an upside on licensing. It looks like there's some risk on royalty, given the industry data we have. The net of all that is that we are happy with the second half. What that means in number terms is the consensus coming into these results is $745 million for the full year. I mean, if you bank the second two overperformance and hold the second half, you're obviously in $760 million territory. In many ways, that is more positive guidance or more specifically positive guidance that we've given in the past, where we've tended to say at least in line with expectations, but we've rolled in the current quarter outperformance.
This time, we banked the current quarter outperformance and then confirmed expectations, which in many ways is more positive than saying at least in line with something that includes a current quarter overage. That's kind of how we're looking at the second half: positive for licensing, positive momentum for royalty, but some uncertain macro indicators out there, and it's just too early to say. If there's any further need of that, I'm sure we'll touch it on the Q&A. With that, I think we'll move straight on to questions.
Warren East (CEO)
I think maybe I just got his hand up first.
Didier Scemama (Analyst)
All right. Morning. Didier hear from RBS? Two quick questions. You've answered the third one that was on the licensing outlook for the second half, Tim, so thanks. I'm just looking at your microcontroller business up 100% in the context of a market of 10%. Presumably, some of that is a bit of overbuild based on the commentaries that we've seen from some of your licensees, but still a phenomenal performance. Do you think that looking at the profit warnings that we've seen from companies that have either not used Arm or not used Arm too much, they are also suffering to some degree beyond the macro environment and OTO's weakness and so on. Do you think they are suffering also from the fact that Arm is taking market share? That would be my first question. The second part of the question is related to OpEx in Q4.
Tim, you've alluded that you're going to continue to invest in R&D. Should we expect OpEx in Q4 to be up versus Q3? I mean, taking into account maybe grads, recruitment, and this kind of stuff.
Warren East (CEO)
Okay. On the microcontroller one, the short answer is yes, we are taking market share. This is an architecture issue. We are seeing demand from the end customer base who want their traditional microcontroller vendors to offer them Arm-based solutions. If you go back 12 months or so ago, you saw Freescale, a very strong microcontroller company, launching a handful of ColdFire, their own architecture microcontrollers, alongside 200 Arm microcontroller parts. I think absolutely companies are responding to demand from the engineers in the design community who want the Arm architecture.
Didier Scemama (Analyst)
I completely understand the caution you're putting on royalties in the second half. Given the backdrop of the substantial outperformance that you have relative to the market and cycle, perhaps you are a bit too cautious even on your royalties outlook in the second half.
Warren East (CEO)
As Tim just said, if you look at the second quarter shipments from the semiconductor companies, our data points are pointing to a few percent of sequential increase. Typically, we have a $10 million for the last few years uplift. All we're saying is that wherever we look right now, there's macroeconomic uncertainty. We're not just exposed to microcontrollers. We are exposed to consumer products that are purchased by consumers. There is uncertainty about consumer expenditure. Therefore, maybe that $10 million uplift is going to be somewhat muted, or there is a risk that it's going to be muted. That's why we're sort of being a little bit cautious about it.
Tim Score (CFO)
Yeah. I mean, in terms of OpEx, I would expect a small sequential increase because the headcount will be increasing gradually through the second half.
Warren East (CEO)
Yep, I think we've got one. Never mind, he'll get there.
Brad Simpson (Analyst)
Thanks. It's Brad Simpson of Harrity. I just had a quick question on the licensing. Tim, you suggested it'd be some questions about extra clarity. Last quarter, you talked about a normalized quarterly licensing figure of about $45 million going forward. You just printed $58 million, which is great. Looking ahead, you've said two-thirds of the intake is non-mobile for licensing. You're suggesting the pipeline's pretty good for orders. Why wouldn't we see licensing revenues in the sort of $60 million for quarterly run rate going forward?
Tim Score (CFO)
I think what I've said is 40-45 is what I said six months ago. What I see in the market is 47-48. We've just done a 51 and a 58. We're painting a picture of a high backlog and a robust pipeline. I don't think that necessarily translates into 60s at this stage. Licensing is inherently lumpy. The good news for us is it's being lumpy on top of a range. I think, again, we kind of need to be cautious. We're signing some very substantial deals with some very substantial players. If they occur on a Monday, not a Friday, our quarterly results can look very different. I just don't think it's right to be too bullish. You've probably never heard us be as positive about a licensing outlook at a quarterly presentation than we've just been.
Precisely how that pans out into the numbers, we will see.
Brad Simpson (Analyst)
Did backlog actually rise in the quarter?
Tim Score (CFO)
No, it's flat.
No. Flat.
Flat sequential.
Brad Simpson (Analyst)
How much of the revenues you booked in the quarter was licensing you actually booked in the quarter rather than from backlog?
Tim Score (CFO)
About 50%.
Brad Simpson (Analyst)
Thanks.
Warren East (CEO)
Thank you, François. Next.
Thank you, Warren. It's François from Morgan Stanley. The first question is on AMD. I think if I remember the call last quarter, you were feeling quite bullish about AMD potentially transitioning to Arm. I don't think we've seen anything so far. Obviously, you can't pre-announce anything, but what's your feeling on this quarter? The next question would be on the A15 license you've signed with, I guess, a new customer for servers. Obviously, we are all very eager to hear from Arm at some point. To 64-bits and multi-threading flavor of the A15 or whatever it's going to be called, A17 or whatever. Shall we expect the design cycles to have shrunk? Because I think from the A9-A15 or from the, yeah, it was something like two years and a half, you've announced A15 last year.
Shall we expect something at the end of this year or next year, or will it take longer?
Right. AMD first. AMD was hitting the headlines when we did our quarterly presentation last time. We were asked questions about AMD, and we said that we've been in ongoing discussions with AMD about microprocessors for the last 10 years or more. We haven't persuaded them, but discussions continue. Over the last quarter, they're in the business of selling microprocessors. We're in the business of selling microprocessor designs. It's our job to try and sell microprocessor designs to companies building microprocessors. We talked to AMD. We've got nothing specific to announce on AMD this morning. On servers and Cortex-A15 and what comes beyond Cortex-A15 and so on, I'm sorry, we will announce new technologies when we choose to announce them and not at earnings presentations.
We know that in order to fully target the server market, there are some technologies that aren't in the current Arm Holdings portfolio, things like 64-bit. We absolutely anticipate being able to target the full spectrum of server applications in due course. We'll announce the appropriate technologies at the right time.
Is it more a question of time or more a question of budget allocation for these 64-bits? It takes a long time anyway.
It does take a long time to develop these new microprocessors. We started developing Cortex-A15 quite a long time ago now. Typically, it's sort of somewhere between two and four years to develop a new microprocessor, a totally new microprocessor rather than a derivative. That's been ongoing as far as we have future technologies that we haven't talked about publicly in the works. We've been growing, as Tim said, we've been growing our headcount mainly in R&D or development of our microprocessors. We grew by nearly 200 people last year. We grew by 100 in the first half of this year and will grow by some more in the second half. I don't think it's a question of budget. Honestly, there's so many hands gone up, I have no idea who is next.
Sam Hudson (Financial Analyst)
Thanks. It's Sam from Redburn. Just a quick question on your slide on royalties from licensing. If I remember from Q2, you were showing about 770 licenses sold up to now and showing royalties from it. This time, you're showing 800, whereas you've signed about 68 licenses. I'm just wondering if there are some license orders or licenses which you have kind of removed as not generating royalty. That's my first question.
Warren East (CEO)
When we refer to a number of licenses out there, we are actually talking about licenses that we believe will generate royalties in the future. There is always a little bit of correction of the total as companies disappear, as some licenses effectively wither on the vine. That's updated continuously. If you follow the arithmetic in absolute detail from quarter to quarter, you will see some dropping out. We're now at over 800 licenses.
Sam Hudson (Financial Analyst)
Just a second question is on the physical IP, the POPs solution. I'm just wondering, what is the current attach rate you have with sort of Cortex-A9 or Cortex-A15 sort of licenses? Is that a good way to look at it? If you could just give us a bit of an idea on the licensing value and sort of additional royalties you would be able to achieve from that.
Warren East (CEO)
Okay. In terms of attach rate, I think at the last set of results, we talked about having, certainly if we look back over the POPs, about 50% of the number of Cortex-A9 licenses, there's a POPs as well. In the quarter just reported, we sold six Cortex-A9 licenses. We sold four POPs. It's a bit better than 50%. I would caution the law of small numbers. 50% is probably quite a good number. We happen to do a bit better in the current quarter. In terms of royalty generating potential, then physical IP royalty on a chip will be roughly half the typical royalty that we would earn on a microprocessor. I say roughly because it's calculated on a different basis. It's calculated as a proportion of wafer cost rather than individual die costs. It depends on die area yield and all those sorts of things.
Roughly half of microprocessor royalty. If there's a POP and a Cortex-A9, you can roughly take your expected royalty and multiply it by one and a half.
Sam Hudson (Financial Analyst)
Your physical IP is much broader than just your POP solution. If a particular chip was on POP and physical IP, are we talking about a larger number than that?
Warren East (CEO)
No, no. POP is essentially a packaging exercise, a product marketing exercise. POP is a predetermined collection of Arm physical IP that gives a deterministic result for implementing one of our microprocessors. From a royalty point of view, it's simply saying if a microprocessor is implemented with a POP, it means Arm physical IP is present on the chip, and therefore, a physical IP royalty is earned on the chip.
Thanks.
I'll let the microphone just sort of move back to wherever.
Gareth Jenkins (Head of Research)
Thanks, Gareth Jenkins UBS. Just two quick ones if I could, or maybe not quick. On your earlier slide, you showed a picture of the Motorola Atrix, which I think is a smartphone laptop, also the ASUS headphone, so smartphone tablet. Do you think there's a danger that you end up with some cannibalization where you end up with one process unit driving multiple devices going forwards? Secondly, in terms of Mali and the ecosystem, can you tell us how many Mali licenses are now shipping and whether you think your previous comment around tens of millions of Mali shipments towards the end of this year as a run rate, there's upside to that given where we see ourselves today?
Warren East (CEO)
Okay. I think we show pictures on the slides of interesting products. There are some interesting hybrid-type products out there at the moment as the world grapples with the notion of computers and smartphones and what's a good form factor for consumers. Frankly, we always say we're not experts in end product design. We enable these companies with the microprocessing technology. The form factors that are going to prove to be successful will be the ones that prove to be successful. As long as we enable as many of them as possible, we'll be in good shape to benefit. I'm sure we're going to see some of these products come and go. That's not a particular comment on the two that you mentioned. It simply means that from a royalty point of view, we're delighted to be in those products.
We're not really sort of pinning massive hopes on particular form factors being successful. Your other comment was about, or question was about Mali and the number of people shipping and my confidence in the tens of millions of units shipping this year. I remain confident in tens of millions of units shipping this year. I think we've seen great success with things like the Galaxy S2. On its own, I don't think that's sufficient to cause me to want to sort of up the estimate. I think we'll stick with tens of millions of units this year. As a reminder, what's good about those tens of millions of units is that it drives ecosystem development. We see ecosystem, the maturity of the Mali ecosystem as being the most significant retardant on growth of our Mali presence.
Seeing tens of millions of units shipped this year is good for our ecosystem, which is good for future volumes of Mali.
Nick, Nicholas up from RBC. Warren, you've been quite helpful in the past talking about the percentage of royalty per chip based on the chip price going forward. You've indicated that could run up to around 3% as you look forward a couple of years. Could you just update us on your thought process around there with the latest licenses you're signing in terms of how you think that could develop?
There is the question referred to a discussion we had back in February about royalty rates per chip theoretically reaching about 3% of the chip price with, as a reminder, an inflated royalty rate for higher-end microprocessor cores such as Cortex-A9 and Cortex-A15 being in the 1.5%-2% range rather than typically around 1%. With an addition of 1% for a Mali graphics processor, with an addition of about 0.5% for physical IP, you take all that together and you get about 3%. There is no change to that picture. We've continued to sell Cortex-A9s and we've got a couple of Cortex-A15s as well. The royalty rates that we're achieving are consistent with what we said before, i.e., they're a higher royalty rate than the more traditional levels that we've seen for Arm 7s and Arm 9s. There isn't really any change.
I would just point out one number in the release where we talk about the volumes of Cortex-A products now being 4% of our total volumes. Although we get very excited about these higher royalty rates on the Cortex-A products, remember, only 4% of the 1.9 billion units were actually Cortex-A products in the last quarter. When you're looking for evidence of this effect in our royalty numbers, it's still a very, very small effect.
To be granular, if you were to take a current product which has the highest Arm royalty on it, what would that number be?
I wouldn't sort of hazard a guess because we supply across such a range. There are chips out there that are Arm-powered that are selling for hundreds of dollars a go. Under some license agreements, there are sort of royalty caps on multi-hundred dollar chips and that sort of thing. It's just not a question I can answer.
Jonathan Plant (Markets Analyst)
Hi, it's Jonathan from Liberum Capital. Just trying to dig in a bit deeper into the microcontroller growth rates, the 100% that you reported. Presumably, there are two factors operating there. One is replacement of 8-bit and 16-bit microcontrollers by 32-bit microcontrollers, which leads you to grow faster. The second is your customers like the ST, the Atmel, who have moved from proprietary cores to Arm cores. That latter factor, I'm just trying to get a fix on that. What percentage of those customers' cores do you think are today shipping with Arm? I'm saying if it's 20%, it's still a long way to go. If it's 80%, then there's less migration within that customer base. I'm just trying to get an idea for the latter part rather than the former part. Second question is your confidence on licensing into the second half of the year.
Is that coming from the traditional large customers where you have presumably more visibility and therefore lends to more confidence, including on the backlog, which you said will be sort of at least flat at the end of the year? Is it coming from the sort of spread of Arm across a large number of small companies in new product areas? Which is the bigger weighting in the licensing confidence for the second half of the year?
Tim Score (CFO)
I think that one first.
Jonathan Plant (Markets Analyst)
Yeah.
Tim Score (CFO)
I mean, the answer to your second question, Janardan, is really both. The confidence about second half licensing is partly based on the entry backlog into the second half. Obviously, we have a view of how that backlog is likely to mature into revenue given engineering deliverables, etc. That's one factor, which as you've heard today, typically can account for about 50% of target license revenue. That's important. When you look into the detail of the opportunity pipeline, it really is a combination of re-equipping existing licensees, but also a size of opportunity pipeline with other licensees who may be looking at Arm for the first time. It's a volume thing as well as a value and existing licensees thing.
Warren East (CEO)
On the microcontrollers, you highlighted two factors: the replacement of 8-bit architectures and the replacement of in-house 32-bit architectures. Primarily, it's a replacement of 8-bit architectures that's driving the growth at the moment. We do have agreements in place with companies like Atmel and companies like Freescale where they have in-house 32-bit. I honestly wouldn't know off the top of my head exactly the sort of, if I look at Atmel's shipments, for instance, what proportion is Arm-based, what proportion is their in-house AVR based. I think it is quite low, the Arm penetration at the moment amongst these companies. I don't see it shifting very rapidly. It's much more of a sort of steady progress as end customers demand variants from these people based on the Arm architecture. There are certain applications where some of these players are very, very established.
It's a much easier win for Arm to displace an 8-bit solution than it is for Arm to go and displace those established 32-bit processors. I think we've got a long way to go on the second factor.
Thanks. Simon Chappell of Global Science. I just had a question about operating profitability. There used to be this 40% number out there as a target. Given licenses, there's so much better in royalties. I'm prone to predominantly to see that. Going forward, royalty is actually growing faster. I guess my question is, what are you managing the business towards now? Are you just hoping to grow OpEx significantly enough to keep that 45% run rate, or is it exceeding your expectations? How should we think about an underlying level of normalized margin?
Tim Score (CFO)
I mean, we don't specifically manage the business for margin. If you look at this model, the way of developing and creating value is actually by designing technology that is ever more broadly applicable to end markets. That is going to drive margins, profitability, and cash. I think it was about 2007 when we were about 30% that I said that this business can sustain 40% and more in the medium term. Everyone said, what's the medium term? I said two or three years, X cycles, and that sort of thing. We're there now. My comments really are, if you look out five years plus, I expect the margin to continue to increase. The pace at which it increases will depend obviously on things like cycles and exchange rates.
It will also depend crucially on our investment opportunities at any given time to create the next wave of licensing and therefore royalty. It may be in five years' time that this is a more valuable business with a 50% margin than it would be with a 55% or 60% margin in terms of its longer-term prospects. That is kind of how we look at it. We're looking at how can we complement this business model. How can we make sure that we achieve market share penetration in a sense as rapidly as possible? Because we know if we achieve that, the margin will flow. This gentleman at the front has been very patient.
Sandeep Deshpande (Stock Analyst)
Yeah, hi.
Warren East (CEO)
Not as patient as this one over here, but he's next. Sorry, Sandy.
Sandeep Deshpande (Stock Analyst)
Sandeep Deshpande here at JPMorgan Chase. A couple of questions. Firstly, on the back again to the licensing revenues, exiting the year, you could do almost $200 million worth of licensing. The base will be very high for the next year. Can we understand in terms of the licensing revenue that you are doing for the year, how much of it is truly, breaking it out between what were your existing old customers and this is new business which you have generated so that the base itself has risen as opposed to essentially that you're selling to the same customer, maybe higher value product, etc.? That's one question. The second question I have is on the way you have shown your key operating numbers in the current quarter. You have taken off some costs associated with Linaro, I think.
I don't understand why we should be taking off costs associated with Linaro when it is an ongoing expense. Finally, with regard to royalty growth, clearly into the next year, you're going to see some of this associated with Windows 8 on Arm, etc. Can you help us? It's actually a more question associated with how you sign your licenses. You mentioned it a minute ago, but do you have actually fixed royalties on any devices or is it always a percentage? Thanks.
Warren East (CEO)
Okay. Shall I do the new customers and the royalties?
Sandeep Deshpande (Stock Analyst)
Sure.
Warren East (CEO)
The first question was about breaking down the license space and breaking it down between new customers, new applications, and sort of re-equipping semiconductor companies in the traditional space. It is actually very difficult for us to break this out because you get a customer who is an established Arm semiconductor partner with a track record in phones and smartphones and then tablets. They'll use the same license to build a product to perhaps target servers. How do we deal with that? It's a very difficult sort of breakdown to do. What I can point you at is if you look back quarter by quarter, we always report the number of licenses sold, the number sold to new people buying an Arm Holdings license for the first time. Of the 29 sold in the last quarter, I think we had 12 new people. Was it nine? Nine new people.
20 licenses were sold to existing customers, nine to new customers. That's probably quite a high proportion. In a normal quarter, it might be more like 20 and 5, that sort of proportion. As we target new applications, do we always have percentage royalty or do we have some sort of fixed royalties? When you're only signing 29 licenses in a quarter, each one is essentially a little bit unique and negotiated in a unique way. We always take account of the target end markets that the semiconductor company is going after. Most of those license deals, however, end up being a percentage. The royalty is a percentage. We do have a few fixed rates out there. It's probably becoming an increasingly small proportion of the total which have fixed royalty rates. Most of them really are percentage.
Tim Score (CFO)
On Linaro, just a reminder, this is an organization that Arm and a number of its licensees set up about a year ago to develop optimized software and tools for Linux-based products. The reason we've presented it in the way we have is that $6.8 million that you see there is our contribution to Linaro for the next two years. Therefore, there will be no other profit and loss entry for Linaro for another two years. In a sense, for transparency, it seemed easier to treat it consistently with how we treated it in Q2 2010 and Q3 2010 when we showed it as a separate column. That's why it is the way it is because you can argue about the normalized, but the fact is it isn't going to be there for the next two years.
Sandeep Deshpande (Stock Analyst)
Wouldn't it be more fair to have it just every quarter because it is an ongoing expense?
Tim Score (CFO)
The situation is that the accounting requirement when you're committed to the contribution is that you have to expense it there and then. It's $6.9 million. It's transparent. There'll be nothing else for two years. In a sense, how you evaluate the headline number, the data is there for you to analyze. As I say, because we had to account for it in one go, we've shown it separately.
Just a quick one on Mali, actually. You've sold, I think, 20 Mali licenses in the last three quarters, but I don't think you've sold any to new customers. Is that something that concerns you or investors should be concerned about at all?
Warren East (CEO)
Yeah. When we say new customers, we mean people who don't have an Arm license at all. If I look at the five Mali licenses sold in the last quarter, for instance, I think there were two of those five to people who didn't have Mali before, and three were existing Mali licensees upgrading their Mali license. They all happened to be Arm customers.
Okay. Thanks.
Kai Niklas Walter (Senior Analyst)
Thank you. It's Kai Niklas Walter of Deutsche Bank. Just a quick question on the mobile royalties, which I think are roughly 60% of PD. It looks like the trend in the handset market is there's exploding growth in sort of low-end 2G sort of China Shanzhai market type devices, which probably would have a lower royalty opportunity for Arm Holdings. Obviously, the smartphone tablet type devices are growing fast. The feature phone segment where Arm Holdings does have exposure to is sort of falling away. I'm just wondering if you've ever done or looked at the split of that 60%, whether we could get a stab at what roughly the exposure by segment is so we could maybe understand the future trajectory of that part of the royalties, please. Thank you.
Warren East (CEO)
Yeah. If you look in the appendix, you'll find our analysis of opportunities for 2015. There, you can see what we expect the number of smartphones, the number of feature phones, the number of low-end voice phones to be. I'm not quite sure what page it is because I have a printed version. Maybe Ian can help us with the page number. Slide 20. Yeah, slide 20 in the appendix. That's where we break down what we think the market is going to break down in 2015. When we update these numbers every year, we show you what we think the snapshot has been over the last 12 months. We'll give you our updated version of the five years out view. It does change. Our sources are the same sort of industry analysts that you look at. The world changes.
We also use some of our bottoms-up intelligence of communication with the handset manufacturers as well to generate these numbers. I'm just conscious of the fact that we have about five minutes left before we have to disappear. We've only got a few questions.
Andrew Gardiner (Financial Analyst)
Andrew Gardiner with Barclays. Just first, on the royalty side, you've cautioned us about being too aggressive on the Cortex-A royalty rates and factoring that through to revenue by saying that it's only 4% of units currently. To put it into context, can you help us understand what % of revenue it may be at the moment? Also, on the PIPD POPs, the roughly 50% attach rate that you're seeing, what are the kind of customers that are going for those? Are those perhaps smaller licensees that may not have the full design capabilities in-house? Are you also seeing interest from some of the larger semiconductor players? Thanks.
Warren East (CEO)
Yeah. The first one is estimates of Cortex-A in terms of royalty revenue as well as volume. We tend to not talk about that in public. The reason we don't talk about it in public is when you put the two pieces of information together, or rather when our customers put the two pieces of information together, they can reverse engineer exactly what royalty rates their competitors are getting. It all becomes a bit commercially sensitive. I'm afraid I'm going to leave that as an exercise for the analyst community. What we do say is that these Cortex-A9, they're going into high-end smartphone chips. You can get a read of what high-end smartphone chips sell for. We're talking $15+ for an apps processor with something like that, maybe $15-$35. We do say that typically, the royalty rate is greater than 1%.
Typically, you're in the north of 1.5%. We give you some basic information, but we don't talk about the answer. I'm sorry on that one. I'm afraid I forgot the second question.
Andrew Gardiner (Financial Analyst)
POPs.
Warren East (CEO)
Oh, POPs. Yeah, people buy POPs for two reasons. One, if they don't have the design capability in-house. You're right, that tends to be smaller companies, and so POPs do enable smaller companies to build competitive implementations of things like Cortex-A9. We're also selling POPs to, looking at the last quarter, some more established players. Unfortunately, they don't want us to be public about it. More established players who do have the capability would want to use a POP because it gives them a deterministic result, and it's a time-to-market to a different openness to the result. That's why they buy it. Yep, I think we're still going.
Tim Score (CFO)
Is there anyone who hasn't asked one yet who wants to?
Warren East (CEO)
No, I think the answer's no to that.
Okay.
Andrew Gardiner (Financial Analyst)
Just a quick follow-up on Cortex, on Cortex-A. Warren, can you give us a sense for the portion of your licensing so far at Cortex-A that might be licensed to non-traditional semiconductor players? Players that haven't really shipped silicon traditionally. When you look at the business, you've typically up to now licensed cores and physical IP. Is there any scope to increase that to license a full application processor, like a white label platform for non-traditional semi guys to license? Thanks.
Warren East (CEO)
Yeah. We have sold Cortex-A9s to some people whom you would regard as, you know, non-traditional. I would say it's a very small proportion right now, so probably around about 10%, something like that. The other question was more interesting. It was about, you know, would we consider extending licensing to license a sort of white label platform? The answer is that if we look back over the years, then we started off licensing, you know, a very bare microprocessor core. Over the 20 years that we've been doing this, we're now licensing much more of a microprocessor subsystem than we were. As the subsystem around the microprocessor becomes more inextricably linked to the performance that you get out of it, the semiconductor companies are comfortable with Arm doing that. They want Arm to provide that, and they don't see lots of that as such a differentiator.
They seek to differentiate in other parts of the system on chip device. If you project that forward, there may be a time when that level rises all the way to the apps processor. We're nowhere near that time at the moment. Our business is dependent on being able to sell next generations of products on re-equipping lots of our semiconductor partners. The last thing we want to do is do their job for them because we don't have the expertise that they have. The real competition is other architectures, and we want to sustain our business. We're not going to cannibalize their business. Long term, there is a trend in that direction. I have no idea how far that trend will go. We continue to evolve our business model as the world outside changes.
We need to do it carefully so that we maintain the advantages of real expertise differentiating in the right places to ensure that Arm-based system on chip devices are competitive. I think we have, let's make these the last two questions.
Tim Score (CFO)
No pressure. Better be good.
Let me think. I think Tudor, they throw an interesting 40% market in portable computing in 2015, if I remember correctly. If I do the calculation at the end of Q2, the Arm architecture in portable computing as defined by Arm netbooks plus notebooks plus tablets is at about 23% with obviously zero or close to zero in netbooks and notebooks. Can you maybe share with us what was the assumption on the market share in netbooks and notebooks to get to that 40% market share? That'd be great.
Warren East (CEO)
Actually, I can't. Tudor was quoting an analyst. That was based on an external analyst, and it was indeed based on a proportion of tablets versus a proportion of netbooks. I have no idea what the assumption was behind there. Disappointing answer to the penultimate question, I'm afraid.
This question hopefully is a bit more specific. You can give me an answer. It's on the Patent Box Tax. This is something which obviously the Labour Government, I think, introduced as an idea of reducing the tax rate from about 24%, 28% down to 10% in the U.K.. Now, if I understand it correctly, it's effectively on revenues earned from patents. Considering yours is an IP business, I'm assuming that you will benefit from this tax rate reduction. What is your view on when this could come in and how much could your overall tax rate of effectively 20% go down to from this perspective?
Tim Score (CFO)
Our overall tax rate forecast for the year in this release is 26%, which is obviously a combination of your rates around the world. We already have a U.K. rate that's signposted to go down to 23% in two or three years' time. We would expect the patent box regime, should it be implemented, and all the expectations are that it is actually going to come into law to benefit companies like Arm. We are involved in that process to try and make sure that happens. There's lots of detailed implementation about precisely which of your revenues and patents, et cetera. I think if it comes in, the overall message is Arm's medium to long-term tax rate is going to be considerably lower than it is right now.
Warren East (CEO)
A much better answer to the last question. Thank you very much, everybody, for coming along. We'll be back in October with Q3.