Status Update

Dec 18, 2013

Slides

Thank you for standing by and welcome to the ARM Analyst Conference Call. At this time, all participants are in a listen only mode. And I must advise you that this conference is being recorded today, December 18, 2013. I would now like to hand the conference over to your speaker today, Ian Thornton. Please go ahead, sir. Okay. Thank you. Good morning, good afternoon, everybody, and thank you for joining this conference call on arms opportunity in the Internet of Things. Joining me on this call is Charlene Marini, who's ARM's VP of Marketing for Embedded Technology, which includes this emerging market. Also on the call is Francois Munier, who's the Semiconductor Analyst with Morgan Stanley. Francois was recently the keynote speaker at a Internet of Things World Forum. And so I thought he'd be a good person to run the fireside chat element of this presentation. In a moment, Charlene is going to run through some of the slides. And if you haven't already downloaded these slides, they're available from www.arm.com/ir. And we won't be taking questions directly from the phone, but if you'd like to e mail a question through for Francois to ask to Charlie later, then please e mail investor. Relationsarm.com and we'll sort out the technology at this end. So with that short introduction, I'd like to hand over to Francois just briefly to give a few opening remarks. Francois? Yes. Thank you, Ian. It's Francois. Yes, of course, we are very excited about the Internet of Things opportunity. Yes, of course, it's going to be very big. And it's interesting to see so many companies talking about it, but it's even more interesting to see companies doing something about it like ARM. It is very important for ARM. And according to our own investor survey, ARM is the most exposed semi company according to our clients. So that's pretty important for ARM for the next 5 years and beyond. So I will now hand over to Charlie. Thanks, Francois. Hi, everyone. So if you have the slides in front of you, I'm going to move right along to Slide 2 on the cautionary statement and suffice to say, please take sufficient time to familiarize yourself with this slide. And then with that, I'll move on to Slide 3. And really setting the context here for us and I think broad agreement in the market, IoT span sensor to server. And across this span really existing hardware and software paradigms are being disrupted. We're going to talk a lot today about the nodes that are really an interesting area of IoT and one of the main drivers, if not the main driver of IoT. But it is going to have impact all through that network back to the server. And really we are in a unique position in that the ARM architecture can span this entire range of applications and performance points. If we move on to slide 4 then, the trends driving IoT. From our perspective, certainly the cloud and the ability to connect to processing resource cheaply and easily for analytics is a very important trend within this context. The other one certainly is mobile Internet and the usage models and the business models around the mobile Internet that anytime, anywhere access through your handset or tablet. And along with that, the app culture that's built up, particularly with the rise of smartphones. And if you look at the types of businesses that are being developed today, usage models, a lot of the innovation that's occurring in the mobile Internet world is around apps and that's going to transfer and is transferring into IoT and Embedded and of course decreasing hardware costs. Then on the next slide, an example of those decreasing hardware costs. And this is an extreme case, but I think it's very important to note what can be achieved in high volume in the embedded space with something like the Cortex M0 microcontroller. And so price points that really make broad deployment of these technologies applicable. Moving on then to the next slide, some examples of early IoT types of implementations. The ARM farm, 1 on the left here is an agricultural field equipment type of model and device. And you can see that again it connects to your mobile Internet, the tablet, the smartphone. The idea here is to be able to accurately control irrigation systems through sensors that detect humidity, rainfall and automate then the irrigation systems. The one on the right Scanadu Scout is a medical monitoring device. So we've seen a lot of emerging devices this past year around medical monitoring. But this one is very much intended to be at the high end of that range. So measuring blood oxygen, heart rate, blood pressure, stress, respiration, so all the applicable vital signs. And then of course pairing with your smartphone to bring that data out to the user. So it is already emerging. IoT in some ways is here. But of course, if we move to the next slide, a lot of what the industry is talking about is what the Internet of Things can become and what the vision can be and will be as we move forward. And so certainly various projections, I put the Gartner projection here 26 1,000,000,000 installed units by 2020. I think everybody has seen kind of the Ericsson and Cisco numbers of $50,000,000,000 and 1,000,000,000,000 in those sorts of timeframes. So it is a market that is emerging and it's very difficult to pin down. And one of the reasons of that is certainly that the characteristics of this market that I wanted to bring out. Unlike the mobile market or the consumer electronics market, where we've been able to look at one device type and project volumes based on a single device type. IoT is very similar to classic embedded and it is a long tail market. And it's an even extreme form of classic embedded in that with those trends that I mentioned earlier, we're going to see an exponential increase in the ability to create different hardware and then apps that sit across different types of hardware. So really just an explosion of both hardware and software as you can see here. So one of the things we're certainly looking at at ARM is how do we enable this market. Low cost, tiny sensors are very important and as the Gartner numbers show, we're getting there on cost. But there are many other things that we feel need to happen to really enable this market to reach its full potential. So in addition, secure standardized Internet and web to the tiniest devices. What's very important here is no matter where you are in this continuum from the tiniest of sensor back to a server, we need to have a standard framework that decouples pretty much the devices and the creation of those devices from the software and the applications. And standards are going to enable that. Authentication and trust, so having billions of devices out there, can we trust these devices? Are they easily hackable? How do we know that the device we're addressing is the one we want to be addressing if we're back in an application sitting on a server. Universal data model semantics, this is something that's not talked about too often, but just like having a standardized protocol and way of addressing from a network layer communications protocol. We also need a common set of ways to address the resource in objects. So a good example of this is a temperature sensor. How does an application know when something says it's a temperature sensor that its resources are that it can give the temperature within such parameters and how do we know that maybe that also can provide humidity reading as well. So in the World Wide Web context of the Internet today, this is done in XML. And you have a person sitting at the other end of a web page that can interpret that. When you're talking about small devices out in the field, the machines themselves and the server itself on the other end needs to interpret that data. You won't have the human in between that's providing context and interpreting that data. And then easy and open development. So as I'll talk about later, this trend in terms of an app culture moving into embedded means that a typical embedded developer is not going to be looking at devices anymore and saying, I need this type of processor, I need this much memory. Instead, they're going to be coming in and saying, I want to create this type of service. I want to create this type of system. And they're just going to want to start developing that service and system. So if we move to the next slide then, this is really how we view the IoT architecture and I'll spend a few minutes on this slide. Certainly, on the left hand side, the nodes and devices need connectivity linking back to that network. And this is again where standards become very important. You want to be able to deploy hardware and for that hardware to be accessed by multiple types of applications and not to be locked in to one type of service and application. And then the second point is certainly when you deploy nodes, you might not know the network topology ahead of time. So if you think about how we deploy systems today, they're planned out by a certain topology and that means that you know I'm going to have this hubs or repeaters or access points and gateways and my network is going to interact and be fairly static. These networks won't be static. And so we need to decouple the topology planning from the deployment. And so that leads back to the need for common standards. Again, you don't want to do that in different ways based on different deployments. And if we look from an ARM point of view of what we're doing today and how we're serving this market, certainly the processors, ARM Cortex family spans from these very small nodes through to the server, the networks in between, the access points and the gateways. SensiNode, the recent acquisition from ARM provides software layer on top of this for standards based end to end security web data objects and management. So on this diagram, that's really the connectivity between these nodes. And then the left hand portion of the software sitting on the server, so the security, the communication, the discovery, the way in which you manage those nodes and communicate with those nodes. And really what we're focused on is enabling this little data, the nodes and the connectivity to them bringing that back up through to the data storage and analytics applications that are going to be sitting in what we're referring to as big data. And we see that as the biggest challenge in IoT is really getting that little data back up to the processing engines on the server side in terms of storage and analytics, so that the data can be used efficiently. And one of the things that we believe that if you move on to the next slide is it's quite important to have intelligence in the nodes. Nodes need to be able to interact with their network. They need to have some autonomy. They can't just be dumb devices feeding all of their data back to the cloud. Data will need to be filtered. There will need to be some event processing locally. Decisions might need to be made locally. And so that really leads us to the need for intelligent nodes at the right cost point and with the right security components on them. And so an example here of ARMY Heart Cortex M0 plus solution and this is just one example. There are others from our silicon partners of course. But the free scale kinetis KL2 really very quite tiny device. So less than 2 by 2 millimeters squared. And this can really go anywhere. When we move to the broad ecosystem, I've highlighted some of our silicon partners below. You'll find that there are a variety of solutions and in fact over 2,000 SKUs of ARM based microcontrollers in this space with varying cost points, varying performance points. But at the extreme, the great thing is we can reach the smallest of nodes with the necessary intelligence to enable these intelligent sensor node networks. Moving on to the next slide. Then with MCUs really the center of intelligence in this market, the other important points are certainly the connectivity. And so the radios that enable the connectivity and sensors and actuators. The ability to sense from our environment, but also act where needed in that environment. So with at 2012 over 2,000,000,000 Cortex M based devices shipped and with a focus in these areas. And we see these areas as the main growth areas for IoT in that context of the end nodes. Moving on to the next slide then and talking about the end nodes and the development and trends there. As I mentioned, the end nodes will be a combination of MCU capabilities, sensor and radio. With these three elements, there will be an evolution to reach the minimal size and cost in PowerPoint. So today, these are mostly discrete solutions. We have MCU vendors, vendors for Bluetooth, Wi Fi and ZigBee and then of course also MEMS or analog sensor. The trend that we see moving forward is certainly multi chip packages or multi die packaging where these components can be integrated into a single chip. We don't necessarily see integration on the same die. Several reasons for that. Certainly, the fact that analog and flash don't scale down as fast as Digital Logic, but also things like MEMS where they go through a very different process. And because of that most of the MEMS value chain are IDMs who own their own fabs. So there are experiments taking place with kind of fabless MEMS sensor companies and people are trying different business models in terms of where intelligence is integrated. And I think experimentation will continue as people look to address the trends in this market. Moving on to slide 12, SaaS. If we're looking at the evolution in IoT and how do we really break down the billions as I call it and provide some clarity, Certainly devices will be increasingly connected. And the question there is what is the attached rate of connectivity? And it will depend on the business model and the end to end platform development. A good example of this is lighting, which I'll talk about in another slide. But it is primarily a replacement market. Some solutions already have connectivity today, it might not be standards based connectivity. And so moving from the proprietary legacy connectivity through to a standards based connectivity. A lot of that will depend of course on the economics and in terms of the pricing of connectivity solutions. Embedded Intelligence for this standardized secure Internet of Things, 32 bit intelligence will be a requirement. And so that transition from 8 and 16 bit that we've been talking about for several years now to 32 bit intelligence, IoT is an accelerator to that trend. And the rate of that transition will very much depend on the deployment of capable secure intelligent nodes. Gateways, controllers and hubs. So this will really depend on the deployment types and topology choices. If legacy systems stay in field longer and replacement cycles tend towards only slight upgrades to those legacy systems, not as much intelligence in the end nodes, Gateway controller hub volume would be higher than if there's a rapid shift to IP to the node, which enables very simple end to end systems and less need for controller hub gateways within the systems. And so on the right hand side some projections from IHS on shipments of new connected devices in various markets. And this is fairly in line with what we're seeing in the good split between consumer applications and what I'd call broadly commercial applications, so medical and industrial. A few key areas in each of those, which I'll talk about now moving forward. So then moving on to Slide Slide 13, there on Cities. This is one of the areas where some applications are coming to the floor and certainly smart parking is an area that has had quite a bit of promotion with deployments in at least in the U. S. And San Francisco and San Diego. Streetline is a company deploying networks in San Francisco using DUS Network, which uses a Cortex M3 solution for the intelligence in that small end node. But other applications like car fitness, the ability to tell when a car needs a smog check, when the tires or pressure might come to a dangerous level and need repair. And of course, building automation and lighting, these are two areas that due to some good push by regulations and government are seeing rapid evolution. Moving on to the next slide, example of a street lighting application. And this is an actual deployment based on the SensiNode product today. And really to point out that again as we've talked about it's a combination of end nodes. The ability of those nodes to self route and interact with one another a link then back to the cloud this case in this case cellular and then processing in that cloud for actionable information and automated management. And looking through this, as you can imagine something like 50 nodes or 100 nodes will need that gateway to compile information and send it back to the cellular cloud. Slide 15, an example of commercial lighting. And so this is in building lighting in a commercial context. 3 to 4 types of devices that we would see sensors and in this case occupancy sensors and environmental sensors And then also the ballast themselves being connected for automation. And so again some projections and these are all potential Cortex M sockets as we see that market. So whether it's the connectivity in the sensors, whether it is sensor analysis, so a type of sensor fusion or its connectivity within the light bulb itself back to an access point. And then of course with that access point backhauling to an end system. And a partner using ARM based solutions is PhotonStar LED. So again, an example here of the ballast light bulb, the access point and then being able to access that information through an array of screens, whether it's a handset or tablet, laptop, PC. The other area we're seeing a lot of activity is in the home, especially in the U. S. And European markets. So a projection here of 50 Internet connected devices to be in the home of an average family of 4 in 2020. And so that's a fivefold increase from what's projected to be in home today. And that's quite a bit of growth. And so where are we seeing that growth? Well, if we move to the next slide, this is really where you do see the power of this combination of mobile Internet and apps and cloud. Smoke alarms, security cameras, smart meters, toys, lighting again and health monitoring, active areas throughout. And so just some indication, of course, smart meters have been talked about for quite a while. But those really being kind of the cornerstone of a lot of the in home energy management solutions that are evolving. 25,000,000,000 smart appliances in 2020 and 12.2 1,000,000 home automation systems to be projected in 2017. And so seeing hub type of devices like the NEST of course, seeing almost do it yourself type of platforms, smart things is a good example of that, allowing people to program their own homes and automate their homes in the way they want. And of course, even things that are popping up daily, I put a picture here of Haiku, which is app for groceries of being able to scan barcodes and manage what types of products you have in your home and they're going to implement an online then grocery ordering system as a way to then link that to an automated delivery service. So daily, I think we've all seen new and exciting services and applications that are being built up in the home. And home is probably the greatest example of why we need standards for the proliferation of devices. A good example is, if I put in alarm system with a certain service provider, if I change service providers, if I want to add applications on top of that sensor, I don't want to have to switch out the hardware. In an ideal world, I should be able to add services and those services can access that hardware. But of course, we're not able to do that today. So Slide 19, wearables quantify itself, 2 very popular trends. We're seeing quite a bit of design activity in these areas and again the diversity. So, on the left hand side, I think we've all seen the watches, the glasses, the health monitoring patches. But it's things like jewelry, posture monitors, clips, even fabric that will be emerging. And so an estimate of 70,000,000 wearables shipping in 2017. Based on I think how the market evolves in 2014, We'll see how big a trend wearables will be. Some of the use cases, I think, are quite compelling. Some of the use cases still need a bit of work to really bring out and harness the power of these devices. Moving on to Slide 19 and kind of then bringing it back to how do we really enable this wide variety of end nodes? How do we get to those 1,000,000,000 and trillions over the next few years. And as I alluded to in the beginning, it really is about open platforms. If you look at the handset and how that market has evolved, the ability for someone to create an app on the weekend overnight have a new business idea has been very powerful. And this is moving into the embedded market. So instead of a very corporate focused view on embedded where the hardware is chosen, the software then is made to fit exactly with that hardware and the context in which it's going to be deployed is put into that market analysis and marketing requirements document for that device. Moving to a world where someone will say on the weekend and I would argue we're kind of already there with things like Arduino and Raspberry Pi, I want to create this and just go off and create it. And for that, our ARM embed device development platform will be a platform to enable this plug and play on both the hardware and software side. We had recent announcements for instance with Nordic to enable Bluetooth LE connectivity. So and another one with u blox to enable cellular connectivity. So to be able to really create an MCU based platform with the types of connectivity, the types of sensors you want to attach to that and then the libraries that go on top, so that you can start programming within days instead weeks, which would be the normal course of events, in embedded. And so this really will allow broad experimentation and development. And of course, if we move on to the next slide, that's being mirrored by new ways to develop and grow businesses. And so things like Kickstarter, but even away from Kickstarter, the capital needed to create these new types of businesses and services is fundamentally at a different level than with the old world of single device categories, expensive hardware, more closed systems. And so I think we'll only see more and more of this type of activity. And that will transition back into the corporate world as well where within some of the industrial companies that we've seen, white goods companies that we've seen dominate in the past, this more innovative quick turn rapid development type of approach. And so Slide 21 to sum up, we've really looked at the market and trying to understand kind of where the simplicity will come out, where we feel those billions will be in the next few years. And so looking at that, our estimate is $3,000,000,000 of these end nodes in kind of the 2017 timeframe. With some upside for the hubs and intelligent systems, but that's very difficult to predict as we see how the fluidity between the network, the end nodes and the servers transpires and evolve. And of course, wearables being an identifiable trend that we think will have volume and will continue to progress in the next few years. So with that, I'll hand back to Francois. Yes. Thank you very much. And we have I think about 20, 25 minutes now in the call. So let's actually talk about the last page 21. If I do the sum, it's around 30,000,000,000 chips in 2017. How much is genuinely a new opportunity for ARM compared to the usual slides you're giving? And how much is a reclass? So the $3,000,000,000 and then add kind of the $20,000,000 and the $70,000,000 of new opportunity, the $20,000,000 $70,000,000 So the embedded connectivity, the hubs and intelligence systems and the wearables, we took out as new opportunity. Okay. Thank you. In terms of royalty rates for all those chips, is it going to be the usual 1% royalty rate on relatively low ASP? Or is there any way you can charge a little bit more maybe at the fixed royalty rate rather than percentage? Good question. Certainly, the highest volume nodes will be those very low cost nodes. But I think it's important to take a broad view that there will be different types of nodes. So again kind of to that end view of the long tail market. And so there will be higher end nodes that require more intelligence. And so we'll see a range of cost points in ASPs even for these nodes. And then of course looking back at the hubs, the intelligent systems that's a different type of class of systems typically a Cortex A system. So that will have higher ASPs even than those or highly intelligent nodes, I should say. So I think we'll continue to see kind of that royalty rate within this context and the variety will flourish within that. Okay. In terms of technology, it looks like it's more like a reuse of Cortex M0, which has been developed for quite a long time and maybe older Cortex A. So basically, the incremental costs for you are very low. And therefore it's likely to be extremely good in terms of return on investments compared to new opportunities like I don't know servers or networking which are coming along? Right. And I think that's the interesting thing about the embedded market. The embedded market at the base thrives on stability and having solutions that might be in market plus 10 plus years going on. And so if you think in terms of these end nodes, certainly they'll be infielder in 7 years, 10 years. And so the ecosystem that's built up around the cores, I mean, as you mentioned, Cortex M and existing Cortex M is going to need that longevity and it's had that longevity in the past if we look at ARM7 and ARM9 and so moving to Cortex M. So we've shown that we can have that longevity in the embedded market. And as you say, building billions upon that framework, I think is an interesting and very exciting opportunity for ARM. Now moving to the wireless part of things because I think a lot of those nodes will have to be wireless. So of course, there is ZigBee, there is Bluetooth, Bluetooth Smart, there is 6LoWPAN, which you're working on with your recent acquisition. What do you think is going to be the most prevalent technology or it's going to be a bit of a mixed spot? It will be a mix. What we're advocating and what we feel strongly is that the link layer, so whether it's ZigBee, whether it's Bluetooth, whether it's Wi Fi, they'll have different characteristics, there will be different needs depending on range and these types of things. But we would like everything to be IP based. So having IP over ZigBee, Bluetooth, smart, etcetera. And so 6LoWPAN is a great technology that enables IP all the way to the tiniest of nodes by providing a very minimal code size implementation of IP. Okay. So it's basically something which standardizes the communication to the node and to some extent doesn't matter if it's whatever wireless connection. That's what you would like to see? Yes, exactly. As long as there's the same protocol way to speak with that end device and we can achieve that by having IP throughout. Okay. So also is there anything you can do as ARM with as part of your ecosystem to make all the APIs more standard and all the glue around the sensors more standard as well. Is there anything you can proactively do to make IoT happen quicker and bigger? Well, that's always hard to say ahead of time, but I can tell you what we are doing and certainly in hopes that it will enable IoT tap in quicker. So back to your previous question, we do feel that for IoT to really take off there needs to be a standards based approach. And so having IP to the node is something that we're strongly advocating and we'll continue to strongly advocate. And also just in general adopting standardized solution. So the IETF and other standards bodies have created and endorsed several standards that go along the stack and IoT. So 6LoWPAN is one example. CoAP is another one. So once you get to the application layer, then how do you have a common way of communicating at the application level? Security, how do you ensure that you have a common prototype of protocol security? So by working with IP and web based standards, we feel that this simplifies that end to end configuration, management, deployment, accessibility of those And that's very much what we're advocating. And the SensiNode product line implements those standards based approaches. So if someone wanted a shrink-wrap solution to help them deploy in a standards based way, That's really what SensiNote enables. So when I hear the word standardization, I can also think about the word commoditization, which is probably needed on the hardware side for IoT to happen. So do you think commoditization is a good thing or a bad thing? I mean, ultimately, we need to have the ability to plug and play these types of devices. And that is going to mean, as you said, standardization. And the price points in these markets, you will need commoditization of some parts, but you also need differentiation. So the devices will need to have common programming frameworks, but certainly I would prefer to have a choice of the temperature sensor and which one is more accurate or which type of end device is going to provide the more accurate data for any given application. And that's just one example of the differentiation. So it's just a common thing as new markets grow, we need to have areas of commonality and whether you and that might mean commonization in some parts, but that enables innovation in the other parts. Okay. So another thing that my clients talk about is basically the security issues around the Internet of Things where basically it looks very good on paper, but people are in general, especially for their personal data really scared about someone hacking and getting access to their data, access to their home because the chips or the wireless connections might not be secure enough. So what can ARM do to improve this and actually make people more confident about the Internet of Things? Yes. I think there's a lot that can be done there. Most of the technologies exist and a lot of the technologies are actually being and a lot of the technologies are actually being implemented by our partners and by the value chain. So for instance, in end nodes having key exchange mechanisms and encryption mechanisms is something that we're advocating. And many of our partners are already implementing those capabilities in the hardware. The sensor node software takes advantage and actually demands that type of hardware capability. So one is advocacy and second is certainly building up an ecosystem and enabling various parts of the design chain from our silicon partners to our OEM partners to easily implement and access the necessary components for security. So in terms of health care application as well, I think people are quite scared, but the privacy. So is it something which that the consumer general is ready to accept? Or do you basically we need to I mean we as in the IoT industry need to teach consumer that everything is fine and they will be safe? I think there are always be varying views from a consumer point of view, and it might be generational. That could be one out, it could be cultural. So certainly a very complex problem, but what we can do is really education and being very diligent about looking at different use cases and what's required from different use cases in terms of trust and security. So you mentioned health. A lot of the health applications, I think, at least in the 1st or second generation are really going to be chronic management and chronic monitoring. And I think the view is that everyone sees these data going back to the cloud and sitting on a server somewhere. Well, in reality, it could be that quite a bit of that is local. It might be a local cloud that I have because I don't feel comfortable with it going back. I want to own that data and keep that data and I only want to enable my doctor to see it. And so that could be one case. Another case might be the service provider. It might be that I trust a hospital or medical practice to have access to that. And if they're providing the service, then I feel comfortable with it. But I might not feel comfortable with broad based service provider having access to that data. So what model am I going to be able to access as a consumer? And does that fit with my desires for who I want to have access to that information. So I think there are several ways for it to work out and I don't there's a clear view of how it will work out. Do you have any view on who will own the data and how people could actually monetize the data that will be generated by those tens of billions of devices? I know it's outside of the scope of ARM, but if you have any view that would be very interesting. I guess I would pull that back because that's a fairly broad question to say in this 1st or second generation, I think the data will be fairly siloed until there is broader cultural agreement alignment education to your previous point on data and how it will be used and who can gain access to it. And there needs to be mechanisms for people to express their preferences and to ensure those preferences are adhered to. So I think we're kind of far from that. And so without that, I think you go back to the model where this is my healthcare data. I know that my healthcare provider has access to it, but I know that they're not enabling anyone else to have access to it. I'm not letting my data just go into a broad cloud somewhere. Okay. Now talking about all this data, which will have to be managed and stored somewhere, do you expect to see any ripple effect into the server market? And obviously, do you expect, I'm sure, to participate in this market taking market share from Intel? And is it something which is suitable for your processors? So the overall answer is yes. And I would expand on that to say there's going to be transformation through the network to the server. And it's just a very exciting time in that, how when you do have all of these nodes in these different networks and topologies and you do have the different service models, There's then quite a bit of a mix. And so you say, well, how much processing do I need at the edge of a network then? How much local processing do I need? And do the types of processing that I am undertaking change, I. E. They need more network throughput, they need more storage capabilities. And so this really is disruptive and you'll see that there are different proposals of how things might evolve. But in the end, I think it's a great opportunity and we are going to see different paradigms not only in the server area with different types of clouds, big clouds, small clouds, local clouds, public clouds, but also in the network. Okay. So basically, that could be a change from today. But that's probably going to take quite a long time. Now if I just come back to the nodes themselves, of course, there is ARM which dominates today, but you're not the only one to talk about the Internet of Things. MIPs and Imagination are talking about it. Even Intel wants to participate. What is your competitive advantage? And do you expect to dominate as much as you dominate today in the mobile market? So the value around ARM has I think a universal appeal across markets. And so it's the same for IoT. 1st certainly is the business model and the ability to create a technology that is common for a software developer, but enables a broad set of end devices. And as we know with IoT, we are talking about an amplification in terms of need for those different types of devices, but with a common addressable set of software components. So I think that certainly plays to the ARM business model. And then certainly it's also the ecosystem that we've built up. Arm has been strong in the embedded space really since we began, so over 20 years history there. And the tools, the software that has been built up around the ARM ecosystem are all basic things that are enabling this higher level accessibility and programmability for a broad set of developers. And so it will become I think more and more about the ecosystem actually, in that in order to rapidly create new devices, in devices, in order to rapidly create new software and systems, that basic stuff needs to be there along with some higher order abilities to mix and match and provide rapid integrated components. So as with any new market, I think that there are opportunities for entrants and competitors. But I see ARM is very strong moving forward in this space. Very good. Maybe one last question about microcontrollers actually. Of course, you're very strong in 32 bits, but some people are wondering because those nodes I mean, some of those nodes would be really low in terms of processing power. Why there is maybe there could be room as well for some of your competitors in 8 and 16 bits? Or is it because Cortex M0 is so cheap anyway that it doesn't make any sense to use 8 or 16 bits anyway? I think the pricing is critical, but the other critical component is back to the Internet of Things and why this is different than traditional embedded. And in traditional embedded, you would deploy something in a fixed network, it would have a fixed set of parameters. In IoT, you're deploying something and it might have a different network, it might be used in a different way, a year after its deployment, 2 years after its deployment, 3 years after deployment. And that's pertinent to the tiniest and cheapest of sensors as well as to more intelligent nodes. So that is going to drive 30 bit, 2 bit intelligence. And certainly when the barrier to adopting 32 bit from a cost point of view is so low. And back to your point on security, to have security in these types of nodes, you need to have some intelligence. Okay. Very good. Thank you. Okay. Thank you very much indeed, Charlene. And thank you, Francois, for leading through the Q and A there. To everybody else, I hope you found this call useful. If there are any further questions or any feedback, please let me know and we'll get back to you with answers to questions as soon as we can. So in the meantime, I hope you have a very Merry Christmas and relaxing break. And we wish you all a very successful New Year. And we'll be back with you for ARM's Q4 and full year results on the 4th February. And with that, we'll hand you back to the operator. Thank you very much indeed everyone. Thank you. That does conclude our conference for today. Thank you for participating. You may all disconnect your lines. Thank you.