Thank you for standing by, and welcome to the Arm Holdings Q3 2015 conference call. At this time, all participants are in a listen-only mode. There will be a presentation followed by a question and answer session, during which, if you wish to ask a question, you will need to press star one on your telephone keypad. Please be advised that this conference is being recorded today. I would now like to hand the conference over to your speaker today, Mr. Ian Thornton. Thank you.
Thanks, Stephanie. Good morning, everybody. This is Ian Thornton, Vice President of Investor Relations at Arm. On today's Q3 results conference call, we have Simon Segars, our Chief Executive Officer, and Chris Kennedy, Chief Financial Officer. Then we'll open up the call to a Q&A session. As a reminder, the presentation and press release can be found on the Arm Investor Relations website at www.arm.com/ir. Before I hand over to them, I have to read out a few words with respect to this conference call and what we're about to discuss. The contents of this conference call are being directed only to those of you who have professional experience in matters relating to investments. The information communicated on this call is being made available only to investment professionals.
Any person present on this call who does not have professional experience in matters relating to investments should not act or rely on the contents of this call. The following conference call will contain forward-looking statements which are other than statements of historical facts. The company's actual results for future periods may differ materially from these statements, as they are based on current expectations and are subject to a number of risks and uncertainties. On this note, I'll hand over to Simon and Chris.
Thanks, Ian, and good morning, everyone. Thank you for joining our Q3 2015 results conference call. I will run through the business highlights. Then I'll hand over to Chris to provide some more detail on the numbers. Following Chris, there'll be some time for your questions.
At the beginning of this year, we told you that we would continue to gain share in our target end markets of mobile computing, enterprise infrastructure, and embedded intelligence, and that we would target new and growing markets by investing in future technology. That is exactly what we've done in 2015. We've seen the latest Arm technology be rapidly deployed across smartphones and tablets, increasing the royalty percentage per chip. We've seen more companies start to ship Arm-based chips in networking and servers. Our share of microcontrollers and smart cards has continued to grow. Throughout this year, we have announced new technologies and ecosystem relationships that extend the leadership position of the Arm partnership in markets where we are already strong and advance us further into markets where we are pioneering. Q3 was another quarter of progress against this strategy.
We saw particularly strong royalty revenue growth driven by more chips with higher royalty rates as Armv8, Octa-core, and Mali shipped in ever higher volumes. We saw a major semiconductor company ship its first Arm-based chips for base station equipment. We saw a major semiconductor company demonstrate its first chips for servers. In the quarter, Arm signed lead license agreements with partners for the next generation of products, and we announced investment plans to accelerate our roadmap to 2020. In August each year, we hold the Arm Partner Meeting, and this is now a major industry event. It is a conference that we host in Cambridge, where over 600 senior managers from our partners across the ecosystem gather to discuss the future of computing and jointly how our roadmaps can deliver it.
We saw tremendous engagement between silicon, software, and cloud companies and a real enthusiasm for them deploying the Arm architecture across all application areas. The Arm Partner Meeting informs our strategy, and this year at our Capital Markets Day in September, we unveiled new investments that we believe will generate around GBP 200 million of revenue in 2020. We're investing in mobile to meet the demand for continuous innovation in the handset market. We're investing in networking and servers to accelerate our market share gains. We are investing in Internet of Things businesses to create new revenue streams. The Internet of Things will smarten our cities, homes, places of work, and bring new efficiencies to healthcare, agriculture, and transportation. It will change the way we interact and bring business opportunity and economic growth. To deliver on this potential, ultra-low-cost sensors will have to implement very complex security protocols.
This is a technical challenge for our industry and one that brings opportunities to Arm and our partnership. By delivering cost-efficient solutions that enable secure systems, we will create brand new markets for our processors and software. In July, we announced the expansion of our security offering with the acquisition of a company called Sansa Security. Together with Arm TrustZone and mbed device platform, Sansa's technology will enable trusted embedded devices that boost the protection of data and content in any connected application. Also during the quarter, IBM, Microsoft Azure, and Salesforce.com each announced that their cloud-based analytics platforms would support mbed. I'll now discuss the royalty drivers in the different parts of the business in more detail, starting with technology revenue drivers. Arm's royalty revenues are reported one quarter in arrears, so our royalty for Q3 was generated from chips sold by our licensees between April and June.
Our partners reported that they had shipped 3.6 billion Arm processor-based chips, up 20% on last year, including 30% growth in microcontrollers and smart cards, 20% growth in smart mobile devices, and 20% growth of chips into networking equipment.
We saw strong royalty revenue growth in mobile, driven by continued adoption of Armv8-A, higher core counts, and Mali graphics technology. Processor royalty revenue was up 37% year-on-year, and physical IP royalties were up 31% year-on-year. Chris will give some more color on year-on-year growth later on. Across mobile, home, and enterprise segments, our partners shipped 215 million Armv8-A chips, which was approximately 35% of our Cortex-A volumes. At the start of the year, we expected that Armv8-A would appear in around 50% of smartphones in the final quarter of 2015. We actually passed that milestone in Q3, when around 60% of smartphone chips were based on Armv8-A. Shipments of multiprocessor chips with our big.LITTLE technology doubled sequentially. We expect the amount of Arm content in mobile devices will continue to increase through 2016.
Armv8-A penetration has further to go, multiple vendors have recently announced octa-core solutions aimed at mid-tier devices. Our graphics technology has also continued to gain share. We now expect that Mali shipments will be at the upper end of our guidance for full year 2015, which was 600 million-700 million units. Yesterday, we announced the latest edition of our range of Mali GPUs, the highly efficient Mali-470, which addresses entry-level smartphones, wearables, and IoT devices. Announcements from our partners indicate that the diversity of Arm chips will continue to increase over the long term. Earlier this month, STMicroelectronics announced that it will transition some of its automotive products to the Armv8-R architecture, Qualcomm demonstrated its first Arm-based server chip. Now to licensing.
We signed 38 processor licenses in the quarter, 14 of those being with new customers. We also saw continuing interest from OEMs who are building up their in-house design expertise. In Q3, we licensed to OEMs in the mobile, home, and enterprise sectors. Six of the licenses signed were for Cortex-A class processors based on the Armv8-A architecture, of which two were for lead license agreements for future processors that we are still developing. We also signed six more licenses for Mali, including our next-generation graphics processor, codenamed Mimir. We also signed 18 licenses for Cortex-M class processors, taking the total number of licenses signed for this technology now to over 330, with more than 220 companies. To finish, the success we have seen so far in 2015 all stems from the investments we have made over the past 25 years.
As announced at our Capital Markets Day in September, we are accelerating our investment in areas such as enterprise and IoT. Most of this investment is in the form of additional engineers. We are hiring a mix of industry veterans and recent graduates, as well as adding experienced design teams via acquisition. Our ability to recruit and retain top talent puts us in a great position to continue to execute on our strategy and deliver superior returns for shareholders over the long term. I'll now hand over to Chris, who will provide some further details on the numbers.
Good morning, everyone. Hopefully, many of you will have had a chance to have had a look at this morning's earnings release. As usual, the quarterly roadshow slides are available on our website. Overall, Q3 dollar revenues at $375.5 million were up 17% year-on-year. Within that, processor licensing revenue was $125.9 million, 5% higher than a year ago. Following 5 years of accelerated licensing growth, which grew at 29% compound between 2010 and 2014, we've seen a slower growth in 2015. We continue to expect licensing revenue growth of 5%-10% per annum in the medium term, and Q3 was within this medium-term range. Physical IP licensing revenue was $19.2 million, down 14% year-on-year.
Physical IP licensing has declined in Q3 2015 compared with prior quarters as we complete agreements relating to 28 and 20 nanometer nodes and are starting to transition to delivering the next generation of technology at 16, 14, and 10 nanometers. We now expect Q4 physical IP licensing revenue to be at a similar level to Q3. Group order backlog at the end of Q3 2015 was down about 7% sequentially. Q3 license revenue growth included a 60% contribution from backlog, which is at the upper end of the normal 40-60 range that we typically see. Looking at the medium-term outlook for license revenue growth, we expect quarterly sequential movement to backlog to continue to be lumpy. Processor royalty revenue was $185.6 million, up 37% year-on-year, in a period when the industry declined about 2%.
This is the largest outperformance of Arm's royalty revenues versus the industry that Arm has seen since 2003. This outperformance is partly due to the weak comparator. You'll recall we saw an industry correction in 2014 that resulted in slower than normal growth in Arm's royalties. The underlying outperformance is due to recently launched Arm v8-A-based chips, which delivered a combination of premium chip pricing and an elevated royalty percentage per chip. If you look through the cycle to Q3 2013, which was also a strong quarter, we see a 17% two-year compound annual growth rate in processing royalties, which is a trajectory consistent with what we believe we can deliver in the longer term. Normalized OpEx in Q3 was £108 million, in line with our guidance in July. Most of this increase reflects ongoing investment in R&D, including the operating costs of the Sansa acquisition.
The Arm team has grown 19% over the past year to 3,850 people, with the majority of the extra employees being R&D engineers specializing in processor and physical IP development. Q3 costs are also impacted by the strong dollar, with about a third of our cost being in U.S. dollars and therefore being translated into GBP at the higher rate. As a reminder, as a rule of thumb, a 10% movement in the U.S. dollar to GBP exchange rate has a 15% impact on EPS. Normalized operating expenses in Q4 2015, assuming effective exchange rates similar to current levels, are expected to be in the range of GBP 117 million-GBP 119 million, reflecting ongoing organic investment in the business and a contribution from recent acquisitions.
The group's normalized tax rate in Q3 was 16%, and the full-year normalized effective tax rate is also expected to be around 16% as we continue to benefit from the reduction in U.K. corporation tax rates and the phased introduction of the Patent Box tax regime. Now on to the outlook. Arm enters the final quarter of 2015 with strong royalty momentum and a healthy licensing pipeline. With industry and customer data underpinning the short-term outlook for royalty revenues, we expect group dollar revenues for the full year to be in line with market expectations.
Thanks, Chris. We will now go to Q&A. As usual, if I could ask you to just ask one question at a time, we will be able to get everybody's question in, and we can come round again if you have follow-up questions. Let's start with the first question.
We will now begin the question and answer session. If you wish to ask a question, please press star one on your telephone and wait for your name to be announced. If you wish to cancel your request, please press the pound or hashtag key. Your first question comes from Gareth Jenkins. Please ask your question.
Hi, Gareth. Gareth, you might be on mute. Can we go to the next question maybe, Stephanie?
Yes. One moment please. Yes, our next question comes from Matt Ramsay from Canaccord Genuity. Please ask your question.
Thank you very much. Good morning, gentlemen. Congratulations on the really strong results on the royalty side. Simon, I think I was a little bit surprised to see the 60% penetration of v8 within the smartphone ecosystem already. I know you guys had given some updated targets on where you thought that penetration could go exiting this year and next. Maybe you could give us some updated thoughts on that, given maybe the faster adoption of v8 in the mobile ecosystem than we had maybe expected in the past. Thanks.
Sure. The uptake of v8 has been very strong. When we entered the year, we thought that we'd see a raft of new devices launched at Mobile World Congress, and that happened, and then for them to ramp through the year, so we'd be at about a 50% penetration level come the end of the year. Good news is that happened sooner than we thought. We've passed that. We're at the 60% level in Q3, and we think come the end of the year, it could be as much as 75% as the kind of exit run rate. What we're seeing is the adoption of v8 across the breadth of semiconductor devices that go into mobile. Many low-end devices being launched with v8, and we're seeing rapid uptake in the market. This is good because it's a driver for increased royalty percentage rates.
The chips as they're introduced, new products from any semiconductor company typically command a higher ASP at the start of their life cycle. You're getting a compounding effect there of the higher chip price and the higher royalty percentage. That's not going to last forever, obviously. It was very strong in Q3, and we think that the year will end strongly on this. Still further to go through 2016, and we'll see that penetration continue.
Thank you. If I could sneak in a follow-up for Chris. Obviously, the OpEx side is ramping as you've outlined in some new investment initiatives for Q4 and next year. You've now sort of laid out where Q4 OpEx is going to be. I'm not sure
Consensus fully reflected some of the increased operating costs you guys have laid out at the Analyst Day. Maybe you could walk us through how you see OpEx expanding into next year and maybe what run rate we should start with to model 2016. Thanks.
Yeah. You're right. I don't think consensus fully reflected what I said at the Analyst Day. We said we'd add GBP 5 million of cost in Q4. That obviously gives you that exit rate of GBP 117 million-GBP 119 million. Really that should be your starting point for looking at where 2016 will be, obviously factoring in inflation within that and the ramp-up of costs that we flagged at the Analyst Day through the year.
All right. Thank you very much, guys.
Thanks, Matt.
Your next question comes from the line of Kai Korschelt from Merrill Lynch. You may now ask your question.
Yep. Hi, can you hear me?
We can hear you. Morning, Kai.
Excellent. Yep. Morning, gents. Yeah, just one on how to look at the multi-core penetration rates. I appreciate the v8 penetration, but it seems there's on top of the v8 driver, there's another step up in royalty rate once we go north of four or quad-core. I'm just wondering, would you happen to have any data on what you think the penetration rate in mobile is currently of chipsets with more than four cores? Because it seems like we're only at the beginning of that penetration cycle, particularly in the mid-range and in the low end. The other question was just quickly on licensing. I just wanted to clarify, should we expect Q4 licensing revenues broadly in line with the third quarter? Or should we see the typical seasonal uptick versus Q3? Thank you.
On the penetration of octa-core, we think in Q3 about 20% of the smartphones that were shipped were octa-core. Excuse me. If you look at phones being launched, and we tend to crawl over this data and look at analyst reports, we think it was closer to 40% of the phones launched, so this is across all SKUs, were octa-core. We think that one in five rate that exists right now is going to grow. A bit hard to say exactly where it's going to go to, but we think that the trend is in the right direction for further adoption of octa-core and big.LITTLE. The second part of your question was? Seasonal royalties. Was seasonal licensing.
Licensing. Yep.
Yeah. We do expect to see a bit of seasonal uplift in Q4. We normally do see a stronger licensing quarter in Q4, we do expect to see some growth there. I think compared to where consensus is maybe, we might see a stronger royalty and maybe not as strong licensing compared to what is in the models today. As we've said, we expect to hold the full year numbers.
Thank you.
Your next question comes from the line of Andrew Gardiner from Barclays. Please ask your question.
Hi, Andrew.
Thanks very much. Good morning, guys. Just to follow on to those first two questions. We've talked through v8, we've talked through octa-core. You also mentioned that Mali is trending towards the higher end of your expectations for the full year. Is that purely smartphones? Are there other devices where you're seeing adoption rise and just how are we thinking about that as we go into next year as well?
Mainly it's growth in smartphones. We're quite strong in digital TVs, although digital TV sales generally seem to be a bit flat at the moment. It's mainly growth in smartphones. We think medium term growth prospects for Mali is very good. We've seen a very strong response to the launch of our Mali-470 product, and that's something that we're designing specifically to target wearable devices, IoT devices, things where you need some level of graphics, but you don't need full-blown, all bells and whistles, high power or high performance rather. For this year, as I said, we're going to be at the upper end of that GBP 600-700 range. We may be even a smidge over GBP 700 if things go well in Q4, with continued growth into next year.
Thanks very much.
Your next question comes from the line of Sandeep Deshpande from J.P. Morgan. You may ask your question.
Thanks for letting me on. Simon, could you talk a little bit about, you've talked about the 25% market share in the server market, you said that at your meeting that you held in August, that you've seen a lot more interest in the cloud. In terms of how we should be looking at the roadmap of Arm adoption in the server. Any particular milestones that we are looking for? That's my first question.
Okay. Yes, at our Arm Partner Meeting in August, we cover a wide range of technology. There was a lot of interest in Arm in the data center generally in bringing data into the data center and then processing it there, where people are looking for alternate solutions, people are looking to achieve greater levels of efficiency, particularly power efficiency, as they scale out the processing around IoT and obviously with the trends behind mobile.
From the silicon that has been introduced to date, we've had good progress. We're seeing deployments of devices. As we explained at the Capital Markets Day a few weeks later, we see that momentum as validating our strategy. Hence now is the time for us to accelerate further and make further investments on software side, and at the system level, and helping with benchmarking, and all sorts of activities that are going to accelerate deployment. That leads us to a view of being able to get to a 25% market share in 2020. In terms of specific milestones, I think looking for silicon devices, it was great to see Qualcomm's recent announcement of the chip that they have built. That looks very high performance. Look for the deployment of that, look for the announcement of other silicon devices from other partners.
It's about deployment. We'll see more trials, better benchmarking, and volume growing over time. We'll obviously give updates on that as we go through our quarterly results over the next couple of years.
Thanks, Simon. My follow-up, just a quick follow-up on this issue of your royalties. You clearly had a very strong royalty quarter. The market was concerned because of the ongoing semiconductor inventory correction. There were these public announcements from many of your major customers indicating slowdown, whereas you've seen significant strength, not just in units, but also in royalty per device. Considering what they said to the market and versus what they have reported to you, have you done any sort of analysis between was there actually a slowdown and you have grown because of growth in other markets which were not there before earlier on? It's mainly the growth has come from the royalty per device, which is not entirely true, of course, because your units have grown as well.
Could we try to understand what was actually happening underlying as such?
I think the growth has been multidimensional. We've seen strong growth in embedded, and we've done lots of licensing of Cortex-M to lots of companies. Those devices are coming through, and volume is growing. Obviously, they're low-cost devices. We're seeing, despite the slowdown in the growth of handset units, greater Arm content, as we've explained previously, and that's helping royalty dollars. Growth in other markets as well, such as networking, which from a relatively small base was still 20% year-on-year in terms of unit growth. It is quite a broad set of markets where we're seeing growth, which is a reflection of the diversification of where Arm technology is used.
At the Analyst Day in September, I was trying to point out how the licensing has evolved over the last many years with companies taking Arm technology to address markets other than mobile, that is starting to see it come through. We're less exposed to any one market. Mobile obviously still being the largest in terms of dollar contribution. If you look through the press release and you see the breakdown of the units by different market segment, you can see that in round numbers, about 40% of the units are in mobile, 60% are in other markets. Those other markets are growing strongly for us. It's the breadth that is helping us.
Thanks, Simon.
Thanks, Sandeep.
Your next question comes from the line of Jerome Ramel from Exane BNP. You may ask your question.
Yeah. Good morning. One question, Simon. You mentioned ST Micro is indeed moving to Arm processor for automotive. When we talk to Freescale, they basically say the same, that their client are also asking them to shift from PowerPC to Arm. We previously had identified automotive market to be a great opportunity for Arm, but probably beyond 2020. Do you think we might see a kind of acceleration from that perspective? Maybe you will see automotive contribution to accelerate before the end of the decade?
Well, certainly I won't be complaining if that is the case. Yes, we're seeing adoption of the Arm architecture in different places and with different companies, and definitely Freescale is an important significant licensee of ours who've been using more and more Arm technology in more and more end markets which is very good for us. The thing about the automotive market, though, is designing cycles are very long. The rigorous testing that you have to do for safety reasons and robustness because of the range of conditions that cars are used in across the world, it does mean that the cycle is very long. It's not quite like consumer electronics. It means that we think about that on a much longer timeframe. Clearly, cars are changing. Cars are becoming a bit more like consumer devices.
There's a lot more electronic content going into cars, and we do see it as a big growth market. It's on a slightly longer trajectory than some of the other markets that we're targeting.
Okay. Thank you very much.
Your next question comes from the line of Amit Harchandani of Citigroup. You may ask your question.
Good morning, gentlemen. Amit Harchandani from Citigroup, and thanks for taking my question.
My question relates to licensing. We saw in this quarter, again, some OEMs taking up Arm licenses. You're talking about a 5%-10% growth medium term. Could you give us a sense of how licensing is shaping up between your direct customers or the chip makers that you originally licensed to, and some of these newer customers, maybe from even newer geographies? That's my first question. I have a follow-up. Thank you.
The majority of licensing that we're doing is to semiconductor companies. What we see excuse me, OEMs wanting to do, in some cases, is get involved more in the design of the chips that ultimately get built and manufactured by traditional semiconductor companies. In some cases, they want to do the whole design themselves and leverage the disaggregated foundry space to get chips built. It really is a variety of approaches people are taking. The majority of licensing that we are doing is still to semiconductor companies. Over the next few years, that may change. Obviously, consolidation is a theme of the semi industry at the moment, and I think that will play out over a number of years.
Good news is that, in terms of that end equipment, it's good from our perspective to see more OEMs directly interested in that technology, greater software content in the equipment that they're building, that's especially true in networking, and a greater adoption of the Arm architecture. We think that helps cement our position. The more code that's written for Arm, the greater the position that we're in. We are supporting that, and we are supporting this or working through this transition of the semi industry that I feel is kind of going on at the moment.
Thanks, Simon. As a follow-up, if I may. At the Analyst Day as well, you talked about the mobile computing opportunity, and one of the topics that we've talked about is use of Arm in content creation devices or clamshell form factors, or basically stepping up as the Arm performance improves. Could you give us a sense of how talks are shaping up on that side? What's the level of customer interaction, or how do you see the adoption shaping up with regards to sort of clamshell or content creation devices or laptop-like devices based on the Arm architecture? Thank you.
We're seeing some products starting to be announced, which there was one of the recent devices that Google announced with their new Nexus lineup with more clamshell-like form factor. We're seeing other companies build Android Books. Again, more 2-in-1s based on Arm. We think that the performance is there. We think software is coming together with more processing, more data storage, and online access to some of the more traditional productivity apps. This is a trend that suits us well. As a user of computing, you can be carrying around with you a much thinner, lighter, and lower power, and longer battery life device, which is going to enhance productivity. We think performance is getting there. We think the devices that we see in design are going to be really interesting.
We think this is another market that Arm can potentially gain some reasonable share in.
Thank you, Simon.
Your next question comes from the line of Francois Meunier from Morgan Stanley. You may ask your question.
It's Francois, actually. Right. I've got two questions, if I may. If I look at your v8 units, about 20%, I reckon, are from non-smartphones, so it's about 40, 50 million units. Could you maybe describe a bit more about what's really hot within those 40, 50 million units? That's the first question. The second question is about the data you have access to and we don't have access to, which is the underlying prices of chips going to smartphones. What is the trend as prices stabilized?
In terms of v8 units not in smartphones, I think you're about right. I think about 20% aren't in smartphones. About half of those are in tablets. 10% are in other things. Other things might be digital TVs, might be networking equipment, might be some server chips, small volume, but some there. You're about right with that split. Just as we saw with v7, the v7-A processors that we introduced found their way into other markets. As we've seen greater success in enterprise we are evolving our roadmap to add features into our cores for enterprise that are more targeted towards that market. We are, again, off the feedback that we get from our partners at our Partner Meeting evolving our roadmap to support what they want to do with our technology.
Your second question was about the data that we have and you don't, we're going to keep it that way. In terms of ASPs, as I was saying earlier, with new products being introduced, we're in a period now where Typically, as is typical when new semiconductor devices are introduced, they do come in at a premium. Some of the devices with a lot of content, high core count, high graphics capability, video capability, very sophisticated devices, you are seeing some of those with higher ASPs than some of the more mature devices. You can get 3G entry-level baseband plus apps processors for a very small amount of GBP these days. The high-end devices are four or five times that right now. I'm sure that pricing will come down over time because it does.
As is the way, the semiconductor players in this space are always looking to put new functionality in to reset that pricing and maintain ASP. We're in a period right now where, on average, it's probably gone up a little bit. That's probably temporal and we'll be back to what I think is, when you look back over the last few years, fairly flat pricing, fairly flat ASPs for smartphones.
Thank you, Simon.
Your next question comes from the line of Gareth Jenkins of UBS. You may ask your question.
Yeah. Hi. Thanks. Hopefully, you can hear me this time, gentlemen. Thanks for taking the question.
Yeah. We can. You got the mute button under control now.
I think I was kicked off, actually, but there we go. Just a couple, if I could, one high level and one maybe a bit more detailed. I just wonder, as you move to non-traditional business models with mbed server in particular and security software, whether you feel the need over and above what you've already announced in terms of investment into channel distribution, new sales teams to kind of directly address the OEMs that you're dealing with. I think you cited you're working with GE on street lighting. I think in your networking slide, you talk about OEMs rather than chip makers. That's the first question, maybe a quick follow-up on something else.
Yeah. Our traditional sales channel is targeted at semiconductor companies, we've got a very efficient mechanism for licensing semiconductor IP to semiconductor companies. As we branch out from that, yes, we are expecting to morph our sales channel accordingly. Different customers, sales cycle is different approach. We are learning our way through that right now. As we are evolving some of this, we have ring-fenced some of the R&D expertise to create the products, we have ring-fenced some commercial resource as well to go and target this new and different market for us. As hopefully we are successful with this, we'll grow the investments in both the R&D and the channel at the same time.
Okay, great. Just the quick follow-up. Just on the base station product that you've just launched, which I presume is HiSilicon, could you talk about the core counts or the processor counts on board a typical base station? Presumably, the chip ASP is materially higher. Could you maybe give an example of one of the products you've just launched? Thanks.
This isn't products we've launched. This is products-
Yeah
that our licensee are selling, I won't comment on your speculation on customer. You might be right, but you might not be either. The chips that go into that space are, as you say, quite large, high core counts. You're probably in the 16 to 24 core arena for, I'm just thinking off the top of my head for some of the chips I've seen in that space. I can think of a 16-core Cortex-A15 device that is targeting that area. We'll see that kind of core count with the v8-A processes as well. That's typically the kind of processor mix right now.
Thanks.
Yeah. ASPs, obviously, with that larger die size, you would expect them to be a multiple of what you get for a smartphone ASP.
Great. Thank you.
Your next question comes from the line of Eoin Lambe of Liberum. You may now ask your question.
Hi there. Yeah, thanks. It's Eoin. A question on the cash generation. Profit went up a lot year-on-year, I think 28% year-on-year. The cash generation declined, I think 5% year-on-year. I was wondering what the mismatch is there, why the decline in cash generation? Is it the unwind in deferred revenue and more licensing coming from backlog? Do we expect the cash generation to go back to historical levels?
Well, Eoin, it's Chris here. First thing is you're looking at a quarter. Within a quarter, you do get the impact, as you say, of backlog reduction and therefore lower cash conversion on that. You've also got just simply timing of cash receipts within a quarter, which are going to vary from quarter to quarter on the timing. I think the thing, in the end, cash and profit is the same thing over the course of a license. That's what we're expecting. We've had a number of years where cash has been greater than profit. As I said, the normal over the medium term, cash equals profit.
Okay. Thank you. Just a question on consolidation. There's lots of M&A amongst your customer base. How does that impact Arm? I presume it has no impact on royalties, but does it have any impact on your licensing? If your customers are consolidating, would there be any sense in people further up the supply chain, like an IP vendor like Arm and an EDA company, getting together to combat the consolidation further down the line?
In terms of consolidation in the end markets, one thing that we're seeing going on at the moment is companies coming together to form very sizable companies who are going to have the scale to invest in leading-edge technology, and that is a good thing. What we've sometimes seen as consolidation happens is you might have two companies who have licensed some technology, might look at their entire portfolio and say, "Actually, it makes sense to enter into a subscription agreement and make a bigger commitment to Arm technology." That can have the effect of accelerating royalty growth further. From that perspective, it's a good thing. Obviously, you can look at it very simplistically and say, well, with fewer companies to license, therefore licensing must come down.
I think that is too simplistic a view when you think about what the challenges are going to be for building complex leading-edge chips on advanced technologies in the future. On the other side of it, there are many startup companies who particularly are targeting IoT right now. We did licenses with 27 companies in Q3. 14 of those were companies who had taken a license for the first time, and most of those were in the IoT space. Most, not all. There is a fair amount of new companies coming into the space. Not all of them might turn into multibillion-dollar semiconductor companies, but some do. I can think back of companies who are significant licensees and revenue companies for Arm now, who once upon a time, we'd all never heard of, and they came from seemingly nowhere.
I think there's quite an encouraging trend of more companies looking to smarten the products that they're building, add intelligence. With the growing opportunity for intelligent devices, I think we'll see that continue.
Just on the merits of an EDA company and an IP company coming together, is there any merit in that if your customers consolidate?
Well, the EDA industry is pretty consolidated at the moment anyway. There are three big ones and a few small ones, but the big ones keep buying the small ones. That seems to be fairly consolidated. In terms of the efficiency for us, there are a relatively small number of companies who we partner with to make sure that whichever tools flow our customers want to use, they get a great result. There aren't obvious synergies of a combination of EDA and IP, or between us and the EDA companies. Don't want to talk about the past judgment on other IP companies. For us, I think the position that we have right now is quite good.
Okay. Thank you.
Next question comes from the line of Lee Simpson of Stifel. You may ask your question now.
Thanks so much for squeezing me in there. It's Stifel, incidentally. Two, if I could, actually. Just I guess you've had a chance to look at the budget for 2016. It's quite a departure from what we understand, at least the developments around Mbed Server. Really as you gap up that customer base from, let's say, just over 300 to 500-- Sorry, not customer base, the employee base, towards a sort of 500 run rate. Just trying to understand how quickly you get to that level. Is that stepped up opportunistically as in when you see the right acquisitions to come in, maybe for sales channel, software licensing experience? Is this done in respect to the sales line, how that matches to the software licensing that you're doing?
Well, as we think about the expansion of the software side around IoT, it's certainly not driven opportunistically. We're thinking about the platform of technology that is going to be required for IoT devices. We're not trying to solve every problem for IoT. We are doing what Arm does well, which is look for horizontal technologies that can solve a number of problems in a number of different markets. We have a view on what portfolio makes sense for us to have. We will look to develop that organically and through acquisition over the next couple of years. That is the approach that we're taking to it. It's not one of just opportunistically seeing a chance to grab a company and doing an acquisition. We're much more thoughtful than that.
Glad to hear it. Just also maybe on, we're hearing about some operators now rebelling against the use of OpenStack deployments. It just makes me wonder whether or not we're actually going to see a gap or a slowdown at some point in NFV rollout. Whether or not this could be something that you see coalescing now behind OPNFV instead.
Yeah, hard to call at the moment. I think NFV generally is still in quite the early days. I think there's a lot of complexity to NFV. I think when you get there, the nirvana of NFV looks great. The technical challenges of actually deploying it are quite complex. As you say, there are a number of competing standards. We have people deployed looking at that, working out how best to optimize the different software solutions around Arm and working closely with the partners in Linaro to make that a reality. So we're staying close with it, and I think we've got very good support across the industry for creating those optimized software solutions for Arm.
Mm-hmm. If there was a premature ending, let's say, in initiatives on OpenStack, you don't think that creates a sort of air gap in any rollout schedules?
I wouldn't have thought so, no.
Okay, great. Thanks so much.
Your next question comes from the line of Achal Sultania from Credit Suisse. You may ask your question.
Hi. Thanks. A couple of questions, Simon. First, on the penetration of big.LITTLE. In the press release, I see that you talk about 70 million units using big.LITTLE, which is about 10% penetration of your Cortex-A shipments. Is it fair to assume that what do you think that this number potentially could be long term, given what you're hearing from your customers in terms of adoption and the technology benefits that they're seeing with big.LITTLE? Could it be as big as, V8, not now, but at least in three, four years' time?
I don't think big.LITTLE lends itself to every market. Although interestingly, we were in a discussion with somebody looking at the data center space recently who was wanting to be able to scale up and down power based on workload. Of course, big.LITTLE is a great solution for that. It's not something that works for every market, so I don't think you'd ever get to 100%. We were talking earlier about the growth of octa-core. That is typically implemented in a big.LITTLE configuration. You're going to see big.LITTLE growth sort of tracking with octa-core to an extent. At the very price sensitive end, you're going to see lower core counts, and I think you'll see lower core counts for some time to come.
I think big.LITTLE certainly has plenty of room to grow, but I don't think you ever get quite a one for one match in that I don't think every V8 implementation will be big.LITTLE.
Right. Understood. One maybe longer term question. I think you talk about royalty rates still having room for improvement even after the big.LITTLE and V8 impact as you invest in new technologies like Artemis and Mimir. What are your customers exactly looking for in terms of functionality improvements from here on? Because I presume that the smartphones and tablets are already becoming very complex and advanced now. What exactly in terms of feature sets are your customers asking from you to deliver on in these newer technology platforms?
Well, it does depend on the market obviously, but generally speaking, people are looking to use new technology to deliver more performance with greater efficiency. Now, greater efficiency can come from improved system design. It can come from scaling a processor, just learning constantly about how power is used in end systems and optimizing around that. If you look at the way our product portfolio has evolved over the years, we have now a lot of technology around what we call systems IP, which is all the plumbing that you need to actually put a chip together, which is sort of hard to show on a block diagram, but there's a lot of complexity in it, in terms of how you actually put the key building blocks together. We're going to continue to evolve that.
We're going to continue to add functionality in media processing because in handsets, in clamshell devices, there's a lot of media processing going on, be it video, next generation video protocols, more graphics, more sophisticated user interfaces. Just fundamentally, in that space, there's a need for more and more processing power as your device tries to work out what it is you want without you having to explicitly tell it. In the kind of mobile computing space, we see the need for more processing power, and more power devoted to the kind of human-user interface side of that. Obviously, in other markets in networking, it's all about efficient throughput. It's all about flexibility over the software stack so that equipment can really benefit from SDN and NFV.
In the kind of data center end, it's all about the ability to integrate the right size processes with other acceleration features. People are looking for technology solutions to help make that a reality. It's very broad. Overall, it's about right-sizing the performance, enabling the right compute engine to do the right task, and continually delivering more performance with better power efficiency.
Thanks. Just one clarification. I think you mentioned the 75% exit run rate for Armv8 penetration within smartphones. Was it by the end of this year or next year, the 75% exit run rate?
By end of this year.
Yeah. Got it. Thank you.
Thanks, Achal.
Your next question comes from the line of Ajay Anand from Mizuho. You may now ask your question.
Thanks. It is actually Ajay Anand from Mizuho. Simon, I had a question on the networking market. I think you said 20% growth this quarter, which is a little bit slower than the 30% that we saw in Q2. I appreciate it is difficult to draw trends from quarterly numbers, but I wondered perhaps it is a case of tougher comps, or is there any other reason which explains the variation? Could you talk about how you expect the growth to trend say over the next 12 to 18 months? In the press release, you talked about a major new licensee starting to ship chips to the base station market. Would you expect similar sort of growth, a 20% type of growth, say, over 2016, or would you expect faster growth? Thanks.
As you say, from one quarter to the next, I wouldn't get too hung up on whether, 20, 25, 30. The trend is upwards on that. Seeing more semiconductor devices announced is good. This is a market that is obviously got a smaller supply base than some others, so there aren't hundreds of companies in this space like there are with embedded, nor is this as mature a space for us as it is with, say, smartphones. We are very pleased with growth. We think the long-term trends for this market bode very well for Arm. Our adoption will grow. The deployment of Arm-based semiconductor devices by equipment companies will continue to grow.
We show the analysts, say, the color chart, the SMARTIE chart, if you will, which talks about equipment companies using Arm. We will provide regular updates for that, which we think will take us towards our goal of 45% market share in 2020. Yeah, one quarter to the next, that is going to be lumpy. From what we can see, progress is very strong.
Thanks.
I also have a short
Your next question comes from the line of Brett Simpson of Arete Research. You may ask your question.
Yeah, thanks very much. Just a couple of questions on v8-A royalties. You talked about 250 million units shipped in the quarter. Back of the envelope suggests that translates to about 40% of PD royalty sales. Does that sound about right? Can you give us some indication what the average royalty rate for v8-based chips was in the quarter, and what scope do you think you have to get this rate to rise much from here? Thank you.
In terms of percentage of PD royalty, that feels high to me, to be honest. We haven't worked that out. I don't have that off the top of my head, but that feels high. In terms of royalty rates of v8 chips, excuse me, we haven't disclosed that. The short answer to your question is no. We think that the blend of average royalty rate for Arm will go up as a greater proportion of chips move towards v8. There's scope for that in more deployments of v8a. We have some scope for that in the deployment of ARMv8-R as well. We can see some licensing going on there, too. We think that'll move us towards a higher blended royalty rate over time.
Okay. Simon, just maybe a follow-up on the licensing side. You typically launch a new architecture every four to five years, I'm just looking through v8, launched in early 2011. Is it fair to assume we're going to see, at some point fairly soon, a new architecture, v9 architecture? If so, what do you think Arm needs to add in terms of feature sets going forward to drive uptake on that next architecture? Thank you.
We're always looking at what architectural features we might want to add. It is very important that architectures are stable because we want to make sure that we build an ecosystem that supports an architecture, if it changes too frequently, you end up with a lot of fragmentation. When we introduced v8, we announced that quite early in its cycle because of the architecture licensing that we're doing, the move into more of the enterprise space, the need to get the software ecosystem on board. Having done that, I think that potentially there could be a longer time or to that sort of four or five-year cycle before we introduce the next variant of the architecture.
We've always got work going on looking at how we add efficiency and all the things I was talking about in answer to an earlier question, how we can do that from architectural features as well as through implementation. Just broadly speaking, we're looking at the evolution of mobile computing. We're looking at the evolution of networking and features that are going to support doing that in a more efficient way. When we have details to talk about that publicly, obviously, we'll be out in due course.
Maybe just lastly, a housekeeping question. What portion of licensing sales came from backlog versus terms in the quarter? Thanks a lot.
Yeah. In Q3, that was about 60%, that's at the upper end of the range. Again, from one quarter to the next, that does move around in the sort of 40%-60% range.
60% from backlog?
Yeah.
Okay, great. Thanks very much.
Thanks, Brett. We're on to the last question now.
Once again, if you wish to ask a question, please press star one on your telephone keypad and wait for your name to be announced. Your next question comes from the line of Robert Sanders of Deutsche Bank. You may ask your question.
Hi, guys. Thanks for squeezing me in. I think this relates to a question from before, but I didn't quite catch the answer. It's just regarding wafer pricing and how that's impacting die costs and therefore your royalty revenue. When you think about 14 nanometer foundry pricing, it's probably about as much as double as 28 nanometer was. I'm thinking that that could be quite a meaningful driver of die cost and therefore your royalty. Have you actually tried to quantify this in terms of how it impacted your Q3 revenue? I'm just interested to think about how sustainable that is. Second question would just be if you can actually disclose how many of your customers in the quarter represented more than 10% of sales. Thanks.
In terms of wafer pricing, certainly every time there is a new process introduced, it's more complex than the last. There are more processing steps, the price of a processed wafer goes up, there is industry data out there. If you look at the GSA, for example, they publish reports on wafer pricing trends. There's quite a lot of data in there. FinFET is definitely the most sophisticated process that the world has ever seen, one would expect wafer pricing to be higher. How much of that gets passed through to the end selling price is down to our customers, and we have no control over that. What we're seeing at the moment, as I said earlier, is an uplift in ASPs more around the functionality that's being introduced more than anything else.
I would expect wafer pricing to go down over time, just like anything else. It tends to cost more when it gets introduced and is new, and goes down over time. We don't really factor in process wafer costs when we think about how our ASPs move. ASPs typically are more driven by the functionality of the chip, rather than the bill of materials.
On your question on customers, we don't give the quarterly data, but there were none in 2014, and for 2015, it'll be in the annual report.
Okay, thanks a lot.
There are no further questions at this time. Please continue.
Okay. Well, if there are no more questions, thank you everyone for joining our call today. We will obviously be back in February with our full year results. For those of you who are coming to our analyst event at TechCon, we will see you out in California on November the 11th. Thank you for your time today.
This conclude our conference for today. Thank you for participating. You may all disconnect.
Thank you, Stephanie.
You're welcome. Have a nice day.