Good morning, everybody. Thank you very much for coming. I looked at the number of people in the room here, I said, "Oh, it's quite depleted." Tim's getting with the semiconductor vernacular now, he said, "Well, it's only partially depleted." There we are. Good morning, and welcome to our results obligatory statements. We're going to start off, I'm going to give an update on the business, a romp through of what we've been up to in 2012 and in Q4, in particular. A little bit of a longer-term view towards the end, Tim will follow up with numbers in the usual way. Let's start off with the highlights for Q4. Q4 was a fantastic finish to 2012. We saw a continued momentum in licensing. 36 licenses in the last quarter.
That's another year of over 100 licenses in the full year. At the start of the fourth quarter, we held our TechCon event in Santa Clara, at that event, and around that event, we launched our V8 products, the Cortex-A50 series. Indeed in Q4, we saw some more licensing of those V8 products. They're launched, we're sort of relaxing a little bit on the lead partner licensing as we get closer to silicon on those devices. That was a good start to the quarter. I'll come back to some of the detail of licensing a bit later. In terms of how shipments performed in the industry at large, we saw a continuation of Arm outperforming the semiconductor industry.
We've got a few slides or a few pictures in the slides later showing this gap of outperformance is increasing as Arm is targeting a wider range of end applications. It was particularly pleasing to see when we step back and look at the year as a whole, growth in Cortex-A and Mali shipments, and we've got some statistics on that. We normally talk about the outsourcing model and the increasing traction there, it was another quarter of good licensing for Mali and for Processor Optimization packages. All of that came together and enabled us to deliver increases in revenue profits. The normal pie chart that we put up there, you can see the split of revenues, revenue's up 21%, profit's up 16%, and Tim will talk about the numbers there. I'm going to step back a little and look backwards five years.
One of the key drivers for what's been going on with Arm's business over recent years has been the PC era, and more importantly, the sort of beginning of the end of the PC era, such that now we're talking in our statement this morning about the post-PC era. We thought it worthwhile having a little look from a sort of longer-term perspective. If you go back to 2007, there were about 400 million devices shipped in the year that connected to the internet. Most of those, a good two-thirds of those, were effectively PCs, and the Arm-based smartphones really were only a small portion of that. There are a number of devices with other architectures as well. A few years ago, 2007 is conveniently five years, it's also the year of the launch of the iPhone.
It sort of generally started the smartphone wave. It is instructive to think about the applications. At that stage, there were thousands, measured perhaps in the tens of thousands but only just, thousands of third-party software applications running on the PC, which provided a very powerful ecosystem around those products. Fast-forward to today, we will call it the post-PC era or the beginning of the post-PC era, you can see the number of devices connected to the internet has grown by a factor of four. There is now 1.6 billion devices connected to the internet. That is, as many of you have heard me say, it is from the palm of your hand, sort of two-inch screens, four-inch screens, right up to 84-inch screens in digital TVs. These are all internet-connected screens.
When we look at the architecture sat behind there, you can see that the Arm architecture is sat behind about three-quarters of this massively increased volume. PCs are still there, of course, people still buy PCs, people will want to continue buying PCs going forwards as well. What we mean is not that the PC is finished, but actually, digital products are now much more than just PCs. It is instructive to look at the applications and the strength of the ecosystem because we have moved from a situation of thousands, even low tens of thousands of apps, to now millions of apps that are available. Of course, these millions of apps have been downloaded billions of times. Some 1.4 million apps between iOS and Android available at the moment, about 40 billion downloads.
The huge growth from 400 million to 1.6 billion has really been driven by smartphones, tablets, and an increasing diversity of form factors. What is going to happen going forwards? It is a very competitive market, who knows what is going to happen to those market shares. What we would say is that this is all opportunity for Arm architecture going forwards as the 1.6 billion devices grows by a further factor of about two and a half to, we think, about 4 billion internet-connected screens over the next five years. Obviously, we are not making any particular comments here about the strength of the ecosystem. Clearly, people are going to be building on those 1.4 million applications, software developers will be free to choose the platforms that they choose.
That is the sort of context around PCs, we know that PCs, tablets, smartphones, smart TVs, is a very important area. It is a very competitive space, it is a very important area for our business over the next five years, thought it was worth that slightly longer term perspective. Back to the business, we will look first at royalty revenue, processor royalty revenue, just under half of Arm's total revenue. For Q4, that was up about just a bit over 20% year-on-year. The industry for this particular period, marginally down year-on-year.
That outperformance of the industry, 24%, shown on the chart on the right. We're breaking that down into approximately 11% market share gains, approximately 7% due to the customers choosing to adopt Mali alongside our Cortex-A products, and about 6% due to the higher value of the Cortex-A products in the first place. That's effectively what happened in Q4. It's instructed to note that that sort of roughly 20% stood for quarters throughout the year. What did that mean in total? Q4, Arm's business usually has a little seasonal component, and we quite often see an uptick in Q4. We saw that uptick, we said at the half year it's likely to be a bit more muted than it has been in the past.
It was a little bit more muted than it has been in the past, but that still produced two and a half billion units in Q4, making a total of 8.7 billion chips for the year as a whole. When we look at the outperformance on royalty for the whole year, that comes out at 19%. As I say, it's about 20%. The chart on the bottom right is a chart that you've seen many times before, and we continue to extend that to show the continuity. Arm is not impervious to the cyclical market here. When the industry shipments fall, Arm shipments fall as well, but not by as much. Rather, when industry growth declines, Arm growth is held back a little. That outperformance is actually increasing over the years.
It's moved from a sort of 10%-15% outperformance to more like a sort of 15%-20% outperformance. What are the sorts of products that are driving that? These are the sorts of products that we see. The reason for this slide is twofold. One is to show that this huge range of products, the dotted lines refer to some kind of connectivity. It doesn't have to be screens which are connected to the internet. Indeed, as we look forward and contemplate the Internet of Things, then the vast volume of devices probably won't have screens. I visited CES a few weeks ago, and I've been going to CES for the last several years. If I think about CES in 2010, 2011, then it was very much about the bottom right-hand corner of this slide.
Consumer-oriented products, a great place to go, collection of nerds, and so on. What we saw in 2010, 2011 was a few microprocessor controlled light switches and security devices and so on. Quite often you could look at these and you could tell that, "Hey, there's an eight-bit microprocessor in here" to give somebody a sort of push button interface instead of a rotary interface and that sort of thing. Not really very smart. 2013 was a different CES. For me, CES 2013 was all about this whole slide. Clearly, it's sort of been a gradual process from one place to the other. It hasn't been a sudden switch in 2013. We really are seeing a plethora of healthcare products, metering products aimed at energy efficiency, home security products, and so on.
You can tell these devices are much smarter devices because they are connected to the internet, either directly or indirectly via some form of mobile phone type hub. That connectivity is enabling people to talk about services which apply on top. Obviously, the commercial potential in those services is what drives the volume adoption of the end product. That is clearly going to be a driver for Arm royalty going forward. It has clearly been sort of part of the story in 2012, 2013. In 2012, we said it is 8.7 billion chips. I am not going to go through this slide. This slide has a few pictures, examples of some of the new products in 2012.
Really this slide is an advert for the tabular slides which we have included in the appendices at the back of the book and will appear on our roadshow slides and will appear in the investor relations section of our website. As has become the habit over recent years, Jonathan and Ian, and the team have updated the numbers, which includes looking at revised numbers from Gartner and other analysts. We have put three slides in. We have restated 2011 slightly. In particular, there is some updates in the networking section and the embedded section, so that the total volume available for us in 2011, we now think was a little bit more than we thought a year ago. That means that our market share is a little bit less than we thought it was a year ago. We are restating that down to 29% for 2011.
On the same basis, in 2012, our market share has increased from 29% to 32%. When you actually do look at the slides in the appendix, you will see that is a continuation of the theme for the last several years, where market share increases by between 2% and 4% per annum. 2012 was no different in that respect from any of the others. We would expect that our market share should continue to increase based on the design wins that we are seeing, but recognizing that it is a competitive world, and just because things have increased by 2% to 4% over recent years, we have to make our own judgments about the rate of that increase in market share going forwards. The slides are there. They are updated.
We are also showing a slight increase in the volume when we look out five years, and we are now expecting 2017 total volumes to stretch to about 41 billion units. The other interesting observation about that slide in the appendix is that the feature phone sector, we are saying effectively feature phones will turn into low-end phones and low-end smartphones. There is a zero in that line now. That is royalties. Let us switch to licensing. Licensing is really what creates the opportunity for future royalties. The quarter saw nine new licensees, which is something which has been happening as we have been targeting a wider range of applications. We are bringing more companies into the Arm partnership. We now have over 320 Arm partners. Just under half of them are shipping products these days. It was a good quarter.
We saw, as I mentioned at the start, increased licensing of our V8 technology. We're very much at the start of the wave of licensing of V8. Embedded in the 36 license agreements, there were 15 Cortex-A licenses, which is good to note. Along with those Cortex-A licenses, another two companies adopted big.LITTLE. We now have 16 people on big.LITTLE. We saw licensing of Mali graphics processors for phones, tablets, and TVs. That's the story on licensing. Continuation of this sort of increasing the number of licenses out there creating opportunity for future royalty. Another year of over 100 licenses in the chart at the bottom. The chart on the right-hand side is again a chart with which you're familiar.
You can see that the latest licenses from the last several years are just over 400, actually, but I was going to say nearly half of the installed license base. It is a little bit less than half. Really not contributing a huge amount in terms of royalty revenue so far. That's clearly a great potential, and of course, many of those licenses, the more recent licenses, have been for the higher value products, the Cortex-A type products, and 15 of them in the last quarter. Some of the drivers behind licensing are, of course, the new markets. It's worth just sort of commenting here. As you know, our processor roadmap has several different sections. There's the applications processor, the Cortex-A products, and the Mali products. These really are about screens.
Screens of all different sizes, they're starting to be about servers and some networking equipment as well. We anticipate apps processor market opportunity in 2017 about 4 billion units. At the bottom of the slide, we talk about the Cortex-M family. These are the microcontrollers, and these are embedded industrial, medical, security, and so on. This is the Internet of Things. Screens are not obligatory in this space. The value of some of the chips is obviously much lower than for application processors. The volume is significantly higher. We're anticipating a market opportunity there of over 20 billion units in a few years' time. The devices which either have screens or the Internet of Things devices are connected to the internet. Increasingly, the volumes of data which are used and generated by these devices is putting demand on the infrastructure.
The servers and the network equipment that connects things are taking too much energy. That is stimulating demand for people to use energy efficient technology. I'm not saying Arm is the one and only answer to this, but Arm is one of the tools in the kit bag for these equipment providers to drive down the energy consumption of the infrastructure and server products. That, amongst other things, is driving demand for our Cortex-R series of products. We anticipate between 10 and 15 billion units of market opportunity in that space in a few years' time. It's those significant opportunities which are attractive to our semiconductor partners, which is what's driving our license business at the moment and driving interest, and we talk about a healthy pipeline going forwards. Let's switch to physical IP.
It was a year of great execution for our physical IP business. Physical IP is definitely now, you can see in our Processor Optimization Pack chart on the top right-hand side. Now, it's inextricably linked to our processors. Many of our partners are choosing to use Arm physical IP to create deterministic solutions so that they know what they're going to get out of an Arm microprocessor implementation. The arrows in the graphic on the bottom right-hand side of the slide, when you look at that in detail, show that using appropriate optimization packages people can achieve implementations which achieve higher performance in smaller area and therefore cheaper devices, and more power efficient devices as well. That applies to our microprocessors, the red and the blue part, but also to our graphics processors as well. We introduced a graphics processor optimization package in 2012 as well.
If I look at the underlying physical IP technology that we licensed to the foundries on the left-hand side of the slide, it was another quarter where we did some more platform licensing. The encouraging thing here is that Arm physical IP is now absolutely at the leading edge of semiconductor process technology. The new platform licenses sold during the quarter were engaged on 14 nanometer work with the leading foundries. If you look at where this stuff is starting to generate royalty, then we saw our first 28 nanometer royalties as well during this quarter. Clearly, Arm physical IP is being adopted more because you can see our underlying physical IP royalty outgrowing the foundry growth as well. 17% plays 7%. That's it for physical IP, and that's it for a brief run around the business as in 2012.
I think it's worth because some of the costs that we're bearing in 2012 and some of the investments that we're making in 2013 is actually about a longer term story. This is about addressing the sort of challenges that face our business in the medium term. Really, we are investing in R&D. You can see us building the headcount. This is because the technology that we're working with is getting more complex. It's getting much harder to implement. We're facing more uncertainty in those technology developments. Rather than simply face that uncertainty, we're investing in people, and we're investing in tools to get around some of those challenges. When we deliver on those challenges, then consumers and businesses are able to benefit because they have even more capable devices that are even more power efficient.
One of the other reasons we need to invest in technology development is because it's a competitive environment out there. I've just painted a picture showing fantastic market opportunity. Obviously, that's attractive to everybody, not just Arm. We absolutely need to make these investments to ensure that our technology is thoroughly competitive. We've always invested in partnerships. Arm tends to work with a few leading edge companies whom we describe as the thought leaders. What you do is work closely with these thought leaders. They've got their eyes on the future, the pulse on what's actually happening. Then we can apply that learning to the fast followers. Now, the interesting thing is, of course, that thought leaders aren't thought leaders forever. Industries change and evolve, we do need to invest more than just in the thought leaders.
That is why we will spend time, we will spend money, and one of the reasons we are increasing the headcount is not just in R&D, it's also investing in partnerships with some of the fast followers, some of the players who are in enabling technology that sit alongside these companies to enrich our ecosystem. There is some investment there. You also saw us in 2012 invest in some patent risk mitigation. The MIPS patent portfolio is an example of this, where we acquired some rights to the MIPS patent portfolio. Obviously, MIPS are in the business of microprocessor licensing. That patent portfolio was potentially, particularly a risk to Arm's business. We needed to invest in that, in acquiring those rights. As our technology becomes more prevalent, clearly we are exposed to a wider range of technologies, therefore a wider range of other people's patents.
That's why we need to continue to invest in patent risk mitigation. Of course, with an eye on the future, people often ask us in one-on-ones, "How do you guard against disruptive technology? How do you know when somebody isn't designing something which is going to undermine your future microprocessor business?" Of course, the real answer is we don't. The right answer is that even though you don't, you spend a heck of a lot of time and effort worrying about who those people might be, keeping tabs on universities, investing in the appropriate startups, interesting enabling technologies that probably half the people in the room haven't heard of. That we need to do because they might be technologies that help us at some stage in future. You see us with seed funding different technology companies.
As we're exploring new businesses, you see us apply the classic Arm partnership approach, working with consortia. Things like the Linaro consortium, moving out from just mobile into the enterprise space. You see us do some acquisitions. You also saw us launch a joint venture in the security space. There are some future challenges, and that's what really lies behind some of the investments that we make today and why today's cost doesn't necessarily match up with today's business. I've gone on for long enough. The summary at this point is the business is in very good shape. Our license revenues are continuing to be strong, and it's really new product development, a competitive environment out there that is stimulating our partners to invest in new Arm technology. That trend continued during 2012. It's set to continue in 2013.
Shipments of Arm-based chips, the results of the labors of those semiconductor partners and Arm, that's continuing to outperform the industry, so we're continuing to invest in the ecosystem to enable that. It's very encouraging to see the extension of the outsourcing model in our graphics business and our physical IP business starting to generate additional royalty revenue. The value that we get when somebody ships one of these chips actually is starting to increase, and we saw that as a trend in 2012. As I just said, we're continuing to invest in new technology. The good news is we're able to do that at the same time as increasing our revenues and profits. Tim's going to talk about that now in some more detail. Thank you.
Thanks, Warren. Morning, everyone. I think Warren's given a fairly comprehensive overview of the quarter, you'd probably be relieved to know I've got four slides which are not going to repeat everything he said, but it's probably just going to give you a little bit of guidance on the models and a little bit more insight into the Q4 P&L. We've talked about the revenue overall up 21% in the quarter, year-over-year. Very strong in licensing, 26%. Strong in royalty, 17%, Warren's talked about the relativity to the market. We know that it's Cortex-A shipments and Mali helping to drive that.
I'm however, going to focus more on the, if you like, the trajectory of a 21% revenue increase becoming a 16% PBT increase, becoming a 10% earnings increase, because important to understand that in terms of when we look forward into Q1 and the rest of 2013. The underlying costs in Q4, as we said in the release, were about GBP 73 million. There were two issues in there that increased it up to GBP 79.7 million as reported. One of those is the mark-to-market charge, which Arm followers will be aware, tends to move around in the ±GBP 2 million, depending on what the quarter end exchange rates are. For those eagle-eyed on December 31, the dollar blew out to 162.50 for sort of a one day only offer, then came back to the 160 and below when people came back to work.
Very transient, but actually had an impact on our mark-to-market. There was a small charge there. The more important one was related really to our Q4 performance, both in terms of the overall corporate bonus that is paid to all of our employees on one year revenue and profit targets. Because the Q4 performance was much stronger than we had expected by the end of Q3. We were obviously accruing a bonus, expecting a slightly lower result. When we got the Q4 result, we needed to true up the bonus accrual that we've been making through the year to cater for that result. That was part of the uplift, obviously that goes away. The other one was sales commission payments to our sales force.
You've seen in the release that the order backlog was up 25% sequentially, in a quarter where licensed revenue was as high as it's been. That tells you that the bookings that were in the quarter was very high. Therefore, sales commissions again were higher than the normal ongoing level. A combination of those two factors meant that we did GBP 80 million of pre-tax profit. In a sense, those won't recur. The corporate bonus would only be at that elevated level in 2013 if we shot through the budget again. You can expect if we did our budget, the bonus next year will be lower than we've seen in 2012. If we shoot through our budget, which probably would be a good outcome for us all, the bonuses will be higher.
Certainly, bonus payments in Q1, Q2, you can expect to be lower than we've seen in Q4. The other issue we had in Q4 was the tax rate. I've been messaging a sort of 25% normalized tax rate for 2012, and that's what it was in the nine months to the end of September. There are a couple of issues in the U.S. One is a short-term timing difference, which was the timing of the legislation of the R&D tax credits in the U.S. You can only realize the benefits of the U.S. R&D tax credit when the legislation is passed. In all previous years, the legislation has been passed in Q4. This time it was passed on the 2nd of January.
Slightly annoying because I'm having to stand here explaining to you why that benefit gets taken in our Q1 numbers, not in our Q4 numbers as normal. That was one issue. There's another piece of law in California that was passed in November, which changed the way that companies who operate in California, how their income gets taxed in California. The good news is the change in methodology means that less of our income going forward is going to be taxed in California. The short-term impact of that is we have some deferred tax assets that we need to, as it says in the release, de-recognize because we had been assuming higher in California going forward. Because we're going to be paying less, not all of those deferred tax assets will be recovered in the fullness of time. We needed to write them down.
Again, a one-off issue. Good news, we're going to be paying less tax in California going forward. Putting all that aside, that Q4 noise on tax, as we know, the Patent Box regime that we've talked about before, we talked about at the Analyst Day, comes into effect in April. We get 60% of the benefit of that in year one and 10% per year in the subsequent four years. For Arm, a lot of our profits are earned from qualifying patents as defined by that regime. Therefore, as of now, my guidance is that the full year effective normalized rate will be about 20% in 2013. That's the kind of Q4, why the P&L is the way it is. I suppose the other thing to say is you also see in the release that the guidance for the underlying OpEx in Q1 is $75 million, $77 million
I said a few minutes ago that underlying it was GBP 73 in Q4, removing the incremental bonus and the FX charge. GBP 73 going to GBP 76, reflecting some overall employee cost inflation that tends to kick in at the beginning of the year. Obviously, we've been hiring people, so there's a sort of full quarter effect of hires that we've been making. Takes us into that sort of GBP 75-GBP 77 range for Q1, we think. That's assuming a sort of neutral mark-to-market. Clearly, if the quarter ended tomorrow with the dollar sterling where it is, we'd have a mark-to-market credit in Q1. Obviously, there's a lot of water to go under the bridge before 31 March.
Standing back and looking at the full year, again, I don't want to repeat what Warren has said but it's been another year of investment in people and in business infrastructure. Net headcount up 276 in the year, about 80% of those people being in research and development but also commercial feet on the street, back office, support functions, accommodating the growth that we're seeing and that we're going to see in the future. That investment will continue in 2013 maybe broadly at the similar level, we'll have to see how the overall environment plays out. It's very early on in the year to make specific judgments about our level of recruitment. We would envisage something broadly similar to what we've seen in the last couple of years.
Notwithstanding that investment, of course, we've seen very strong net cash generation, especially in the second half of the year. We're recommending a full year of final dividend for 2012, up 35% on last year's final, which overall would be a 29% increase in the full year dividend, taking it to GBP 0.045. Licensing and backlog. You have heard me talk about $35 million of licensing being the base plus or minus. You've heard me talk about $60 million and $65 million and $70 million. This morning you've seen us report $85 million. I'm sure you're all interested in my guidance as to what the base is for 2013. I think the answer to that question is about $75 million. Again, ±. Are there going to be quarters in 2013 where our licensing starts with an eight? Probably. Could there be one where it starts with a six? Maybe.
Should we panic if there is? No. Licensing is inherently lumpy, the backlog is high. The underpin of licensing is higher, obviously, with that record backlog. $75 million ± is how we see it based on backlog and based on opportunity pipeline of licensing opportunities that are either in flight now or we would expect them to be in flight later in the year. Of course, other things will come onto the radar later that we're not yet seeing. What that chart, which we introduced a couple of quarters ago, was really there to show because I think the question. We've always said licensing is a mid to high single-digit revenue growth stream over periods. Before the downturn, it was 9%. It then went down, it then bounced up, it's been growing strongly in the last couple of years.
The question was, have all your Christmases come at once? Is this just a bounce back out of the downturn? Is it sustainable? What we're saying is, yes, from this elevated base, we're still guiding, if you look out multiple years, mid to high single-digit on top of this base that's been growing at more than 20% run rate for the last couple of years. One of the main reasons we say that, of course, is that chart on the right, which shows how the backlog has grown over the last five years and how licensed revenue has grown over the last five years. There's quite a big gap to close, which of course is closed by the recognition of license revenue over time. $75 million ± mid to high single-digit growth rate sustainable over multiple years, we believe.
The outlook. We came into the year as we have said, licensing pipeline looks good. Order backlog is high, up 25% in Q4 based on the sort of new products and new markets that we are going into, but it does remain lumpy. $75 million ±. In terms of royalties, we grew royalties 21%, as you saw in Q4. These are processor royalties. We grew processor royalties 17% in the full year 2012. The market in Q4 2012 versus Q4 a year ago, which is obviously a relevant relationship when looking at our Q1 royalties, was slightly up, flat to slightly up. Based on all that, it is reasonable to expect that Arm's royalties in Q1 could be 20%ish up year-on-year. Q1 royalties last year were $93 million.
That would take you somewhere into the $110 million, $112 million, $115 million area if we were continuing to grow at around that sort of 20% plus. Taking that license revenue guidance, looking at recent royalty growth, looking at the development of PIPD, development systems, in the round, we think that is around $250 million ± likely for Q1. Normalized OpEx, as I said earlier, $75 million to $77 million. The reminder at the bottom that it is very early on in the year. We are living in low growth times, to be kind. Whilst we may think that sort of catastrophic risk has reduced, we are still living in low growth times. There can be negative influence on consumers and on the semiconductor industry. We need to be careful.
Assuming it doesn't deteriorate significantly from where we are now, we would be confident in saying that our full-year revenues will be at least in line with where they currently are, which is about $1.030 million. I think with that, we will open it up for Q&A. Thank you.
Okay. I think we got the first one there who's towards the front on this side. Couple in the front row there.
Morning. It's Nick James from Numis here. Just two questions. First was on the OpEx. You've highlighted the need for investment in the business. I guess just to understand your thoughts on the longer-term potential for operating margin expansion. Is the fact that now we're kind of looking at other markets beyond the very high volume mobile one, mean there's less of an opportunity for operating margin expansion as we've seen previously? The second one would just be in terms of looking at the level of outperformance we can expect this year. Obviously, last year was a very strong growth year for high-end smartphones. It now feels like the growth in smartphones is moving much more to the lower end. Does that change the degree of outperformance you see relative to the semiconductor market?
Okay. I'll do the first one. There's no real change to how we see the investment proposition here. I think on a quarterly set of results, we're sort of focusing on the costs and why we're investing. It doesn't really change the view that there's operating leverage inherent in this model. You've got generic technology that's becoming more and more applicable to more end markets. We don't see step changes in our cost structure that mean that everything we've told you about margins before has to be rewritten. I think there's a very valid question about, you've got 45%-46% margin today, what's it going to be in five years' time? Is it going to be 50? Is it going to be 55? Is it going to be 60?
Warren's point is, in five years' time, the relationship between our revenue and our cost, they're not going to relate directly to each other. One's going to be based on costs incurred in the past, and one's going to be about revenues to be earned in the future. It could well be that this business is much more valuable in terms of its potential to generate profits in cash with a 50%-55% margin than with a 60% margin, which may mean that royalties are a higher percentage of revenue, the business is more mature, et cetera. We don't think of it specifically in terms of margin. We think it in terms of our investment opportunity and our ability to generate profit. You should certainly expect margin to be higher and continue on this gradually increasing trend.
The second part of the question was about royalties, shipments, and outperformance of the industry. Did we feel that with slowing growth in smartphones, the outperformance was likely to change? I think, the slowing growth in smartphones is arithmetic. There are estimates out there, by the way, for still significant growth in smartphones in 2013. I don't think we're going to see a massive change there. I would also point out that as some of this smartphone growth is fueled by effectively low-end smartphones. Sometimes that's perceived as a bad thing, strangely, for Arm's business, which is of course a complete fallacy because a low-end smartphone is typically replacing a simpler phone. Arm is still earning perhaps five times as much royalty on a low-end smartphone as on a simple phone.
Those changing dynamics in smartphones, I don't think really makes a massive difference. Whereas historically, as we saw on the chart, the outperformance has been in the 10%-15% range. I said, it's in the 15%-20% range. Whether it's 21, 19, 22, or 17, we can't really tell you what it's going to be at the moment. It's going to be in that range rather than the 10% range.
Thank you.
I think Sandeep was next, actually.
Hi, Sandeep Deshpande, J.P. Morgan. Thanks, Warren. Couple of questions. Firstly, on licensing, you've been growing licensing extremely strongly over the last few years. Can we have a comment on what is the long-term growth rate on licensing that you are looking at in your own model at this point? Secondly, on the royalty per device, you are seeing this uplift as you move towards these latest generation Cortex-A processes. At the same time, you saw a small dip from the last quarter to this quarter. Can you make a comment? Was it that some other businesses are growing, which is causing a dilution in the royalty per device? Then finally, this guidance on the first quarter in terms of revenue, Tim, maybe you can make a comment. In the last 12 years, you've never given a quarterly revenue guidance. What's changed?
Did you say 12 years?
Yeah.
Wow.
Okay. Do you want to?
I thought it was eleven in three weeks' time. On that latter point, Sandeep, we don't often specifically mention it, you're right. Funnily enough, we did this time last year, actually. If you look at your Q4 2011 release, you'll see we pointed to 200. It's not unprecedented. Our job is to guide the market the best way we can, and if we think it's appropriate to put in a specific number, then that's what we'll do. Sometimes we do, sometimes we don't. As I say, it's all about guiding the market best way we can. On the licensing, it's really what I said up there, Sandeep.
I think from having had these periods of much higher than historic growth in licensing, when we look forward, look at the backlog, look at the customers, look at the markets we're entering, look at our technology roadmaps, we see a mid to high single-digit % growth rate from here out into the medium term.
The middle question was about royalty rates. I don't think we see a change to the long-term upward pressure with higher value microprocessor cores in applications processes pushing the royalty rate up. In any given quarter, the actual answer is going to depend on relative growth rates of those types of products versus things like microcontrollers versus low cost in connectivity devices that are also going into things like smartphones and tablets. On a given quarter-by-quarter basis, don't read too much into the numbers. The medium-term trend is Cortex-A, Cortex-A + Mali, Cortex-A + Mali + physical IP, multiprocessors. We've got to the stage in 2012 where we're starting to see the Cortex-As with about one in four Cortex-As had a Mali attached to them in the last quarter. We're seeing physical IP being attached as well and contributing.
As we look forward, we'll see big.LITTLE drive things like multiprocessors as well. The long-term trend is there to stay. Goodness, where do we start? Let's have some over this side because I think they were quite keen as well.
Sorry, did you? Yeah, it's Francois from Morgan Stanley. Thanks for taking my question. The first question would be about the royalty rate on average for the company. I think last year it was 1.2%. If you could give us an update of where it is today or for 2012. Second question is about servers. If you could give us a bit of update in this market. I'm pretty sure that the chips are being tested at the moment by most of the end customers being Facebook or whatever. What's the feedback so far from the end customers, and how long do you think it will take for an IT department in a bank or whatever to have a try and basically give up Intel?
Okay. On the royalty rate question, I don't have a specific number for you-
It's a little bit higher
off the top of my head.
It's a little bit higher.
Yeah. It's a little bit higher last year than it was the year before, and it'll be a little bit higher this year than it was last year because of the answer that I gave to Sandeep's question. That trend is absolutely in place.
1.5, 1.3.
Think of those sorts of numbers. Actually, it doesn't necessarily need to be high. Obviously, if the microcontroller of Bow Wave was totally overwhelming, it could actually be a little lower and royalty revenues could be even bigger, right? It's not sort of totally, in a sense, a reasonable comment that it is a bit higher, but it is. The incremental percentages that Warren talks about are outweighing more volumes at the 1%. Gradually going up.
Yeah. As I said in answer to the last question, it is going to fluctuate a bit on a quarter-by-quarter basis. On servers, yes, we're seeing increased momentum. We are at least eight closer, we're 12 months closer to 2014, and therefore, the appearance of sensible, commercially available Arm-based servers. We are more certain the timescales aren't actually changing, as though we are transitioning from there being no servers to some servers, and 2014's when it's going to be sort of something that you can actually notice. In terms of the feedback that's coming back, so far, the experimental work is actually exceeding expectations. The original theoretical stuff has been proved and more, and that is fueling some enthusiasm, and you saw some announcements in 2012, and I think we'll see some more in 2013. Let's move down that. Oh, sorry. Here we go. First. Sorry.
Nizam from Anglo American Heritage. Just wanted to ask two questions. I've seen you've signed or you announced 16 big.LITTLE partners, which is quite a big number I think. I just wanted to get a sense from you as to what sort of applications we should expect to see big.LITTLE, and also more importantly, if we're going to expect to see big.LITTLE in the lower tier of the smartphone market over the coming years. That would be my first question. I've got a follow-up. Thanks.
Okay. big.LITTLE, we're seeing in smartphones and tablets. That's where it's happening first. You've got to have the sort of workload which is quite variable, so that you can have some workloads which are compute intensive and some workloads which are compute light. You can get the real benefit of big.LITTLE by using the appropriate size engine for the appropriate workload. That applies in phones, in smartphones. It applies in tablets. At the moment, it's clearly a sort of leading-edge technology thing. I don't see why over the coming years, the sort of normal trends we see in the technology space, which is something gets introduced at the high end and it trickles down. I think you will see that in lower-end smartphones. It's going to be some years away before it gets to those sort of price points that are sensible, but it will definitely happen.
Great. The other question is, obviously over the last 12 months, there's been a lot of debate about your key competitor driving Moore's Law and process technology ahead of your foundry partners. However, I think over the last few weeks, we found out that TSMC is going to ram 20 nanometer quite a bit ahead in terms of volumes that perhaps the market expected. Samsung is alluding to a 14-nanometer FinFET perhaps as early as later this year or second half of this year. I was just wondering, if you combine process technology from your partners on the manufacturing front with big.LITTLE, is it possible that your performance per watt versus your key competitor, in fact, increase in the coming quarters or years?
Well, obviously, we hope so. We have not really subscribed to this notion that Intel should have a long-term significant technology lead. Clearly, the business model is much easier for Intel to get to a new node ahead of the foundries. The foundries, as I've said before, have to support many customers, and it's a more challenging task for them. 2012 has seen the foundries, the TSMC, Samsung, GlobalFoundries, have all talked about their leading-edge roadmap, UMC is starting to talk about it as well now. The same equipment's available for everybody. There's no reason why Intel should have a long-term sustainable lead. They just have a sort of middle lead because it's easier for them to service effectively one customer instead of several.
As you saw me talk about up there, Processor Optimization packages for 14 nanometers, our physical IP team are engaged with these foundries. You see in the release we talk about Samsung taping out 14 nanometer Cortex-A7 based test chips with Arm and Cadence. It's real.
Thanks. Maybe just one final question.
Yeah, I think we need to press on then.
Just a final point.
Very quickly.
on the embedded front
Get the last thing from me.
Last quarter, Renesas announced a Cortex-A8 microcontroller, and yesterday Atmel announced a Cortex-A5 microcontroller. Of course it's probably too premature to talk about an increase in royalty rate for microcontrollers, but is it unthinkable that medium to long-term, that royalty rate also creeps up, in addition to what we know as a sort of 1% based business? Thanks.
Yes. It's not unthinkable. However, the reason we talk about microcontrollers and the lower ASPs is because there is a relationship between ASPs and volume. In the microcontroller space, when we talk about 23 billion units, what really drives the high volume is the low-end products. It's true today that you can buy Arm-based microcontrollers for GBP 0.10. You also have to pay more than GBP 10 for some Arm-based microcontrollers. Microcontroller covers a multitude of different applications, and some of these devices are quite sophisticated. The more expensive ones are going to be much lower volume relatively. Right. I think we need to start moving towards the back of the room.
Thanks. Gareth Jenkins.
We will get round to everybody.
Thanks. Gareth Jenkins, UBS. Just a few if I could. Warren, you talked about challenges of implementation, I just wondered what precise, apart from security that you mentioned, what precisely you were talking to. Have you had any challenges, for example, with the foundries implementing big.LITTLE, Are there any kind of challenges that you're finding shorter term? Secondly, I guess unit volumes in handsets down 3% for the industry. You're up 5%. Can you talk about your mobile royalty revenue growth, year-over-year, please? Lastly, it looks like your graphic share may be pushing close to 30%. I think this time last year, you talked about a target for Mali of over 100 million units. Just wondered if you could talk to a 2013 target. Thank you.
Right. Three questions. First of all, on the challenges. We have the slide in about challenges. Not to talk about a specific challenge, but simply to explain that execution, Execution I hope in the business has been pretty good and will hopefully continue to be pretty good. I'm simply pointing out that we need to spend some money to make it good. As these devices get more complex, There are many, many more variables involved in creating a processor. big.LITTLE is just one example. If you're trying to simulate 300,000 transistors, it's easier than trying to simulate 1 billion transistors. Much easier. You need a lot more machine cycles and a lot more brain cycles to work out how to, and then to effectively simulate the more complex device. That's it. It's a generic comment rather than anything specific. The next question was about mobile.
I think the best thing we can say there is, when we gave some very specific guidance or specific information a quarter or so ago, we probably made a bit of an error giving too much information to the financial community because some of our customers also can get access to this information, reverse engineer it, and use it in commercial negotiations. We're not going to make that mistake again. I said that once before. If we look at Q4, mobile units were up about 5% and mobile value was up just over 20%, 21%. On graphics and our share, we expect the increase in share to continue. You've seen the graphics licensing. Licensing doesn't guarantee royalty because we have to get design wins as well as get the licensing. The sort of upward trend on our graphics volumes we expect to continue.
It's fourfold increase in 2012. I'm not sure that we're expecting a fourfold increase in 2013. Obviously, easy to increase off small numbers. It's going to be significant, and it's going to follow tablets and smartphones. I think we need to move on swiftly.
Hi, it's Janardan from Liberum Capital. A couple of questions. One is just on your Q4 royalty seasonality. Last year, you pulled out the Q1 trend because you said that you had pulled forward some revenue into Q3 on the royalty side, on the PD royalty side. Normal seasonality previously was for a rise in your Q4 shipments on the actual quarter basis. I'm just wondering, is that seasonality changing now that you're getting a drop in your quarter-on-quarter royalties, and is this the way of the future, which is that your Q1 recognized royalty revenues will always be down from Q4 levels? The second is just on the Windows 8, Windows RT front. Windows RT has got off to a bit of a disappointing start.
How do you see that evolving going forward, and how do you see the Arm ecosystem gaining traction in the Windows 8 operating system, and how long do you think that process could take? Lastly, your licensing is rising very fast and your number of architectural licenses seem to be increasing. Is that a trend that you're going to see more and more, a bigger and bigger share of architectural licenses amongst your overall licensee base, which will keep nudging up your licensing levels in the future?
Yeah. Thanks, Janardan. It's difficult to talk about seasonality and the fact that seasonality will stay the same, et cetera. What I was pointing to there was the relationship that we saw between Q4 royalties last 2011 and Q1 2012 is probably the most recent information we have of how the seasonal impact of the current environment, how it plays out. The sort of 20%+ growth that I was talking about year-on-year are also consistent with what we did in Q4 and what we did in the full year 2012 for a growth rate. Sitting here today with very few royalty reports, and seeing guidance from our partners, which as usual, some is really quite positive sequentially, some is negative. That seems like the best information we have to guide at the moment.
I'm not making any prediction about what seasonality might mean going forward relative to how it's been in the past. Last year was probably a good indicator.
Next question was Windows RT. What do we think about Windows RT? 2012 was the launch of RT, and our expectation was really satisfied that Microsoft proved they could bring a Windows operating system up on Arm technology, and it works, and it's fine. It is a very controlled release that Microsoft have done. They have very tightly controlled the number of chip companies they're working with and the number of chips that they're working with, and the number of OEM customers whom they're working with as well. That is a matter for Microsoft. I believe that their 2013 plans are similarly very tightly controlled. If you want to probe on that, you'll have to push a bit harder with Microsoft. As far as we're concerned, we're pleased to see the platform running on Arm.
We think that the world needs multiple operating systems, and we stick with our usual principle of being pretty agnostic, and the rate at which Microsoft take share or not from other operating systems is a matter for Microsoft. We're going to target the full range of operating systems that are out there. On the licensing forecast and architecture licenses, you have seen a bit of incremental architecture licensing over the last year and likely over the next year or so as well, for two reasons. As we target a newer, broader range of applications, as we get into things like networking infrastructure and servers. We're exposing ourselves to different semiconductor players.
For example, Cavium's license back in the summer of 2012 is an example of that, where they've sort of previously been a MIPS house and adopted Arm for some of their networking applications. The other thing that we're doing at the moment is going through a transition from Arm architecture version 7 to architecture version 8. The existing architecture licensees are thinking about upgrading. For those two reasons, we are going through a period of seeing sort of some increased architecture licensing. Those are the reasons. There's not a fundamental shift in the business model. Simon, I think we'll keep on with the pace.
Thanks. Simon Schafer, Goldman Sachs. Just wondering actually on this incremental upside surprise for licensing. The new architecture, is this yet again another big step-up in the amount of royalties that you can charge? If I think back about the schedules that you had specifically since sort of the initial Cortex generation, there was obviously a step-up, as is the case for v15. How should we think about the type of value that you can extract in your royalties then a few years down the line based on some of this incremental licensing activity that you called out today?
Okay. It's a continuation of the trend. As we've moved into the initial wave of Cortex-A through more sophisticated versions of Cortex-A. The royalty rate has increased as we're delivering more value to the semiconductor partner than we're expecting to be paid for that incremental value. Version 8 is another increment again. We're expecting to be paid for that. Whereas in the past we've done sort of 1%, 1.25% up to 1.75%, generally version 8 architecture licenses, the number starts with a 2. Tim.
Do you want to do Suman next? Here at the front.
Yeah.
Thanks. It's Suman from Redburn Partners. Congratulations on an excellent result. I guess when you have good results, you tend to notice some of the negative niggles as well. If you don't mind, I have two on that. First is on options expense. It's up 45% Q and Q, and up 14% year on year. Could we do something, or could you do something about how to sort the sort of volatility we may see in terms of reported earnings? The second question is to do with the PIPD segment results. When I look at it, and I think I asked you this question in Q4 2011 as well, in terms of what could we see in the next five years in terms of operational profitability of that particular business.
It's still -20%. I'm just wondering if there's a strategic view over the next five years, how you could bring it up to the processor division sort of profitability or even to break even. That'd be really great. Thank you.
Yeah. On option expense, unfortunately with the option accounting, there is volatility. There is volatility specifically when you get big moves in share prices. The fact that our share price moved significantly in Q4 meant that there was, in a sense, the option charge or elements of the share-based remuneration charge went up as well. If you look back over the last sort of eight quarters, it's sort of uniformly below that level, and will go lower again in the future. Now, there are various debates about how this should be disclosed and accounted. What we try and do is set it out very clearly in the same way that we always have done, which is, here's our normalized numbers, here's our option charge, here's the subtotal with the option charge included, here are some of the other non-cash accounting items.
In a sense, we extract the volatility, but we appreciate it's there. The reason it's there this time is because of the spike in the share price in the fourth quarter. I think when we talk about this to investors, I think the most important issue is dilution and how many shares are you issuing to your workforce every year, and that has remained consistent, typically under 1%, is unchanged. Now, that gets translated into accounting rules that are volatile, mainly because of share price movements. PIPD segments, I think Warren explained that a lot of the value of PIPD is to the wider Arm. A lot of the value that the PIPD brings to Arm doesn't actually go through the PIPD segment. It goes through in the processor division.
I think the other thing to note is, as you say, there's a marginal improvement from last year. We've been investing in that division quite heavily because, as Warren said, it's now acknowledged as the leading third-party provider of physical IP. Multiple foundries are wanting us to work on multiple projects at the leading edge, and that's requiring more people, which is why you see that result. Strategically, I think PIPD is more important and more valuable to Arm now than it's been at any time since the acquisition.
Great. Last few.
Thank you. Andrew Gardiner from Barclays. Just again, on the licensing, you highlighted in one of the slides, Warren, the increasing gap between your backlog relative to the licensing revenue in the quarter. With your comments also around the shift from v7 to v8 and more customers at this point having to consider those upgrades and around architecture licensing, do you think that gap is going to widen for the time being before we start to see it narrow? Is that inherent in your mid to high single-digit long-term guidance, or could it perhaps outperform that?
I think the mid to high single-digit long-term guidance stays. I don't think we should look at that graph too closely on a quarter-by-quarter basis. The reason it's over multiple years is because it's meant to show that, hey, there's a trend over a handful of years. We have the backlog to support the increased level of licensing. When you get into quarter-by-quarter, look at the last quarter, it's just shifted quite a lot. That is exactly because of the new technology. We are having this little wave of people licensing new technologies. When we sell somebody a v8 license at the moment, or license to a v8 product, we don't recognize revenue. 100% of that is going to go into backlog.
That is going to cause the gap to widen because you've got backlog growing and license revenue not happening as a result of that order. That's going to happen for a few quarters, and as Tim said in the presentation, license revenue is lumpy. Sometimes that's because orders happen on a Friday, or then there's the end of the quarter, and there's the following Monday. That sort of thing can make a big difference to the actual numbers. Do not look at it on a quarter-by-quarter basis, my advice.
Just quickly for Tim, on the updated tax guidance. I think previously you'd said medium term, you were looking for your tax rate to step down to something around 17%, largely driven by the Patent Box. But with these other changes you're seeing, whether it's the California change and others, are we now even lower than that, or is 17% still a good spot to be?
Once we have the full benefit of Patent Box out in effectively 5 years' time, then all other things being equal, and of course they won't be, sort of mid-teens erring still to the 17%. We are helped by the California issue. On a sort of 5-year view of our tax charge, that's getting a little bit in the noise.
We have room for two more. Let's have these two, and we'll come to you afterwards, okay?
Yeah. Jerome Ramel, BNP Paribas. Just to come back to the manufacturing part of the story from the foundries. It seems that the 14 and 16 nanometer node FinFET from the foundries is not a 14 or 16. At the gate length, it is still a 20 nanometer node. It really seemed to me that Intel had the lead, and more importantly, from a cost perspective, Broadcom and other companies said that for the first time at 20 nanometer node, they do not see a price improvement. I am just wondering if that negative trend might actually be a positive trend for Arm as customers might be forced to accelerate the roadmap and ask for more leading-edge core from Arm.
It is certainly a feasible line of thought. I maintain that what matters is not the geometry, whether it is 14, 16, 20, but what matters is what performance do you get out of the processor, how much area does it take up on the silicon, and how much does it cost to produce the chip? That is what really matters. Companies like Broadcom traditionally operate a little bit behind the leading edge of process technology, and they find that a very successful formula. It enables them to get products out to market much quicker because there is much less uncertainty in the implementation, and that is what works for their business. The good thing about the Arm business is that we supply our processes to a range of different companies, some of whom swear by operating just behind the leading edge like that.
Others choose to operate at the leading edge, swallow the uncertainty and the iteration, and go for the leading edge. What matters to us is semiconductor companies getting Arm chips out at the right levels of performance for the market, at the right price for the market. We recognize that different people have different approaches, and so we have to support all those approaches. Let us have the last question.
I am Mithurjanani, Citigroup. Thanks for taking my question. My question would be on networking. What do you see in the networking space today in terms of competitive dynamics? On a three- to five-year view, how do you see enterprise networking contributing as a percentage of your royalties? As a quick follow-up, in terms of the investment in the MIPS portfolio, do you see an opportunity to make money out of this rather than just purely being a mitigation or an insurance kind of act? Thank you.
Okay. On networking, we do expect to see Arm's presence in networking increase over a five-year period. I don't think you're going to see it be massive out to 2017. 2012 was a year of design wins. It was a year of license enabling with companies like Cavium, like Freescale, like LSI Logic. Some of those have the design wins with their customers. There is a period now of implementation, and then those products will get introduced, and they'll get phased out and introduced in phases. Things like the rollout of 4G networks, which is happening now around the world, is a great stimulus for people to think about, okay, so when we're coming to upgrade this network, how are we going to make it more power efficient and use leading-edge technology? That's where Arm plays a role.
Right now, the things that are getting installed are not based on Arm technology. They're based historically on a lot of PowerPC architecture is out there. For monetizing the MIPS patent portfolio, we did that primarily for what it said on the slide, which is patent risk mitigation. We're not interested in monetizing the MIPS patent portfolio. We're interested in monetizing the Arm architecture, and that's what we're doing. I think with that, in the interest of time, I do apologize if people have got some additional questions. We're going to have to bring it to a halt. Thank you very much.