Hello. Great. Good morning, everyone. Welcome to CEVA's first ever Analyst and Investor Day. Delighted to have you all here. Before we start, I just have a quick video we're going to show you, we made just to celebrate the 10 billion CEVA powered devices to date, and give you a little bit of information about some of the markets that we power. Without further ado, here's the video.
Dream of designing a device. A device that will change our lives. Imagine you can make it happen, make it real, make it fast and powerful. Make it intelligent and make it learn. Make it analyze and predict. Make it fun and creative. Make it look sharp and gorgeous. Make it friendly, user-friendly. Make it efficient and productive. Make it smart and connected. Make it prevalent and extensive. Make it secure and protected. Make it safer and autonomous. Make it sound clear and pure. Make it speak. Understand. Make it free. Truly wire-free. Make it sweat and stay in shape. Make it grow and make it nourishing. Make it all happen with CEVA. CEVA licenses ultra-low power signal processing platforms and AI processors. CEVA's IP powers more than 10 billion devices worldwide. Dream it and make it with CEVA.
All right, there we go. Good morning everybody. For those who don't know me, my name is Richard Kingston. I run market intelligence, investor and public relations for the company, and I'm delighted to have you all here today. Hopefully learn a little bit more about CEVA. For those of you who know us a bit longer, we've been around for about since 2002. CEVA was spun out of a semiconductor company as a DSP licensing division or, was a DSP licensing division within a semiconductor, is now a standalone company. I think over the last few years, we've showed some tremendous growth. The company has diversified into a lot of new spaces and market opportunities, and I look forward today to showing you some of these new markets and areas that we are licensing into.
Before we start, just make you aware, we have legal notice for forward-looking statements in the presentation. The first thing we did today, we announced there's a milestone here for CEVA in that we shipped our 10 billionth CEVA powered device during Q4, which is a pretty phenomenal achievement for a company of about 350 people today. Over the last few years, we've shipped in about a billion devices annually. We're in many of the products that you guys are probably holding in your hands or in your pockets today. I think in the next few years you're going to see CEVA's extensive devices and millions and billions of units continue to increase as we see more smart and connected devices in the world, which is the area that we try to power.
Just before I look at this slide, I just want to say that the purpose of today is not about 2019. It's to try and help you guys see further out down the road on where CEVA's going. Most of the timeframe we're talking about today will be towards 2022 or so, and each of our presenters will give the market opportunities towards that timeframe, what sort of products we're developing aiming for that. Yaniv, at the end of the day, will wrap everything up and give you sort of the forecasts and expectations where we believe that CEVA can be if the market drivers that we predict are happening will go that direction.
We're in our quiet period right now. As a result of that, we won't be talking about Q4 or 2019, but we will be happy to do that when we report our earnings on February 13th, I believe. We were at CES last week. We made some new product announcements, new customer announcements, and even today we had a new press release. We've opened up a new R&D design center in Bristol in the U.K. Bristol is very well known for digital signal processing technologies, and there's a lot of semiconductor companies based there, like ST, Broadcom, Infineon, Imagination, and so on. It's a great new opportunity for us to tap into the R&D base that's based in the U.K. and hope to accelerate the development of our signal processing cores and also our AI processors by leveraging the expertise that's there.
Some of the other products that we announced with customers in the CES timeframe, new Narrowband IoT and Cat-M chip powered by our DSP with Nordic Semiconductor. Sonova, who's the world's leading developer of hearables, are using our Bluetooth in all of their new hearable product range, the SWORD chipset. Many new design wins with the likes of Atmosic Technologies and Bestechnic, who are using our technologies to get into new markets and some very exciting opportunities and products there. CEVA itself, we announced two new products as well at the CES. The first was WhisPro, which is a new neural network-based voice trigger technology, meaning in addition to licensing DSPs from CEVA, we now offer you the ability to license software from us as well.
This is part of our strategy to verticalize in some of the key markets so we can generate more licensing and royalty revenues. The second announcement we made was around a new DSP, what we call a multipurpose DSP for doing digital signal processing and digital signal control. This is an area that Moshe Sheier will talk a little bit more about when we have the sound presentation later on in the afternoon. What's on the agenda for today? At 10:00 A.M., which is in a few minutes from now, Gideon Wertheizer, our CEO, will give you an introduction to the company and some of his strategy and thoughts in the different markets we address. We'll follow that with a session on cellular. Following that, we'll have computer vision, AI, and automotive.
I should say that we're not going to do a Q&A after each session because it will just take too long. What we've decided to do is have a morning Q&A session after the first two business units present, and then we'll have an afternoon Q&A after the second two business units present. If you can hold your questions until the right time, it'll help us keep on time for the day and get through everything we need to get through. After lunch, we'll do the Connectivity and our Sound Business Units, and then after that, we'll have, as I said, the second Q&A, followed by a presentation about China, and some insights into the market there and where CEVA's playing and what the opportunities are. It's a pretty interesting look at things from our perspective.
Finally, Yanni will wrap up the day talking about the implications of the growth strategy and the different knock-on effect from what we're going to discuss within the business unit presentations today. Then hopefully, if we're all on time, we'll be out of here by 2:30 and hopefully you'll have learned a lot more about CEVA that you didn't know in the past. With that said, I'm just going to give you a quick intro to each of the presenters today so you're a bit more familiar when they come up. Gideon Wertheizer, our CEO, who's been in charge of CEVA since 2005, but as part of DSP Group previously, where CEVA spun out of, he's been there since the '90s, and Gideon and Issachar Ohana, who's our head of sales here, were the two people responsible for developing the DSP, licensing the business within DSP Group.
They've been with the company a very long time. Yaniv Arieli, of course, a long history within DSP Group and joined CEVA in 2005 when Gideon took over as CEO. From the business units, Michael Boukaya is here. He heads up the Wireless Business Unit. Again, longtime CEVA employee, came from the DSP Group days as well, for a long time the chief architect of many of the DSPs that are in production today and generate a lot of the royalties. Aviv Malinovitch, who heads up our Connectivity Business Unit. Aviv joined about 10 years ago now and took over the RivieraWaves business and the Connectivity business in 2014 when we acquired RivieraWaves. Before that, he was the VP of operations for the company. We have Ilan Yona, who's the general manager of the Vision Business Unit.
Ilan joined in 2011, was looking after the imaging and vision algorithms and products until we developed the Vision Business Unit. We saw the opportunity there, and he's been running that since we set that up. We have Moshe Sheier here, who's going to be presenting the Sound strategy for the company. Moshe today is the Vice President of Marketing at the company and also runs Sound. Moshe has a long history with CEVA and DSP Group. He asked me not to say, but he did leave for a while, came back again, so he's on his second round here with the company, but we're very lucky to have him. He's very strategic and a great asset to the company. Finally, the last three presenters we'll have, Emmanuel Gresset. He works within the Wireless Business Unit.
He's going to cover the markets and trends within wireless, covering base stations, handsets, and cellular IoT. We have Jeff VanWassenhove. Jeff looks after automotive marketing for CEVA, and today he'll not just be presenting the automotive strategy, but he'll also do the imaging and vision markets and trends for you. Finally, we have Franz Dugan, who is head of sales and marketing at the connectivity business unit working under Aviv, and Franz will take you through the Bluetooth and Wi-Fi businesses. Franz was one of the founders of RivieraWaves back in 2010, I believe, and we acquired RivieraWaves in 2014 and obviously have built a very nice business as a result of having RivieraWaves for connectivity. With that said, I'll pass it over to Gideon to begin the day, and good luck, and hope you get something good out of today.
Okay. Good morning. Let me start by welcoming you here to visit us. For me, it's a historical moment. I recall when I joined this digital signal processing back in 1990, and I spent the whole first week of carrying desks and setting up computers to the point that we are here in Nasdaq and opening in an opening bell. With that said, let's put aside the nostalgia and let me take you to 2013, where we, CEVA, was one of the rising star in the cellular market. We were a leader in this space, taking over sockets by then and Nokia and others. Even in these good days and hay days, we decided to embark a new strategy, and the strategy was to diversify and to take advantage of our DNA in digital signal processor to a new area.
Today, objective is to give you the insight of all this strategy, how it turned out after six years from now, and to explain all those markets. For this purpose, we brought all the executives and the marketeers that lead this effort. I'll take the next few minutes and give you a more overview of what you're going to hear today in more detailed level. Excuse me. What this strategy that I just mentioned, it's all about? Basically, we have developed a technologies, and CEVA is an IP company that process and connect data for IoT purposes. Underneath this process and connect data, there are underlying technologies that the profession of this one is scarce. Think like modern technologies and wireless local area networks like Wi-Fi and Bluetooth and AI and hardware and software.
These are professions that you don't find that many around the world. It's not commoditized like CPU programming or C programming. With this skill set, we decided to focus on the key markets. The first one is cellular. I mentioned that back in 2013, we were just in the handset, and we continue to be in the handset market today as well because people may not be aware of, but the area of mobile internet, meaning things that we have, LTE smartphone that allow you to connect to the internet, is not that fully deployed. There are 3.3 billion LTE subscribers today, and it will go up to 5.4 in the next few years.
This on top of it, these are on top of this mobile broadband that you're going to see it in India and emerging market, there is the cycle of 5G that is coming by itself. That's the good thing about cellular technology. Every few years, there is a new standard, a new class of services that come out, and it recreate these billions of units again coming. Another area that we decided to embark and to work is to go into the base station market. This turned out to be a very successful strategy because we found an area that our specialization to build processor, and we will elaborate furthermore what this specialization, it's something that you cannot find everywhere, and we become a dominant player in this huge space.
The third area that we are targeting, it's relatively new, I would say about two years, and this is what it calls cellular IoT or other people maybe heard the name Narrowband IoT. This is the area that you're going to connect billions of sensors. It's going to be a camera, it could be your smart meter at home, it could be your health scale monitor, and you're going to connect it to the cellular and take advantage of it. This is by itself, it's about 4.1 billion units. You take the CEVA as a handset company, and you use the same skill, or we use the same skill, and to grow a huge other segment in the market and without the need to do any acquisition or big dramatic moves that put the company in an uncertainty for a while.
Let me move to the other area, and this is a collection of what we call smart sensor. I started by saying that we are basically processing sensors. The terms of sensor shape type completely in the last few years. Camera today is not just for the purpose of taking photos. If you try to understand what is the video, detect face, detect pedestrian when there is autonomous driving, this is a sensor. You need to deploy technologies in the area of computer vision AI. People speak about AI, but the first implementation of AI is camera as a sensor. We decided to address this space early on and become one of the pioneers in this space. Another area, and that's, people knows about Amazon Echo and Alexa and smart speakers. Underneath there, it means that your microphone is not just to speak.
It's about making your microphone a sensor, because if you want to recognize voice, if you want to interpret and learn something that you speak before you go to the cloud, then it's a sensor. We decided also to approach this market to our technology, and we will elaborate how we're going to address the opportunities there. The third area, again, relate to connectivity. We have the cellular. Cellular is a long-range connectivity. What about local area network? What about things that you do at home? What about your smartwatch that is connected to your phone? This is the area of Wi-Fi and Bluetooth. For these purposes, we decided to acquire a company, which turned out to be very successful. We acquired back in 2014, a company named RivieraWaves.
It turned out to be very successful because the market is huge, and you can see those billions of Bluetooth devices, $5.3 billion, and $4 billion about Wi-Fi devices. It turned out to be successful because the team that we acquired is a team that had a reputation technology-wise, and they need us for business. They need us for making a one-stop shop of connectivity technologies and sensing technologies. With this set of technologies that we developed, and it is not long time ago, we just started all those development back in 2013. What we want to do, our customers are the chip companies, the semis, and they are under extreme pressure. Time to market, life cycles are short. There is extreme pressure in operation.
What we want them to do is to provide them all those different blocks, different computation engines that they can use, whether it is voice, whether it is computer vision, whether it is connectivity, whether it is AI engines, that they can mix and match between what we are offering them and what they have in-house. We let them focus on their excellence, and we take the rest of the burden. All those blue blocks that we draw, these are areas that this is how we expand it. This is how we diversify. We created by this bigger market to address and a higher ASP, because when you start providing to customer higher value, you can have a higher ASP. One of the things that we are being asked here, people understand the diversification strategy.
You see it in our numbers in terms of the licensing line. The licensing revenue is going up, Yaniv will elaborate. The amount of license deals that we sign per quarter went from one in five to six deals a quarter to north of 10 deals a quarter. You get this one. Then, let's say, the proof of the pudding is the royalties. Where the royalties will come from and when? I want to explain to you the practices in this business. When we embarked in 2013, this strategy, we had to develop IP. Developing IP that will turn out to be lucrative IP takes time. This could take time between two to three years to develop a processor, a DSP processor for computer vision or DSP processor for base station, could take more. That is CEVA investment.
Here we put our R&D, here we put our talent, here we put our excellence. Then this is the phase that we are today. Once you have this technology, it is qualified and fully tested and something the customer can license, we go into the licensing and the deployment. We license this technology. We get some license revenue out of it, but then our customers start working in building the chips with our technology, developing the software with our technology. This could take one to five years. If you have a Bluetooth wristband, that could take a year. If you have a base station chip, that could take five years, and of intensive work and on our side, the game is not over by licensing. We need to support and increase our resources and work with this customer.
What we have, we have this 40, 50 licensing agreement that we are signing today, these are 40 to 50 designs that are embarked, starting from this phase. We have the phase that we are all waiting for, that's the royalties, because royalties is what make IP company a successful company. From that point on, you go into a continuous rate. We have today 40 royalty payers. Through this strategy that we developed starting from 2013, we are now about 60 companies that will add to this one royalty. From where we are today, we basically more than double the amount of new royalty sources into our revenue stream. Not just that, the market that we're targeting is at 10 billion units today, it is growing.
You see that we are expecting 70% of all those markets that we are addressing this product by 2022. We speak of a total available market of 17 billion units a year. As I said, we have the customer, those 60 new customers that are actively developing these new chips into these markets. Another element that we considered when we built our strategy going forward is the life cycle, the lifetime of product. We wanted to have a mix. We don't want to even could be in a multiple market, but be in a short-tail market. What is short-tail market? Short-tail market is that the life cycle of a product, it's one to two years. The good news is that every one, two years, you have a new cycle, you have a new cycle of licensing.
You build relationships with customers, you keep going forward. On the other hand, the requirement is not just give them the same stuff every two years. You need to go up in the value chain. You need to give a more complete solution in order to shorten their time to market. That's, as you can see in the presentation from all the businesses, you'd see how we went up in the value chain for those short-tail markets. Another component that we wanted to address is the long-tail market. The long-tail market is a market that can be like in the base station. Whenever you win an agreement and whenever you get into production, it's a 10 years business that you get all this. On the other hand, when you develop such a technology, you have to have a foresight.
You have to think what will be still relevant in 10 years, which means higher development, more complicated processors, more understanding of where is 5G and next 5G. We are developing technologies today that takes us beyond the 5G. To summarize the strategy, you're going to get more into this in more details. First, our strategic objective in cellular is to, first of all, go to the handset. Handset is a big market, there are two opportunities there. One is what is called LTE mobile broadband. There are still lot of places in the world, starting from India, going to Africa, going to Latin America, going to Eastern Europe, that they still use either 2G or 3G.
LTE, it's a game changer for them, they will go to LTE, you see the investment in the operator and semis are going into this one. This is going to be a low-cost, high-volume market. This is the essence of CEVA. In the feature phone, if you take the history, feature phone, CEVA has 70% of the market. It's not growing market anymore. It's still sizable, but you do have 70. This give you an example of where the strengths of CEVA and its customer. When there is a high-volume market, every penny counts, our technology makes a lot of sense. On top of it, we are coming to the 5G. Our wireless people will show you what we are doing in 5G in terms of building platforms to our customer. That's one element.
The other element is the base station. Here, we have a lot of room to growth. First of all, in terms of existing customer that will go into production, we have two of them. One is already in production, the other one is about to go to production. Because we are ahead of any available technologies that could be in this space, we can go and expand further our customer base, the benefit for a very lucrative market and a long-tail market. By the way, when you speak about 5G is not just handset. 5G is also automotive. 5G is about robotics in industrial. 5G is about edge cloud. There are so many new opportunities, these new opportunities, you're going to see newcomers. Newcomers are an untapped space for us to come and benefit from the IP model.
The third element in the cellular market is the cellular IoT. A new area, billion of units, our wireless people will show you the strategy that we developed, which is unique in the IP space. Set the stage for many, let's call it non-cellular people that want to go into the space, but they are under a trauma that, this market will be again dominated by MediaTek and Qualcomm, this is not the case in the cellular IoT space. Another objective is in the connectivity. We mentioned things that happen in an enterprise, thing that happen at the local area. In the Bluetooth side, this is a growing space all the time, billions of units. I don't recall any space today that can show you 30% CAGR the next three years. It's a very dynamic area.
There are changes in standards out, we are actively running and offering to our customer this state of the art. Another area in the Bluetooth that it was old, but the AirPod of Apple created this understanding, I would say AirPod at the context of assistant services. These are the headsets like the AirPod. These are not just something that you wear when you want to hear music or something that you wear because you want to speak with somebody. It's something that you wear throughout the day, because this hook you to the cloud. This allows you to ask queries to the cloud, all those services that Siri is providing and things like this. This is by itself, it can be a billion units.
When it comes, by the way, because it's stereo, it's 2 billion units because you have two Bluetooth in each of the ears. The third component is the Wi-Fi. Wi-Fi is a huge market. It used to be only your residential gateway that hook you to the internet at home, but it's all over the place now. Every smart speaker is a Wi-Fi. In smart home, you're going to have tons of Wi-Fi. Every smart TV has Wi-Fi connection. These are new areas that we can come out with our vertically value-added solution, not just give them a piece of it. We are giving for Wi-Fi and Bluetooth a full solution. Last, but not least, it's all related to the sensors in the form of cameras or microphone.
Here we take advantage of early on expertise in AI, that we bundle this expertise with our camera offering and with our voice or microphone offering. With that said, that's in a high level. I let our people go and more deeper into these technologies.
Yeah. Can you hear me? Yeah. Okay. Good morning. I'm Emmanuel, and I'll go on by looking at the cellular market and its opportunities for CEVA. Certainly the biggest one is 5G. Why is that? Because it opens up a whole lot of new wireless services and use cases that could not be addressed by LTE. The most common one is the enhanced mobile broadband or eMBB, which is basically LTE on steroid with higher data rates. There are many other use cases. Let's look at a couple of them. Certainly autonomous driving is a big one. Immersive cloud-based virtual reality, which are using some of the key features of a 5G which are new, which are low latency and reliability. On the third type of use cases, there is massive IoT, which includes industrial robotics, smart cities, smart homes, and wearables.
Such services need a new infrastructure. What that means is that the networks needs to be expanded with what's called a heterogeneous network, meaning a combination of Cloud RAN, macro cell, small cell, and femto cell, whether it's indoor or outdoor. A new type of small cell, we'll look at that in detail, but a new type of small cell would be required for wireless local loop or fixed wireless access, which is called a FWA. We'll look at this, but you need dedicated small cells for this. All of these are very promising opportunities for CEVA. Let's look at handset. Which is the big part of it. You can see here on the left diagram. Well, okay, we reached 1 billion overall subscribers today. By 2022, we'll get to $8.6 billion.
What's interesting is to see the shift from overall voice and data to primarily data. By 2022, this will represent 90% of mobile broadband. That's explained on the diagram on the right, by showing that there will be a net addition of 2.2 billion from feature phone to smartphones. That's because of the replacement and the new subscriptions of smartphones in the emerging market, which will shift from feature phone to a smartphone. Primarily, it's about half and half between LTE Advanced, which offers gigabit throughput, and 5G. Let's now look at the cellular modem and the baseband shipments by technology. We'll reach by 2022, 2.7 billion cellular modems shipped yearly. 30% of these will be 5G, as you can see on the diagram here. It is said that every new cellular technology has been deployed faster than the previous one.
Apparently, that's what we're seeing, that 5G will be no different. It will be even faster than the pickup of LTE. I think what should be noticed here is that the huge shipment all the way to still growing until 2022 of LTE baseband, LTE and LTE Advanced. Being added with 5G starting in 2021 and beyond. 30% in four years is very significant. Okay. Let's look at one of these new services enabled by 5G. This little picture here shows that it's very interesting. There are two sides. Obviously, it's wireless local loop, so you're delivering high data rate to the home wirelessly with a CPE that's connected, that's on the outside of the home. You need special base stations that actually use millimeter wave to serve these homes.
What's interesting to see here is that the growth is very large, exceeding 100 million by 2022. The SAM is even larger. It's up to 1 billion just due to the number of households without fixed broadband. Some replacement of wireline technologies that are better addressed with this. 1 billion SAM by 2022 is an extremely attractive market, and we'll definitely be addressing those, both on the CPE side and on the small cell side. Before we get into the size of the opportunity for infrastructure, let me explain a few things here. They are macrocells being used today by 3G, LTE networks. For 5G and LTE Advanced to be successful, you need a lot more than microcell. You need the small cell, which are composed of picocell and the macrocell.
Microcells are the ones we just described for fixed wireless access, and then femtocell indoor. I think what's interesting to note here is that Verizon, on October 1st of last year, launched its first 5G wireless to the home service with a 300 megabit service for about $40 a month in four major U.S. cities. Let's look now on the size of the opportunity on the infrastructure side. There are two things. If you look at the top diagram here, there is already, this right now, over 13 million base station being installed in operation, and they all are using legacy technology up to LTE. They need to be upgraded to LTE Advanced and 5G. That's one big opportunity. It means hardware and software upgrades.
Again, to serve LTE Advanced and 5G, there will be another 13 million, as many, a doubling of base stations and small cell by 2023. If you look at the bottom diagram here, because of the reasons we mentioned for the heterogeneous network and the fixed wireless access. In essence, for 5G to succeed, we need to deploy small cell. Actually, the growth here is huge of small cell, 36% until 2022. They will actually exceed, by 2023, the number of macro cell base stations. About 2 million a year for each of them. The last cellular market we'll look at is cellular IoT. What is that? How do you connect all these devices to the network?
It's through new types of network, which I call the LPWA. So low power wide area networks, which are designed to connect devices for many, many years on a single AA battery. Very low power, but also very low cost, because the complete bill of material of an endpoint should be less than $5. That's what we're already seeing today. A whole lot of integration into single chips needs to be performed here. There are two standards that are being defined by 3GPP for these. One is Cat-M, which is the higher end of these LPWA networks for connecting images and cameras with some mobility support, and then the very low cost, very low power NB-IoT for 10-year battery life at a very low cost. Let's look at the opportunity of these networks.
The forecast here is of starting at $1 billion today, is that it will reach $2.8 billion by 2022, which represents a 29% growth. The market is very segmented here. There are a number of use cases that we saw a bit at the beginning, but ranging from industrial, smart cities. Of course, metering smart grid management has been the legacy cellular IoT technology, started with 2G. 5G and Cat NB-IoT will enable as well industrial and transport and logistics. That's a combination of asset tracking with GPS and NB-IoT to offer tracking assets, cars, people, and containers, for example. Okay. Let me now hand the presentation to Michael, who will go over the strategy.
Hello? You hear me? Okay. Good morning. My name is Michael Boukaya, and I'm running the wireless business unit for CEVA. I will try in the next couple of minutes to share with you our strategy for wireless. All right. Let's start with our position today. 7 billion handsets powered by CEVA today across multiple cellular technology from 2G to 5G. It's a great milestone for our company, and it's a result of very focused strategy and introduction of key technology by working very closely with our different partners worldwide. In addition, six years ago, we have made a strategic decision. We have identified a new growth opportunity in infrastructure segment, in base station segments. We have made a very important decision to focus on specialized DSPs for this market.
As a result, with very consistent strategy and very innovative technology, we have managed today to equip three of the top five vendors in base station industry, which is also a very, very good milestone for us. Like base station, we also identified last year, let's say two years ago, a new growth opportunity, which is called cellular IoT. As Emmanuel referred previously in his slide, it is one of important mission of 5G. This mission targets to connect billion of devices to the internet. For that, we have identified a case for CEVA to introduce a solution to address the smart market. We introduced last year the first IP for Narrowband IoT, which is the dominant technology for cellular IoT.
If you look to this slide, we can say today that CEVA is the only IP company that is able to connect point to point wireless devices from sensor via mobile handsets up to base station. Let's go over our global strategy, in the next slide, I will go in more details about each one of the segments. Our strategy is based on three pillars. Number one is Mobile broadband. Mobile broadband is a very, very big market. This market is growing. As Emmanuel introduced, we expect 8.6 billion of subscriptions in the next couple of years. This is volume. CEVA is here, CEVA will continue to invest in this segment.
The way that we see this market, in term of strategy, you will see a transformation of our proposal from being a DSP-only provider, transform to be a mobile IP platform provider for 5G, which is much more comprehensive than just DSP only. We move from DSP to mobile platform. I will go in more details in the next couple of slides. This will enable us to increase our value add with higher ASP. Number two is base station. Base station, we lead the market. As I said, our presence is an evidence in the big players in this segment. We intend to expand our leadership for the rest of the customers. Also to expand our leadership to the new segments infrastructure.
We will see that due to the nature of the deployment, we already identify newcomers in this market that will need our technology. This segment is very lucrative in term of license and royalties. Number three is, in the last, let's say, 20 years, we have accumulated a lot of experience in wireless. We also recruited talent teams, we want to use this expertise in order to address new markets. One example is cellular IoT market. We have a know-how today to go to the market, I will show you our strategy. More than that, as we saw in 5G introduces new non-handset use cases, new business opportunities. We definitively target these segments like automotive and industrial. Let's start with Mobile broadband in a bit more detail. In one sentence, we want to streamline the path to 5G.
For us, there are three opportunities. Number 1 is, we power a billion of devices, and we are in very good relationship with our customers. We have the privilege to work with these guys in order to migrate from one generation to the other. We made it from 2G to 3G, we made it from 3G to 4G, and we are now making this transition and migration with our partners from 4G to 5G. The second opportunity, it's a very interesting trend, is OEM internalization. OEMs are looking to internalize their modem technology. There are a couple of reasons for that. One of the reasons is to get better margin. The second reason is to reduce the risk to be dependent on one single source in terms of ecosystem. The third reason can be also to differentiate the solution. These guys make modem architecture.
In modem architecture, you need DSP processors. You need processors. The natural choice for them is to license IP technology versus recreating another CEVA company internally. They rather prefer to focus on other aspects in terms of productization. Therefore, for us, it's a great opportunity. The third opportunity is China. Issachar, in the afternoon, will elaborate on China, will give you very interesting insight about what happened in this region. I can say very quickly that China strongly pushes 5G with a lot of budget, and encourages local players to develop 5G devices. This means that newcomers are entering this market, and when these newcomers are entering this market in China, they need technology. For us, it's a great opportunity because we can bring the right IP portfolio to reduce the entry barrier for modem architecture.
As we say, 5G is not only a handset opportunity. It's also new, lucrative, non-handset devices and use cases. 5G introduces key technology assets that open new markets. For example, 5G introduced what we call URLLC, ultra-low latency communication. It means that you have a link connectivity with ultra-low latency. This characteristic opened the market, for example, for automotive. Autonomous car, connected cars, where you need ultra-low latency between vehicle to vehicle to pedestrian, or vehicle to network. 5G opened this opportunity. Same for industrial critical control with ultra-low latency. That's just two examples. Another nice example is 5G also sustains very high bitrate. We talk about gigabit per second. As Emmanuel introduced in the market presentation, this gives the opportunity to replace cable modem, DSL, with fixed wireless access, with wireless connection. This is possible due to the very high bandwidth and very high throughput.
These are just three examples, and these examples mean that we see newcomers from different industry come and they want to make their own SoC, including cellular technology, rather than taking models from Qualcomm or MediaTek. When these newcomers are entering this market, they lack, in some cases, some aspects for wireless expertise to productize products. For us, it's a great opportunity because licensing IP from CEVA will reduce the time to market and reduce the risks. Let's have a look about the technology and why CEVA is the right player to address 5G challenges. We like software. We are a processor company, so we like software. The question is why 5G and our modems will also like software rather than fully hardwired solutions. I will give you just few arguments without diving into technology. One reason is multi-mode architecture.
Even we see during this year new chipsets that will be announced. You will see that these modems will be not 5G only, but 5G with LTE. 5G with LTE Advanced. Remember that LTE and LTE Advanced are still very dominant in terms of technology, in terms of cellular radio access. Therefore, there is a need in your modem to support 5G, LTE A, LTE Advanced, and let's say also 3G. 3G does not disappear. This means that in one single device you have to support multiple modes. A programmable engine, a programmable solution with software solution makes sense rather than gluing a lot of hardware by supporting hardware for 5G, hardware for LTE, hardware for 3G. It will not be cost effective. The second reason why software is so important is very simple. The evolution of 5G.
We are just at the beginning now, the standard is evaluating. We are today on release 15. Next year we will see release 16, and so on. This means that you need to have a flexible solution, not to close now the solution. A fully hardware approach is not upgradeable. In few words, we see, and I can tell you with confidence by knowing this market very well in terms of modem architecture, all 5G modem will be software-based with hardware also, but software-based, what we call software-defined radio. When it comes to software and processors, CEVA is a natural choice. When we look to the different opportunities, we introduce PentaG. PentaG is a mobile platform. It is not a DSP, it's a platform that contains specialized engine, specialized processor, scalar processor, vector processor, and AI processors.
We are proud to say that CEVA was the first in the industry to introduce AI technology for wireless modem in handset, in IP landscape. We were the first. Just to give you an indication of the innovation we put, investment we put in this field. As I said in my introduction, we are moving from delivering one single DSP. Rather, we see here a mobile platform that includes multiple elements to give the right answer to the most sophisticated use case in 5G NR. As a result, the PentaG reduces enormous complexity in designing 5G NR devices, reduce the entry barriers for the newcomers. Only newcomers. We talk about OEM. OEM likes this approach because the PentaG is like a Lego.
You can select whatever you need whatever you lack in your modem, if it comes to a scalar processor, a vector processor, an AI processor, just connect to the legacy solution of these OEMs. By design, we made sure that each of the IP components is designed a very standard way for a smooth integration into legacy solution. Last point, very important point, is the architecture and the platform is very modular. This enable us to address different use cases, not only mobile broadband use case, which is the most important one, but other use cases, new use cases introduced by 5G. Here we see an example. We see here two derivatives of our PentaG. On the left side, we see a derivative of the platform for mobile broadband handsets. Here you can understand what I said.
You see it's not a 1 DSP. It contains a platform of different components. DSP core, vector processor, scalar processor, the CEVA-BX2 that we recently announced, hardware accelerators, AI processor, et cetera. On the right side, on the left side, sorry, you see here a configuration for cellular V2X. V2X stands for vehicle to vehicle or vehicle to network or vehicle to pedestrian. This is for automotive. You can see that we can reuse our design. We can reuse the IP that we designed for addressing different use cases. Let's move to the next segment. Base station. 5G is not an evolution. 5G is a revolution in terms of technical challenges. If you look to the deployment today, you will see on base station, on the towers, 2 antennas, 4 antennas, 8 antennas for LTE Advanced. In 5G, we talk about up to 256 antennas.
256 antennas, or 108 antennas. We work with customers that target their device now with 128 antennas. It's enormous. This imposes new technology challenges in terms of baseband technology, like massive MIMO, how to treat signals for all these antennas. Much higher bit rate, 10 gigabit per second under ultra-low latency, as we discussed before. The challenge here is not a simple evolution, but a very big revolution. On the other hand, you see that, I can tell you that all base station architecture, all base station architecture are DSP-based. All. Why? Because these products have a very long lifespan. It's not like a handset that you can change every 2 years, every 3 years, but something that stays for a long time.
This means that you need the ability to software upgrade the platform from 1 standard to the other, from 1 release to the other, from 1 feature to add more features, you need a software approach. Let's have a look to what's available. What are the challenges in terms of availability? Legacy ASIC platform that was based on Texas Instruments, Freescale Semiconductor or NXP Semiconductors today, or even FPGA-based implementation that don't meet the cost. In terms of cost, but also on power consumption, and does not meet the performance and requirements imposed by 5G. As a result, OEMs require state-of-the-art DSP technology to address these challenges. Here is what we do, and this is our go-to-market strategy, and here is the path to our leadership in this market.
As I said, from 6 years ago, we have made a strategic decision to design state-of-the-art vector processor architecture, especially for this market. We are today the only one that launched 4 DSP cores for this market, 4 generations, from 3G up to 5G. We have established very tight collaboration with 3 of the top 5 vendors in this industry. This collaboration enables us to really deliver and design and architect very sophisticated DSP architectures. As a result, we present the competition very high entry barrier. As a result, we can say that we are today the undisputed number 1 vendor in this segment. Our technology and our DSP cores and vector processor address all RAN architecture from small cell, micro cell, C-RAN, et cetera. The technology is in mass production. The 2 generations, XC3200 and CEVA-XC4500 are in production today.
Our latest DSP core, the XC12, that we designed for 5G is already licensed by 3 of the top 5 vendors for 5G. You can see here we are proud to power Nokia and ZTE. An interesting point in terms of growth strategy. This market is evolving. It will not be just a game of the big players. The nature of the deployment is now different, is a heterogeneous deployment, meaning new type of cells will be deployed. New small cells, femtocells will be deployed in this market, the reason is to sustain the big data traffic and dense area with high download requirements. This we see today, newcomers, new small cellcomers in the market, they all need DSP technology. In addition, we also observe new initiatives, like Open RAN initiative, Open RAN forum. This forum is driven by operators.
The objective of this operators is to reduce the cost. What they do, instead of having a black box from Huawei or Ericsson, they decided to open the device to different components, each one can come with a different piece of the solution. Companies will be specialized in baseband only, other company with RF IP, other with network processors, with standard interface. This is disruption of the landscape. We already see here new vendors and newcomers that require DSP technology. As a result, we see a change, an evolution with newcomers, not necessarily the big vendors, but need DSP technology. Here is my last segment, cellular IoT. This is a very exciting segment.
As I said, one of the important mission of 5G, to connect billion of devices with characteristics of very few bitrate, like sensors, like geofencing devices, asset trackers, smart city, smart meters, a lot of use cases. Even if it's something new and growing, it's taking off now. We see that we have today 69 commercial launches worldwide that deploy either Narrowband IoT or LTE-M, Cat-M. The interesting thing here that we see that 80% of the deployment is for Narrowband IoT, that's the reason we focused on Narrowband IoT. Why do we have an opportunity here as an IP company? Let's have a look. The opportunity here is across the value chain.
In this value chain you have on the top the mobile network operators, you have the endpoint OEM, the module OEM, like Gemalto, Telit, u-blox, the tier 1 cellular SoC, newcomer cellular SoC, tier 1 MCU SoC companies. What we observe here is a transformation of the value chain when it comes to cellular IoT. We see that to get better margin also the target to have different use cases in this market since the market is very fragmented in terms of use cases, even operators want to make their own SoC. True for module OEMs, true for, of course, the classical cellular SoC, also newcomers like MCU companies to make their own SoC to integrate Narrowband IoT modem, integrate AP sensors, GNSS technology for location.
However, these guys are newcomers for cellular, the way to reduce the risk and to have a good time to market is to license technology. We see that these guys talk to us, it opens for us many new licensee opportunities. Therefore, we identify this transformation in the value chain, we have put a very clear go-to-market strategy to go for cellular IoT, which is a mass market. First, we need to reduce the entry barriers, to reduce the entry barriers in cellular, you need to deliver a complete solution, not a DSP only, but a full solution, including baseband, RF technology, protocol stack, all the interfaces, support for light application for IoT. Here is another example of our transformation in our global strategy to move from DSP vendor to platform vendor.
The second point is, if you want to address different use cases, we need holistic IoT endpoint with GNSS integrated, always-on voice command, and combination of connectivity like BLE that we have in-house. The third point is the solution should be very modular. Why? Well, since we want to accommodate the various type of customers. Some customers would like to get, the big cellular vendors, DSP only. Newcomers, for example, for industrial license, a full solution. The solution should be very modular. This encouraged us to invest a lot of R&D to utilize our expertise in cellular and introduce a full NB-IoT solution. We call it Dragonfly NB2. This is the first worldwide Narrowband IoT IP solution, which is silicon-proven today. It's a fully integrated solution compliant with 3GPP Release 14.
Integrate RF design, which is today available in two different processors in 40 nanometer and 55 nanometer. Very smart power management because these devices should be ultra-low power. The good news, we also introduced a part of the solution, GNSS integrated in the same solution. Here we have a block diagram of the platform. Again, you see here a platform and not just a DSP. Of course, our DSP is in the center. Here our CEVA-X1. We specialize the DSP with dedicated instructions to be power efficient for cellular, but also for GNSS. You can see our RF here, which is a dual-mode RF for cellular and GNSS, and other peripherals like smart power management for ultra-low power. As a result, I think it was unbelievable for me, for a new market, we managed to have big players, that adopted our solution.
Nordic adopted our solution for NB-IoT and Cat -M1. ZTE also adopted CEVA technology for Narrowband IoT. An incredible milestone, we have just last year, 10 new licensing of our full solution for Dragonfly for an emerging market. All the indications are very good. I think we have a very strong strategy here. When you look to the competition and the landscape, I can tell you with high confidence that we are the only IP player that can supply such comprehensive solution for this new mass market. As a result, some key takeaways. During the last 30 years, we have developed a unique expertise in cellular for different fronts, for cellular handsets, for base station, for cellular IoT. We are very well positioned to address the upcoming 5G cycles. We're already making it with handset and base station. We continue to do that.
We are in a very good position with very sophisticated technologies in different segments. Last, we utilize our knowledge and know-how in order to enable newcomers in this industry and to reduce the entry barriers for these guys or to partner with us. Thank you very much.
All right, good morning and welcome. My name is Jeff VanWassenhove, and I'm the director of the automotive segment marketing at CEVA. Today I'm gonna talk about some of the most exciting and most promising technology within electronics today, and that's computer vision, artificial intelligence, and automotive. At CEVA, my role is, I work within the larger automotive ecosystem. I work directly with OEMs and the car makers. I work with tier 1s. I work with silicon providers. What I do is really work with these companies and communicate the value of working with CEVA. More importantly than that, we try to understand what are the requirements of next-generation vehicles and ensure that CEVA's IP is aligned to help them deliver these next-generation features and functions. I'm gonna talk about automotive in the second half of my presentation.
The first half I want to spend time about talking about the broader computer vision and AI market. I'm going to hand it over to Ilan Yona, who's the Vice President of the Vision BU, Ilan is going to talk about CEVA's technology and strategy going forward to be successful in these markets. Last week, I was at the Consumer Electronics Show in Las Vegas, there were many announcements about artificial intelligence and computer vision. The thing I want to talk about is maybe a little bit more subtle. I was in a casino, I look up to the sky, there's cameras everywhere you look. I found out that the Bellagio alone has 2,000 cameras in the Bellagio. It doesn't stop within the casino. I'm walking down the strip.
Every 100 yards or 100 feet, there's a camera shooting both ways. I get up to the intersection. The intersection has anywhere from four to eight cameras monitoring traffic. How do I escape this? I'm going to get into a cab. The cab has three cameras. There's one monitoring traffic, there's a camera that's recording the driver, there's a camera that's recording me. If I learned anything last week in Las Vegas, is what happens in Vegas is probably getting recorded. That is a trend that we see today. Cameras are becoming ubiquitous. They're everywhere. In this room, there is three security cameras in this room. Really everywhere that we go, we're being recorded. This is a trend. They're in the vehicles we drive.
Today, to help us keep us safe, to make sure that we're paying attention, we're doing things like SLAM for 3D viewing and 3D reconstruction. Augmented and virtual reality are opening new markets, first starting in the gaming industry and moving really into everyday lives. We have object tracking with drones. Drones have multiple cameras to do things like collision avoidance, object tracking, scene recognition. Cameras are not only recording what's happening, they're doing real-time analytics to tell us what's happening. Then, of course, facial recognition. These are being used today in our cell phones and security and safety systems. By 2022, we expect to see 44 billion cameras in the market. These cameras aren't just recording what we're doing. They're actually doing real-time analytics and real-time applications, and these require computer vision and artificial intelligence.
Today, much of that artificial intelligence is being done in the cloud. If you want real-time applications, we cannot depend on the cloud to do artificial intelligence. The trend is to move these intelligence and analytics towards edge devices like cars, drones, security cameras, and phones. A perfect example of this is automotive. Automotive, it's a safety-critical device. You have to do the analytics and the AI local to the car. We can't wait for the algorithm to be done in the cloud. It's too late. That could be the difference between life and death in automotive. Some of the benefits is obviously low latency. Just like the automotive use case, we look at power. In our cell phones with face authentication, we need to have local processing there local to the device. Privacy is a big issue.
Having edge processing in these devices helps to alleviate some of these problems. Within the greater computer vision and AI market, CEVA's focusing on some key verticals. These verticals are smartphones, surveillance, consumer, automotive, and industrial. The one thing that all of these have in common is they all require smart, efficient, local processing. That's where CEVA's taking a leadership position, is in these areas. I'm going to talk about a little bit these in more detail. Smartphone. Within the smartphone, the camera's becoming a very important key differentiating feature. I think about today, a big moment for myself, big moment for CEVA, and we all want to have a picture of it. If you look downstairs, everybody had their cameras taking pictures of everything. I just don't want a picture, I want a moment.
That moment could be, I just don't want my face, I want it in portrait mode with a really cool filter. That's what computer vision and AI bring to the cell phone market. If you look at cell phones today, we have a standard of two lenses. That's moving up to three lenses. This is not only in just the high-end phones, but it's also migrating into the mid to low-end phones as well. These things that we can use as computer vision and AI are things like intelligent zoom, improving selfies and low-light performance. Soon, augmented reality. Within augmented reality and virtual reality, today it's used mostly for gaming. In the future, it's going to help us to inform us about the world around us. Surveillance and security. The surveillance market is very large opportunity for CEVA.
If I look at China today, China has the Skynet program, and it's used to monitor the population. Today in China, they have 175 million cameras. That's growing to 300 million cameras by 2020. In fact, Beijing today, there is 100% coverage of the entire city of Beijing. Wherever you go in Beijing, you're being recorded, 100% coverage. These cameras are being used for things like government. A good example is our airports, doing things like facial recognition. Municipal governments, police, and transportation. Able to monitor traffic in real time, change traffic flow, and help congestion. Police is a very good example. Before, if you have a security camera, you need a person to monitor that camera. With 44 billion cameras, you're obviously not going to have 44 billion people or 20 billion people monitoring these cameras.
You can actually use computer vision and AI to do real-time analytics. In police cameras or security cameras, they have things like scene recognition that's happening via AI. If a guy's running down the street, is he running or is somebody chasing him? With artificial intelligence in the camera, these are the types of things that we can do. Consumer. Within consumer, there's a variety of devices. These are some of the most important ones, but what I think is even more interesting is within the consumer space, there are areas that we don't even know about today. If you went to Consumer Electronics Show, a guy had a suitcase and he put a camera and a sensor in it, and now the suitcase follows the guy. Within the consumer area, there's a lot of potential for applications that we're not even aware of.
Drones today are doing things like image stabilization, object tracking, the follow me function, if I'm skiing down a mountain, that drone's able to follow me. Smart home. Within smart home, we have smart home devices today, we're using voice. These are now being enhanced with computer vision to do things such as personalization, safety and security. Augmented reality and virtual reality. Today, it's used mostly for gaming, that's the reason, because we have these really large headsets, no one's going to walk down the street with a very large headset. There are a lot of companies that are really focusing on making smaller, stylish AR/VR wearables. Then, of course, the old standby, photography. Obviously, we're utilizing computer vision and artificial intelligence in photography and also in action cameras, this is also expanding.
Things like 360-degree cameras and also cameras that are able to shoot video enhanced for augmented and virtual reality. If you go on a vacation, with a 360-degree camera, you can basically relive that moment in real-time with the aid of AR and VR. Then, of course, industrial. We're on the brink of the next industrial revolution, it's called Industry 4.0. These factories will be smart and connected. They will be powered by AI-based processes. They'll have a lot of computer vision, anything from AI-powered robots to inspection and computer vision systems. What's interesting about Industry 4.0 is that we are doing basically real-time analytics and able to change things on the fly.
If we have AI built into robots, we get input from a machine that has computer vision, it knows how to make the changes to do things like streamline efficiencies and making things more efficient in Industry 4.0. That's largely based on AI and computer vision. Finally, I'm home, automotive. I guess I want to tell a little story about automotive. I was at the Consumer Electronics Show, I said that earlier, but I was at the Consumer Electronics Show this year. The first time I went was 10 years ago. I was working for NXP, we had a booth. We were allowed one table in the corner, and we had one demo. Everybody wanted to know why the automotive guys were even at the Consumer Electronics Show. You fast-forward today, I would say that it's almost an automotive show.
The north hall is completely filled. You have the parking lot that's filled. It's spilling into the west gate. The reason why is because automotive technology is so impactful and there's such a tremendous amount of opportunity there. The semiconductor content market for automotive is to be a $67 billion market by 2023. This is being driven predominantly by a number of growth factors, is one, of course, everybody's talking about autonomous, and I'll talk about that in a little more detail. Before that, there's things like NCAP. NCAP is called the New Car Assessment Program. What NCAP is before you used to get a safety crash rating, so five stars, everybody wants a five-star rating. Before it was how well a car would survive a crash, and that's changed because of active safety.
Now it's how well does your car avoid the crash altogether. Because they're increasing what it takes to get that five star, there's a tremendous amount of growth into active safety. Of course, autonomous driving as well. We're all aware of the major impacts that autonomous driving will have. All of these require artificial intelligence on the edge and artificial intelligence within the vehicle. In order to achieve these things, we need more sensors, more processing, and more intelligence. What has CEVA's strategy been in automotive? Of course, there's the prize out there, which is fully autonomous vehicles, but really there's a lot of opportunity before you even get to full autonomous vehicles. If you look at this chart here, you really only see introduction of level 4 and level 5 driving, so that's basically fully autonomous or highly automated driving.
You only start to see that in 2024 and 2025. Prior to that, there's a tremendous amount of opportunity with a number of applications such as rear backup camera, surround view, driver monitoring, and all of these require AI at the edge, so what CEVA is specializing in. The reason why we took this approach is obviously we can have a quicker time to revenue. The market is much more stable for the lower levels of driving and also for active safety. We're waiting for 2022, 2024 or 2025, so what we're doing is we're taking our relationships and the business that we've won in some of these lower levels of driving and positioning ourself to develop the IP for next generation.
We have time. I talked about NCAP and I talked about active safety, but there's another thing that's happening is a lot of growth is being caused by government regulation. In the U.S., last year in 2018, they required every vehicle to have a rear backup camera. It's an ultra-competitive landscape, so everybody wants to add differentiation. How they're doing that is by adding additional functionality, additional functionality that can be created through the use of computer vision and artificial intelligence. Forward monitoring, so forward camera, that's basically what got everything started in ADAS. This is being required in 2021 in Europe. Europe actually just announced 11 specific safety devices that they are requiring in 2021 to have a level 5 crash rating. Those are things like in-cabin monitoring.
In level 3 driving, the car can drive itself, or it has to hand control back to the driver. They're using cameras in the vehicle to monitor the driver to be able to do that. In addition, the number one cause of accidents is distracted driving, these cameras are also being mandatory in Europe in 2021 as well. This is another growth market. Japan just gave the green light for mirror replacements. You don't even need a mirror in your car, you can use a camera, strictly a camera. These are big opportunities for CEVA, and these are the types of devices that we see ourselves in this first phase of our strategy.
I've spoken a bit about automotive artificial intelligence, and obviously we're focusing on the vision side of it for object detection, free space detection, but there's also uses elsewhere in the car, so virtual assistants, gesture recognition, speech recognition. The reason why artificial intelligence is so important is, number one, if you look at levels of high autonomy, they require highly accurate systems. You're looking at very challenging environments in automotive like weather, occlusion, I'll talk about occlusion in a second, going into bright lights, dark lights. These are very challenging things for automotive. Artificial intelligence can handle these robust use cases. For a fully autonomous vehicle, we need to have 99% accuracy. Okay? In order to get 99% accuracy, you have to have very reliable and high-performing systems. Normally with traditional computer vision, we can get to 80%-90%.
Really artificial intelligence is going to get us to the next level to be able to do the rest of the 10%. This is an example that I wanted to show. This is actually one of CEVA's customers, Nextchip. They're a semiconductor company based out of Korea, and they are utilizing CEVA's XM4 to run small, efficient networks. The number one growth opportunity, in my opinion, in automotive is, of course, we talk about fully autonomous vehicles, but it's really the particular use cases that will need artificial intelligence to get to the next level. An example is traditional computer vision has a really difficult time knowing that that is a car. We're using AI to handle this one particular use case. In addition to things like this, we also can do free space detection.
free space detection is very, very important for autonomous driving, these are all being enhanced by CEVA technology. The strategy for CEVA for success is we have products that have a combination of computer vision and AI. Now AI gets us to the next 10%, but still computer vision is very stable and predictable, it's not going away. AI complements that of the performance of CV. CEVA products are well-positioned because we have tightly coupled products for both computer vision and AI. We've had good success in the smart sensor area, guys like ON Semiconductor and Nextchip are using CEVA products. Finally, I want to talk about the autonomous processing landscape. If you look at really the amount of opportunity for CEVA, a full autonomous driving has, one car has $1,200 of semiconductor content.
This is a combination of centralized processors and edge processors. We're looking at vehicles that have anywhere from six to 12 cameras, very large processors. Today, a lot of people are using these very large and inefficient compute platforms, there's a need to find replacements for these for mass market. Within CEVA, there's very low volumes for some of the first vehicles. When those volumes start to become higher, they're going to be looking for solutions that are very low power and efficient. A majority of the volumes within automotive autonomy of, everybody talks about the full autonomous driving, but L2 and L3, which means there's some autonomy, there's a tremendous amount of opportunity for that. That's a good combination of smart sensors with smart, intelligent processing, this is really the area that CEVA's in.
In regards to CEVA's opportunity in autonomous driving is we have broad engagement with OEMs, Tier 1s, some of the disruptors. If you look at, this is a market that's really changing quickly. Even from an algorithm standpoint, so in neural networks, there's new algorithms that are coming out every day. The processing requirements are increasing. We need more sensors. How we can position ourselves for this market is working directly with these companies. We have very good relationships that we can help develop the next generation systems. That's what I have to add. I'm going to hand it over to Ilan Yona. He's the vice president of the vision business unit. Thank you.
Okay. Hello, everyone. You've heard from Gideon and from Jeff that cameras today are everywhere. Actually, you cannot hang around without seeing a camera. It used to be in the past that we were using cameras in order to do digital photography. These days, we are talking about much more than digital photography. Actually, we are talking about four underlying technologies that drive the advanced cameras. I believe, again, you heard about it, we are talking about digital imaging, computer vision, of course, 3D scanning, that I will elaborate later on, of course, the AI. By the way, when we are saying AI, you can hear that as machine learning, neural networks, CNN, deep learning. It all covers the same area in some kind of computer vision.
To let you understand when we are saying today digital photography or digital imaging, what we are talking about. In this image, you can see that there is a nice picture. You can see the road, you can see the car, there's the building, the tree. You can say that it's not a blurred image. It's a nice image. As you can see, above the building, there is the sky. You can see the sky, but we shouldn't lie because we don't see the sky. What we are doing today with the HDR, this is a technology that by opening the sensor to several exposures, and we have several images, and we have complicated and complex algorithm to mix all these images, we can actually bring the sky back to the image, how the eye see it.
If you can see deeply, you can see the tree is much more vivid, and you can see the resolution inside the tree. This technology of HDR, keep in mind that if I'm talking about, let's say, five years ago, if you were using HDR in your smartphone, no one used this, because when you push HDR, it took between three to five seconds to the smartphone to bring the HDR image. Now, in the technology nowadays, you can do it in less than a second. HDR is a common use in digital imaging. This is another example. Here I'm talking about computer vision, not on the AI yet, but this is traditional computer vision. As you can see, there are a lot of algorithms for computer vision. When we say computer vision, we mean not just to get the picture to understand.
This is what computer vision is about, to understand what's in the image, not just to take the picture. In this simple example, you can see pedestrian detection. By the way, we are talking about an algorithm called Histogram of Oriented Gradients to detect pedestrians. In general, when we are talking about computer vision, we are talking about detecting objects, all kinds of objects, and so on. The 3D scanning. These days, again, we are talking about multiple sensors within the camera, and you need 3D abilities. Meaning in this particular example, you can see that we are having 3D model of an image. When we are talking about 3D, it's not just modeling the object. Modeling the object is very good and is a must for AR, for VR technologies.
We will see later on that we are talking about simultaneous localization and mapping, which is very important area these days for the 3D in order for the robots or for the autonomous car to know where they are relatively to the environment. This is the 3D scanning. Of course, we have the crazy area of AI. Everybody today is talking about AI. The main difference in this, as you can see in this image, we still have detecting objects. As you can see in the AI, in one trained network, you can detect much more than one category of object. In this example, you can see that we can detect a car, bicycle, and pedestrian. You can detect hundreds of different objects within one image when you have the right trained network.
AI is growing, and you heard it from Jeff earlier, and we will see what CEVA can bring to this AI market. Keep in mind that, again, the AI is very important. Everyone is shifting to AI, but we know from our customers that AI is not enough. When talking about AI, they would like to have traditional computer vision aside in order to back up the AI in order for them to work together. Currently, when we are providing our solutions to the market, we insist to have AI and computer vision sitting side by side. The problem is that when we are talking about computer vision and AI, we are talking about the competencies are scarce in this area. Mainly when we are talking about AI. AI is different expertise from traditional computer vision.
These days, when we are talking about Google, Nvidia, Amazon, they know what to do in AI, but they are doing it mainly on the cloud. On the edge devices, this is another story, meaning you have to take care of resources, restricted memory, power efficiency, and more. What we are doing in the cloud is not similar to what you need to do on the edge devices. More than that, we've been in this area of AI for something like four or five years already. We see that the complexity and the required performance is growing by 2x-4x by every year. It's crazy. Four years ago, we had a solution with 64 processing unit per cycle. These days, we are providing to the market solutions with 4,000 MACs per cycle.
Four years from 64 processing unit to 4,000 processing unit, it's still growing. We know that next year and coming years, we will need much more processing units in order to meet the market requirements. With this regard, classical CPUs and GPUs are not relevant. They cannot meet the requirement of low power consumption together with high performance. We need specialized processor in order to do this work, in order to deliver what the market really needs. We in CEVA, we build a holistic solution combining software and hardware together. We believe, and I will talk about it, that hardware only, hardware solution only for AI, is not enough. You cannot actually bring all the solution only with hardware.
Combining the software and the hardware that we are bringing to the industry, we can get the good trade-off between power efficiency and high performance, and most important, short time to market to our customers. When we are talking about the holistic philosophy or holistic strategy of CEVA, we are talking mainly about this diagram, meaning the four pillars that we have in our solution. As you can see here. On the upper side, you can see our hardware platform that we are bringing. On the lower side is the software. On the left side, you can see the AI, and on the right side, you can see the computer vision. Again, it's all connected together. We have the NeuPro, which is the engine and which is our solution for the AI. We have the CEVA-XM.
I'm going to talk about the CEVA-XM, which is, again, it's a family of cores. The CEVA-XM is a family of DSP cores for computer vision. We have this family of cores, and we keep evolving it for four or five years now. We have the well-known CDNN, which is already a benchmark brand in the market of the software that we have for AI. In terms of the computer vision, we equip the XM, the DSP processor, with a lot of software. We believe that without these SDKs, without SDK for SLAM, without SDK for 3D, without optimized library for imaging and computer vision, you cannot get the solution that the OEM, that our customers will really need.
The way that we see it, you have here in the upper level, the OEM application level, which has many applications, many needs that you would like to address to the market. We could say to the OEM or to our customer, "Okay, go with the CEVA-XM, go with the NeuPro. You have the hardware platform." We know that once he has the application level, bringing the hardware platform is not enough. We believe in tightly coupled between the hardware and the software, and we believe that we in CEVA know exactly what this NeuPro, what this XM, and we can bring the fully optimized SDK for him to use in order to bring his applications. There are two levels here, the level of the hardware and the level above the hardware, which is CEVA optimized software.
Again, every customer of us, beside the hardware, beside the platform, beside the IP licensing, is getting from us also the software that related to these products. What's the NeuPro? You heard already that when we are talking about NeuPro, we are talking about that everybody wants AI. Everybody would like to have its neural network. Every market would like to have its neural network. By the way, each market has its own constraints, its own needs. What we have in the NeuPro is portfolio. We are not bringing one solution. We are bringing portfolio of solutions according to the market and the market needs. Here you can see that in this portfolio of processors, we are growing from 500 processing units to 4,000 processing units. Again, in the IoT market, which is the low end, you can use 500 processing units.
On the high end, meaning automotive surveillance, they need really high-end processors. We can deliver to them NeuPro solution with 4,000 processing units. CEVA-XM. CEVA-XM, from our point of view, is the middle of everything regarding computer vision and AI in the vision business unit. CEVA-XM, as I told you, we started with the development of this core. We have several cores from then, five years ago. We are talking about very powerful multipurpose DSP. We know that we have to always keep developing this family of cores because the requirement from the CV in terms of resolution, in terms of complex of algorithm, always keep growing and we have to be updated with our XM family. Nowadays when we are talking about the CEVA-XM, it displaces GPU or proprietary with open platform that we are bringing to the industry.
What we can tell is that it's a big success. Currently, we have dozens of design wins, and OEM are very happy with our solution for computer vision. You can find the CEVA-XM in many markets like the ADAS, advanced camera, drones, surveillance, action cameras, and more. The software. The question is why do we need this piece of software? When we are talking about AI, currently we are talking about neural network. Each OEM, each developer has its own network. He designed the network. He has the network. It's what we call here the trained neural network. It's already there. On the other side, we have the optimized engine, inference engine, something like the NeuPro. The problem is that when the developer design its product, he doesn't think about embedded solution. He doesn't think about the fact that there are no floating point.
He doesn't care about memory restrictions. There are so many things to do with the trained network that just taking the network and put it on the embedded device will not help. It won't work. In many cases, it can say that you need months of optimization for making it to work on your embedded device. What we have in CEVA, the CDNN compiler is taking the network and start working on it offline. We are optimizing the network. We are taking care of the bandwidth. Bandwidth is a real issue in embedded devices. In many cases, it's the bottleneck of the process. We are taking care of bandwidth optimization. We are taking care of quantization of the weights of the network.
We are putting the network in another way, optimizing it, when it's done. By the way, this offline process can take one minute up to 15 minutes. When it's done, we have a network that is ready to run. In this case, we can save months of optimization for the customer. He can do it in 15 minutes using our tool, and then immediately can send the optimized network to our NeuPro engine and run it immediately. By this approach, we can take care for the accuracy, for the performance, and save time to our customers. SLAM. As I said, SLAM stands for simultaneous localization and mapping. Meaning when you have robots, and as I said, autonomous car is also some kind of robot.
When we have this robot and he would like to know to map the environment and to know exactly where he is related to the environment, this is what SLAM is about. In order to do it, of course, you have to do a lot of 3D processing. 3D, 3D. All the SLAM SDK that we are developing to our customer, it goes around the 3D. And all the AR applications, VR applications, autonomous cars, they need SLAM. SLAM is heavy process. You can run the SLAM on the CPU, but you will not get the frame rate. What we are doing in CEVA, again, we are developing the SLAM SDK, and we can tell our customer all the heavy processing of your application, of the SLAM, you can run it on CEVA-XM. By that, again, get much higher performance from your application.
In order to make it very easy to the customer, we are talking about plug-and-play SDK. Very easy integration for them to use. Just to show you what SLAM is, you can see here. There is the environment, there is this room, there is the camera is moving, and while the camera is moving, on the left side, you can see that the algorithm actually map all the points in the room, and know exactly where the robots or the camera is inside the room. To summarize, as we said, the requirement from imaging, computer vision, AI, is keep growing. It never stops. You cannot blink. We are very happy with our solutions. Currently, the vision solutions containing, of course, the CEVA-XM and the AI solutions, we have more than 50 design wins over the years.
The success comes from our holistic philosophy, meaning we are talking about incorporating DSP, AI processor, and broad range of software technologies together. That's what our customers love about our solutions. As you heard from Gideon presentation at the beginning, when talking about royalties, it takes time. Now we expect in coming two years to get a lot of royalties from the vision business unit. We are talking about this coming year and the year after, we expect a lot of royalties coming from the surveillance, ADAS, and drones customers. Thank you.
Thank you, Ilan. We're going to have a Q&A session now with the two business unit heads from this morning, and Gideon as well. If you have any questions, I have a microphone here to ask, and the guys will be happy to help you with whatever's needed.
First question comes from Mr. Gary Mobley.
Thanks, Richard. I wanted to ask about NeuPro. It's been about a year since you launched it, and I think it's, correct me if I'm wrong, but it's generally available for licensing now. Has it lived up to your expectations?
Speak louder.
No, he's speaking loud, it's coming this way. He's asking about NeuPro.
I'm asking about NeuPro. It's been a year since it's been licensed. It's generally available today. Has it lived up to your licensing expectations? If not, what has been the impediment to driving more licensing activity?
Let me, Gary, let me start. Ilan can jump in. AI, the usage of AI in the Edge is relatively new area. When we started that, we started on a DSP first before we announced the NeuPro. The first wave of customers are those guys that I would call them fully integrated, meaning they know to develop the AI application in addition to this one. They are camera guys, drones guys, that they know what the application they need to do, and they know to do what is called training the network, and then they could took advantage of our CDNN technology and get it running on their drone or camera or whatever. What we see today is the second wave of AI users. These are less knowledgeable persons.
Their mindset is, let's put an AI engine in our chip and let our customer, our OEMs to do it. We are now dealing with the second wave. These are customer less knowledgeable. They need our education. They need our support, and they also need our guidance about what's the next things in AI. That's what we do. This process, it's slower than the first wave. We still have first wave of customers that licensing. As Ilan said, this area is still in early days. There are not that many people, lot of people speaking about it. There are not that many that knows how to build the Edge
side of it. The Edge is challenging, has been challenging by itself.
Hi, guys. Suji Desilva, ROTH Capital. One of the end markets I saw that I hadn't seen before was the home router market, this kind of extreme mobile broadband, as you called it. You guys are very strong in the battery-powered, and then you have now the base station, which is high performance. What's the advantage there, and what are the customers demanding? Who you compete with, how do you compete there? What's the opportunity for you guys?
So-
Fixed wireless.
Fixed wireless.
You want to answer about what the, our advantage is in the fixed wireless? The question was, what are our advantage in fixed wireless, which is not a battery operator.
Yes.
Right? You want to answer? Fixed wireless is a 5G use case, and the challenge is building a modem for 5G, and high-speed modem. The fact that it's not mobile still has its own set of challenges in how to deal with sensitivities and beamforming and antennas and stuff like this that are challenges in places like home and/or enterprise. The challenges in modem is not just in the battery side of it or the low power, it's also in terms of getting performance. The performance that you're going, or the bandwidth or the bitrate that you expect in 5G is even faster than in mobile.
Just want to elaborate a bit.
I think-
Yeah. Go on.
Thank you. As you saw, we are able today with our portfolio to really help our customers to reduce their entry barriers. Our biggest advantage is the holistic delivery, what we give today. It's just not the DSP, but as you saw in my presentation, with the PentaG presentation, we are giving much more than DSP. We are giving different kind of DSPs, but also software on top on that, and this can expedite the development of such a complex modem.
Yeah. Question on the PentaG you spoke about. You talked about that opening up new opportunities in, and mentioned like V2X, and new types of semi customers coming, not the classic cellular baseband guys. How are those discussions going today, and how long before we start seeing these new types of non-baseband customers or the non-cellular guys coming for things like V2X?
I believe you mean in the context of 5G. I would say the following, there are a lot of talks, but it's very early days. If you take CES, the operators literally said, "Come back next year to discuss about this additional 5G use cases." No doubt, this will be a cellular V2X. V2X in general, it's additional sensor that people will need to do in order to get to safety, because you cannot rely on one sensor for that purpose. Fixed wireless is another area that will become a key point, because it's much easier to install a high bandwidth with wireless than to install a fiber, for example. These are the things that will happen, but for now, the first use case for 5G would be mobile broadband. The first use case for 5G would be handsets.
Okay, maybe just one follow on as well on the PentaG moving to more of a platform solution. How does that change the licensing or the royalty or the potential value that you can extract?
Yeah. That's a good question about the PentaG, because as Michael explained, the PentaG is not a DSP, it's a modem platform. The thinking was that into the 5G, you'll have several category of customers. The first category of customer are newcomers. People that are coming into the space could be from cellular, V2X, fixed wireless. These are not the incumbents there. We probably heard about OEM internalization. There are OEMs that wants to control the supply chain and doing modem development by themselves. The third element in the PentaG platform, PentaG, it's a modular platform. If you are incumbent, there are incumbents that are not using DSP. When you go to 5G, you have to go to a much broader multi-core or many core architecture.
We go to those incumbents that don't use us and tell them, "You don't have to adopt the whole platform. If you need an IP, if you need this specific DSP, or you need this AI engine that you provide, please license just this one. In this approach, we want to expand not just the new use cases, not just to the new class or new entrants into the market like OEMs, we also want to go to the incumbents and convince them to license even part of the technology. As long as there is a royalty arrangement in this respect, we'll be appreciative to do business with them.
Pay balance. If the percentage of the chip and its spend, for sure, will not change much.
Hi, Joshua Buchalter from Cowen. Thank you for the informative presentations. I was hoping you could talk about some of the differences in the economics between the small cell and macro cell base stations. How different is that customer base, and also how different are the DSP cores that you're licensing to those customers versus the larger ZTEs and Nokias? Thank you.
Mike, feel free to jump in. The macro cell, the landscape there is those big names that we are all familiar, Nokia, ZTE, Huawei, Ericsson. You need to have a scale to be a player in this one. Those guys also going, that's where we are stepping in, that's an example of internalization. They want to control, they need to control their chip inside the base station because that's their secret sauce. They find a good match, a good strategic match with us because we provide them the technology and can tailor-made it to their specific needs. When you expand further into small cells that could work even in a millimeter wave, here, the chips, it's still a complicated stuff, but there's going to be a lot of small cells. It opens up opportunity for standard product, SSP product.
There will be semis coming from network, semis coming from the handsets, that will go into this space, but they cannot redeploy or reuse. If you're a handset maker or baseband in handset, you cannot redeploy it in small cells. It's not as complicated as, or performing demanding as the macro, but it requires guys like us to provide the enabling technology and the experience to go into the small cell.
Yes. Thanks for taking the question. On the PentaG platform, are you internally developing the RF section or are you partnering for that? If you're partnering, who are you working with?
RF PentaG.
We develop.
In that case, we are not entering this market, meaning that our customers are going directly to RF IP vendors. We don't play in RF in that case. We just focus on baseband. Yeah. It's a good question because it's different from cellular IoT. In cellular IoT, we made a strategic decision to deliver RF technology. For PentaG, our customers either have our own solution for RF or working with other partners.
Anybody else? Suji?
Just a quick one as well. On automotive, I think to date, Mobileye's done well. It's a closed system. Can you talk about the opportunity for guys like you, and what the customers are asking from you that gives you an opportunity into the market as the customers evolve the development of neural nets, and how they want to deploy them into ADAS onto automotive? We know what's happened to date, but curious what the customer is thinking about what they want going forward and how you participate there.
Suji-
Automotive. Sorry.
What was the question? I could hardly.
Happy to.
Oh, sure. Automotive, Gideon. What are the customers saying about what they want in terms of the AI models developing? I think today you've had closed systems like Mobileye doing well, but what is the opportunity for folks like CEVA going forward?
Yeah, that's a good question, Suji. What customer are looking is, how do I make it smart? They know how to take the camera. By the way, we have customers that ask us the same question about radars, radar company. They have this radar or camera that they know how to deal with the signals coming there, clean it, understanding, transferring it to digital. It comes to processing. What is there? Do I see a person? Do I see a traffic light? Do I see this? Here you get to the area that we are specialized. Could be-To AI engine because AI networks or deep neural network can detect this one, and when the DSP processing can help in cleaning up. Keep in mind that today in radar, there is what is called vision radar. A radar can see today, not just measure distance.
We are coming to those people and basically enable them to make their sensor smarter or smart at all. I mean, it's not smart. It's going to be. We see this is a very collaborative relationship because for them, it's the first step into this area.
Anybody else before we break for lunch? No? Okay, thanks, guys. There's lunch outside. We'll reconvene at 12:30 back here for the next set of presentations. Careful now. Just hide this. Yep. If you want to do the highlight, you can do it like that. It's not useful. Yeah. Thank you. Yeah. It's always hard to start after lunch. Useful.
You worked with this girl? Is that who you were? Or were you?
Okay. Welcome back, everybody. Next up is our connectivity business unit, starting with Franz Dugan. Then Aviv Malinovitch will cover the strategy for the connectivity group. I'll take a seat.
Thank you, Richard. Can you hear me? Yeah. Okay. Hope you had a great lunch. I know it's not the best slot now. We really may fall asleep. We are going to present you what we do in our Connectivity Business Unit. We'll talk about the market overview, and then the strategy for our product line. Okay, when we talk about connectivity, actually, in our view, we focus only on Bluetooth and Wi-Fi. Why? Because actually these are two huge markets. Bluetooth and Wi-Fi are by far leading the connectivity wireless communication market. You can see the numbers. By 2022, there will be 5 billion devices shipping with Bluetooth and 3.7 with Wi-Fi. It's a huge market. Which mean that for us, for CEVA, it's a lot of opportunities for royalty revenue.
Bluetooth and Wi-Fi are everywhere, for sure in smartphone and tablets, these markets are huge but pretty stable, not growing anymore. Fortunately, they are also integrated in many other type of devices like smart home, wearable, wireless speakers, et cetera. Many different kind of markets. These markets are actually very fragmented. There are a lot of opportunities for new company to design new chips with their secret sauce or differentiated solution. This is, for us, for CEVA, again, a lot of opportunities for business and for our license fees, actually. I'm going now to discuss a bit further about each of the technology, Bluetooth and Wi-Fi. Bluetooth first. Bluetooth is a quite mature technology, more than 20 years old now, but still growing a lot, surprisingly so.
Not for sure on the smartphone and the tablet markets, which are huge but not growing anymore. There are many other market segments, like IoT, consumer electronic, et cetera, which are showing some good CAGR. Like IoT is 35%, which is good. Actually, this is where most of our customers are playing. They are in those markets. That's why we see a lot of growth in the next few years. In term of shipments, if you look at the details, you may not be familiar with the Bluetooth technologies, but there are actually several flavors. There is the old Bluetooth Classic, which is there since the creation of the Bluetooth standard.
Then, have been introduced in 2010, the Bluetooth Low Energy or BLE, which is extremely low power, and which has been created to actually be integrated into extremely low-power devices like activity trackers, other sensors, beacons, et cetera. Then there is the dual mode. Dual mode, Bluetooth dual mode is actually a combination of Bluetooth Low Energy and Bluetooth Classic. What we see actually is that the Bluetooth market is growing mainly thanks to the dynamic of the BLE market. BLE market is really driving the Bluetooth market. That's exactly what you see here. We've actually 24% CAGR for the next five years. I'm going to talk a bit further about Bluetooth Low Energy. Actually, Bluetooth Low Energy is the standard for data and information use cases. We can split all these markets into 3 different categories. First one is location services.
Quite decent market, more than 260 million unit expected this year. This include point of interest, indoor navigation, asset tracking, et cetera. Actually, maybe not many people knows, but Bluetooth and particularly Bluetooth Low Energy, is actually used in many kind of application on this market for indoor navigation, asset tracking. The standard has evolved a lot and still evolving actually. Very soon, a new version of the Bluetooth standard will be ratified and will be announced. With new features, actually, to optimize even further this kind of application. It's actually a great alternative to GPS and other wireless technologies which are used for navigation. The second category is data transfer. That's the biggest one, which is for sport and fitness, health and wellness, PC peripherals or mouse, keyboard, et cetera. Most of our customers actually see big shipments in these markets.
The last one is about device networks, which include control, monitoring, and automation systems. That's also a very interesting category. Actually, we see growth, particularly since the Bluetooth Mesh has been introduced, or lately, Mesh has been introduced, which was a key feature which was missing in the Bluetooth standard. This is a very key contributor to the growth of this market. Now let's talk about Bluetooth dual mode. Bluetooth dual mode is really the technology for audio. When you need audio, you need Bluetooth dual mode. For Bluetooth audio, there are three main markets. First one is a Bluetooth headset and earbuds. They now use Bluetooth dual mode. The good thing is that, we see growth for this market, steady growth.
The good thing, as Gideon, our CEO, said this morning, is that when moving from headset to earbuds, you basically move from one chip solution in the headset to two chips. Which is twice the royalty potential for us. Which is very good on top of the fact that the overall volume are growing. What we see also longer term is that, this solution will move to BLE once actually the audio over BLE will be ratified. This is something which is under cooking and will be ratified anytime soon. Which is good for us. We have this technology so we can address this market. Second market is smart speaker. Smart speaker is actually growing three times faster than the whole smartphone segment.
We see here also a lot of opportunities, not only for Bluetooth, but also for Wi-Fi and also audio, because this device includes all three technologies. Which is very good for CEVA. The last one is about automotive. Bluetooth is really a default features now in automotive. Let's talk about Wi-Fi. Wi-Fi is also a pretty mature and old technology, almost 20 years, but still growing and showing a good CAGR. Again, the smartphone and tablets market are big but pretty stable. While some of the market segment are booming. You can see, for instance, IoT, which show 35% CAGR for the next five years. Which is pretty good. If you look at the shipments per Wi-Fi generation, Wi-Fi flavors, you may or may not be familiar with all this terminology, 802.11n, AC, AX, which is the next generation.
Actually, the Wi-Fi Alliance made a good job recently to simplify the naming of those generation. N will be called Wi-Fi 4, AC will be Wi-Fi 5, the next generation is Wi-Fi 6. One thing to mention here is that Wi-Fi, 802.11ax, will actually take off very quickly. This is a major contributor to the growth of the Wi-Fi market for the next five years and showing a CAGR, which is huge, 258% for the next five years. That's what I'm going to elaborate now. Wi-Fi 6 will be ratified this year actually, in August. This is really a new technology which will be deployed extremely fast. Actually, we already start to see some products on the market which are compliant with this 802.11ax. Like for instance, you see a picture of the latest ASUS access point.
At CEVA, we already see a lot of traction actually. We already signed some customers for AX. What is interesting is that people are moving to AX for IoT applications, not only for smartphone and tablets or access points, et cetera. That's really a game changer. AX actually is known for high throughput, high efficiency. High efficiency, we mean high spectrum efficiency, meaning that on a given access point, you can connect tens, hundreds, or even thousands of device together to the access point. That's the main benefit of 11ax. Not only that, there are some great features to improve significantly the power consumption, that's why we see also a tremendous demand for 11ax in IoT application, in wearable as well.
Actually, 11ax will not only replace 11ac, we see a strong demand for 11ax to replace 11n, which was known to be a low cost, low power flavor of Wi-Fi. I'm leaving to Aviv for the strategy of our products.
My name is Aviv Malinovitch, I'm the GM of Connectivity Business Unit. In the next 15 minutes, I will walk you through our strategy in our unit. First, I would like to start with our current status. Today, we are by far the market leader for Bluetooth and Wi-Fi IP. I would like to share with you some data related to that. So far, more than 1.5 billion devices have been shipped with our technology inside. This is our contribution to the 10 billion milestone, which was announced earlier today. The compound annual growth rate of our shipment was during the last four years, more than 50%. We have more than 150 customers, every year we sign more than 25 deals.
Whenever the Wi-Fi Alliance or the Bluetooth SIG come or make an announcement of a new standard, we are the first to introduce an IP which support the new standard. For that, I would like to give you two examples. The first example is related to the Bluetooth. Tomorrow, January 15, the Bluetooth SIG is going to make an announcement of a new Bluetooth standard, the Bluetooth 5.1. Our Bluetooth 5.1 IPs for low energy and dual mode, both are already available off the shelf. During the last few months, we have signed a few deals with these two both IPs. The next example is related to the Wi-Fi. As Franz just mentioned, the next thing in Wi-Fi is the 11ax. Our 11ax 1x1 IP is already available off the shelf.
As Franz mentioned, we have already signed a few deals related to this IP, and more deals are just behind the corner. Why we are so successful? The first bullet is very important. As you know, most of the IoT devices are battery powered. In light of that, power consumption is a critical factor. We know that when our customers search the market for a new IP, and when they evaluate the different IP options they have, the main selection criteria is power consumption. In light of that, at CEVA, we give a special attention to power consumption. Our design methodology includes special stages, special action to assure the final product is attractive in terms of power consumption.
As a result of this design methodology, some of our customers, when they make an announcement of the new Bluetooth Low Energy chip with our technology inside, each one of them make the statement that their design, their chip, are best in class, best in the world in term of power consumption. The second bullet relate to the integration aspect of the issue. We offer our customer the option to license a fully integrated platform. By fully integrated platform, I mean to our IP already integrated with some subsystem element and with a microprocessor. As you understand, we save a lot by that in the integration work that the customer need to be done. By that, we reduce significantly the risk involved with integration, reduce the integration effort and the time to the market. Bluetooth and Wi-Fi combination is a very interesting combination.
As you may know, there are a few chips in the market that already offer this kind of combo. For example, chips that come from Broadcom. Today, we are the only IP company to offer such a combination. If you continue in this line of thought, you know CEVA has a very wide portfolio of IP. Cross-selling of the Bluetooth and/or the Wi-Fi with some other CEVA IP can be very attractive to our customer. I would like to focus a bit on Bluetooth. I will get back to the Wi-Fi in few slides. We are by far the market leader. We have been licensing Bluetooth since the year 2000. Just during the last two, three years, we have more than 50 design wins. For Bluetooth Low Energy, there are few options in the market.
That is, there are few licensing company that offer this IP. For Bluetooth dual mode, we are the only option. That is, we are the only IP company that offer Bluetooth dual mode IP. If a customer is interested only in Bluetooth dual mode or in Bluetooth Low Energy and Bluetooth dual mode, we are the only options they have. One of the critical factor in our success is the acquisition of RivieraWaves that was done back in 2014. Gideon said a few words about that earlier today in the morning. This was really a critical factor in our success. The market response to the acquisition was outstanding, and this was translated very fast to a high number of new deals. The graph that you see here represent shipment of devices with CEVA Bluetooth IP inside.
The first results of the acquisition can be seen in the number related to the year 2015. If the number related to the 2014 actually represent shipment of devices with old CEVA before the acquisition Bluetooth IP, the number of 2015 represent both shipment of devices with CEVA Bluetooth IP and shipment of devices with RivieraWaves Bluetooth IP. As then you can see, it was a significant step, something like 3x, just between 2014 and 2015. The next implication, the next result of the acquisition can be seen in the numbers related to 2017 and 2018. As I mentioned earlier, there was a significant positive impact of the acquisition, and the number of deals went up significantly. Before the acquisition, each company signed something like five, six Bluetooth deals per year. After the acquisition, this number went up to something between 15 to 20 new deals.
For a company from design start, from the point they license the IP till they reach mass production, till we see royalties, it takes two, three, four years, something like that. It depends on the application. We start in 2017, we start seeing the results of the high number of deals which were signed after the acquisition. As you can see in the graph, it more significantly can be seen in the year 2018, and this will continue forward. I would like to say a few words about some of our customers. Dialog. Dialog is one of the Bluetooth Low Energy market leader. Sometimes they are number 2, sometimes they are number 3. They use our Bluetooth Low Energy IP. Beken. Beken is less known. They are a Chinese company. They are one of the Chinese Bluetooth market leaders. Their production volume is just huge.
We have been working with Beken very closely during the last 10 years. I think they license all our Bluetooth IP. Whenever we come with a new IP, it takes a while, they license it, they put it in one of their design, and after a while, they start shipping like hell. They are very successful. Sonova. Sonova was mentioned earlier in the morning. They are the hearing aid market leader. They have a very high-end product, which include our Bluetooth. The last, it's ON Semiconductor. They license our Bluetooth Low Energy IP. This guy, the first target was hearing aid, but once they realized that their chip is one of the best in the world in term of power consumption, they changed the target, and now they target the general market, and most likely, they will join Dialog as one of the BLE market leader.
What is our strategy related to the Bluetooth? I mentioned earlier that whenever the Bluetooth SIG comes with a new standard, we are the first to introduce IP which supports this standard. It is not a coincidence. This is exactly our strategy. We have two guys from our team who participate in the Bluetooth SIG working group community. In this way, on one hand, we have the chance to influence about the content of the next generation, and on the other, we have full visibility. We know the content, and we know the timing. We know exactly when is the right time to start a new design. We know exactly what to design. We are ready few months before the ratification.
On top of that, we focus on audio over Bluetooth, and by that, we are trying to take advantage of a CEVA leadership position, both in Bluetooth and in audio market segment. Cross-selling. We are trying to promote cross-selling of Bluetooth and some others of CEVA IPs. Last, we are trying to provide more value to our customer. I mentioned earlier the fully integrated platform. We offer also Bluetooth Mesh, we offer Bluetooth Modem, and we offer Bluetooth RF. Getting back to the Wi-Fi. We start dealing with Wi-Fi after the acquisition of RivieraWaves. RivieraWaves starts licensing Wi-Fi back in 2002. Far, we have many licensees, many customers from all around the world. Our Wi-Fi IP portfolio, it is quite wide.
We have many IPs, starting from the low end, from the one by one, which targets the IoT application, going through two by two for smart home and handset smartphone, up to four by four and more, targeting the high-end access point application. Talking about some of our customers, Beken here again, they license our Wi-Fi IP after the acquisition. And Beken like Beken. When they starts production, it is huge volume. Huge volume also in Wi-Fi. ASR is a Chinese start-up company. They license the 11ac one by one, and also Bluetooth dual mode, a kind of a combo, and their target application is smartphone. Celeno, Israeli company, targets high-end access point. They license our 11ac four by four and CEVA DSP. They already started production. Last, with this nice special logo, this is a Chinese start-up company called SeaFlower.
They license the 11ac one by one, targeting the low-end access point. They are already in production and already started the next generation. In Wi-Fi, our strategy is as follows: We focus on the 11ax. This is a game changer, and we want to be a market leader in this area. We are trying to promote CEVA DSP in order to enable a soft application, a soft implementation, which is better fit for access point. The cross-selling of Wi-Fi with other CEVA IP is very compelling, and we also here trying to provide more value to our customer with the fully integrated platform and with additional layer in the Wi-Fi solution, such as SCPIP, security and crypto engine. We do that in hardware, and also we provide integration with Alibaba operating system and with Amazon Cloud service software. Our key takeaways for connectivity activity.
We are the Bluetooth and the Wi-Fi. The Bluetooth and the Wi-Fi are very large market segment. Our offering include a very unique IP portfolio that enable our customer to integrate the Bluetooth and the Wi-Fi with other CEVA IP and to gain advantage. Because of that, we expect our customer base and our market share to increase. In one sentence, we are the world's leader of the Bluetooth and the Wi-Fi market segment. We expect our customer base to grow and to increase our market share. Thank you very much
Hi, everyone. My name is Moshe Sheier, and I'll be talking about a very exciting topic for CEVA of sound processing. In sound, wherever you look, you see statements like this. When it comes to consumer audio, we're back in boom time. Everybody says that from different angles. It's very exciting times for audio. For us at CEVA, it's a huge addressable market with very diverse applications, and it's divided into three main segments. First one is voice-enabled IoT. Everybody wants the Alexa experience today. Voice user interface in smart home, automotive, wearables. That's the thing. That's what everybody wants. That's a major market for us. Second one is headphones. After market headphones, headsets, earbuds. Everybody wants the in-ear, truly wireless earbuds. That's a second huge market. The third one is mobile, of course.
In mobile, more and more AI or sound AI is getting important, and people want to include that. That's where we see our main addressable markets for sound processing. Now, why is it so exciting? Have a look at what happened in the last decade, what we call mobile convergence. We started to use our smartphones and ditched our MP3 players, our fancy audio equipment, and just moved on to use our smartphones. Came the time of the mobile accessory. We all bought aftermarket headsets. We bought Bluetooth speakers to enhance our mobile audio experience. In the last three, four years, and still going on now, smart audio. Smart audio is all about consuming streaming audio together with voice activation wherever we are.
This is what causing this boom time with 15% annual growth rate for the last few years and going into the next few years. That's why sound is so exciting for us and in general. Have a look at the volumes. Going from the modest automotive car volume, 630 million units of voice-enabled IoT, more than 1 billion headphones. We're talking here about active and wireless headphones. 2.2 billion mobile phones. That's the addressable market. There is 4 billion devices here, and this is what we are targeting with our sound technologies in 3 years' time. Voice-enabled IoT, the first market. Let's go through them. Voice is certainly the new user interface. Everybody wants that. Why voice? First of all, it's natural, it's intuitive, it's personalized. It can recognize you and not allow other people to use whatever system you're using.
It is hands-free, eyes-free. You can use it wherever you are, no touch needed. It's device-free anywhere and very cost-efficient and small form factor. All you need really is a microphone and a low-power voice activation chip, there you go. Voice-enabled IoT. That's what we're good at, and that's the first main target market for us in sound. Smart home, we look at that specifically under the voice-enabled IoT. Look at the volumes here that's showing the growth rates of different home segments. The biggest one by far is the smart speaker. 32% growth rate for smart speaker in the home annually. Everybody wants that. It's not just the Amazon and Google devices. It's third-party devices that are looking for the Alexa experience and are looking to integrate that voice-enabled capability. That's our first huge addressable market.
Second thing is headphones are heading a very clear way. They are going wireless, they are going in-ear. We see that every day. We look at the volumes, we're talking about 1 billion devices. We're talking here about the wireless segment and the active wired segment. These two higher ones. About 1 billion, some say a bit more than 1 billion devices in 2, 3 years' time, just for the aftermarket headphone market. We try to dissect that by form factor, what we see is that the in-ear, these two elements over here, the in-ear experience and specifically the in-ear true wireless, the one we see here in the middle, that's having the fastest growth. That is the experience that users are looking for today.
They want to go wireless, they want to go in-ear, they want to have voice activation from the previous segment. That's a major market. Again, 1 billion plus devices are out there looking for sound and wireless technologies. The third one, mobile. Mobile has always been huge, but sound AI is getting very, very important for mobile. Close to 1 billion devices are expected to be enabled in the new breed of smartphones that will come in 2, 3 years' time. Some people have started to use this term, sound processing unit. That's a new term that it can either be a standalone chip or it can be an embedded processor within a larger chip. That's sound processing unit, and that's the new sound AI capabilities, not just voice activation, but biometrics and understanding your environment, where are you, all according to sound.
The sound target markets, we look at what's happening in them, our speakers have become smart speakers. All the Bluetooth speakers have become voice-enabled. They have voice assistants in them. We look at headphones, like I said, they are becoming truly wireless, voice control. Everybody wants active noise control today. Everybody wants the AirPods experience. Changing as well. Mobile, the third market, voice assistant, sound sensing, all the changes that we are seeing and will accelerate in the next few years. The message here is that everything is getting smarter. All these segments are going through rapid growth and are all getting smarter. This requires a lot of DSP technologies, which I'll talk about in the second section.
Just look at this big shopping list of software technologies or DSP technologies that are required, multi-microphone, beam forming, echo cancellation, voice biometrics, voice activation, positional audio, and so on. Many, many things which need to run on a tiny low-power DSP for voice activation or a high-performance DSP for multi-channel and multi-microphone sound processing. Many things which are needed, which leads us into the next session, our strategy. How are we handling all that huge 4 billion devices of addressable market in the next two, three years? We look at that by seeing what we have today. We have more than 30 active sound customers across diverse markets, and that's all part of the royalty streams that we are seeing at CEVA today. It goes from the mobile and headphone side of things, the smartphones, the smartwatches, the earbuds, of course, into devices like these.
These are real devices, some of them you can see out here, which are enabled by CEVA DSP, and the voice-enabled IoT, the more recent domain with smart speakers, with action cameras, which we are in all these types of devices. From speakers to action cams, all powered by CEVA already today. When we look forward, what do we offer our customers in terms of the value proposition for sound? This is built around three elements. The first one is the DSP, the traditional, the legacy, the things we've been doing for many years, ranging from low-power DSPs over here, down to a more high performance, and the recently announced CEVA-BX architecture. That's a hybrid architecture that handles DSP, control, and AI. That's the first element of the strategy. We have software. There is a lot of software technologies needed, as we've seen before.
We offer two main software packages today named ClearVox and WhisPro. That's the noise reduction capability. That's the voice activation capability. These two software packages, which enhance and allow voice activation to be successful from far-field in noisy environments, they are mandatory in all these markets and segments that I mentioned. The third element is the AI. We started with NeuPro, but we are supporting AI within our sound product line as well. We do that by providing the necessary infrastructure, as in neural network compute libraries, as in neural network frameworks, to allow our customers to map, to infer neural networks on the edge on our sound DSPs. That's our three-leg strategy and value proposition for our sound customers. It's very unique. No other IP provider is offering these three elements in the value proposition. Breaking down quickly to a few technologies.
CEVA-BX, our latest DSP architecture for sound processing and general control and AI, was just announced, just released last week. The message here is that we really provide much higher DSP performance, 4x more performance compared to our own previous generation, CEVA-X architecture. Everybody talks here today about the requirements and performance that needs to go high all the time. We did a 4x jump with this family. We're also able to reduce the code size, meaning the memory size required. Today's applications, never mind where you look, you need both, what we call signal processing, as well as real-time control. That means a lot of code. That means people want to map a lot of code, generic code, into the machine, whether it's sound or other general-purpose workloads.
We enable that with that architecture, which is provided using two versions, two variants, a smaller one, a larger, more powerful one of a DSP. That's our most modern, latest architecture, very high level programmable, very easy to use, which helps us to establish ourselves in the first element of the strategy, the DSP. Moving on to the software. Two main elements here. The ClearVox, which is the front-end voice processing software solution. This is your far-field voice pickup in noisy environments for noise reduction. Handles echo cancellation if the device is playing music. This is mandatory for any voice-activated device, whether it's a wearable or whether it's a fancy smart speaker with seven channels. They all need voice activation, and they all need these noise reduction echo cancellation technologies.
We offer that package as two versions, what we call ClearVox Universal for many IoT devices, and a specialized version for headsets. A smaller form factor is needed to go in the very size and power-constrained in ear earbud applications. The second software element, just announced, WhisPro, that's a speech recognition. That's the second critical software element in the voice activation ecosystem. Everybody needs voice activation, always listening, trigger phrase, identifying the keyword of your choice. We provide that in-house today. It is based on a neural network speech recognition technology, and we allow our customers to select and customize their own triggers. Not everybody wants Amazon and Google, which by the way, you can get those triggers from these vendors, but people want the "Hi, Samsung" or the "Hi, something else," and they can get that from us using the WhisPro package.
Second element of the strategy, the software. We have the AI. Sound AI on the edge. I showed you the almost 1 billion devices, sound processing units that will be needed for mobile phones. Sound AI at the edge is mandatory today. People want privacy, they want reliability, they don't want to rely on the cloud, they want low latency, and they want very low power and cost. They achieve that with neural networks. Neural networks involve two sections. We have the deep learning, the offline training, what we call, and we have the edge inference.
Neural network on the edge to classify, to filter different sound events, that's what we enable on our DSPs, and we do that with the infrastructure, the compute libraries, the frameworks, and this allows speech recognition, this allows sound sensing, and it allows our customers to develop their own neural networks for whatever application it is. DSP plus software plus AI at the edge, that's what we're offering to our customers. This is what we go to the market, the one-stop shop for all sound processing elements. We leverage our 25 years of experience and expertise in low-power DSPs. That's the first critical element. We have the value-added software.
Everybody who wants the Amazon experience or the in-ear earbuds experience needs that kind of software for noise reduction, for speech recognition, and the sound AI, the infrastructure to allow people to easily implement sound AI at the edge. Couple all that with the high synergy we have with our connectivity portfolio, which Franz and Aviv just described. Bluetooth for the headphone market and Wi-Fi for the smartphone market. That is what we go to into the expanding, growing, very exciting sound market today. Thank you.
Okay. Thanks so much. We're going to have Q&A now for connectivity and the sound processing. Please, any questions, we will hand the mic this time, hopefully it will work a little bit better. Do we have questions?
Me?
Sure.
Thanks. I'm starting. In the U.S., the voice user interface is dominated by the Amazon smart speaker, Google, and maybe the AirPods to an extent. What are you seeing outside the U.S. in terms of China, for example, as a market? How is voice user interface being deployed in the fashion, the home, et cetera?
I would say the following. If you look on China, it's a replication of what's going on in the U.S. They have their Alibaba, JD, all those guys to do it. For their standpoint, I think going forward, it will be the U.S. It doesn't make sense that you have in your home three types of smart speaker, one for Google, smart for Amazon, one whatever.
To do. They'll try to unify it in a way. All of them, Google, Amazon, they have third-party programs that allows them. There are a bunch of, Sonos, for example, is supporting multiple. That's what we are giving because what Moshe presented, our AI network, WhisPro, is basically, it's a network that supports vocabulary. It's not just Alexa or Okay Google, it's Alexa, Okay Google, or Hey Siri, or whatever, because from our standpoint, we have the technology base to support multiple, and we are going to take the same technology to the next level, and this is what is called ASR, automatic speech recognition. Meaning, think about your car, okay. You want to do multiple things, to control the air condition, to control the radio, to control the volume, much bigger vocabulary. That's where we are going with our neural network.
That's what giving us the neural network. Neural network allow ourselves to unify and to support multiple keywords, vocabulary. That's part of the trend of moving the smart, the voice recognition, the NLP, the natural language processing into the devices for privacy, for sophistication.
Just to add one second. We have customers already in Asia that are adding voice content activation to set-top boxes. We saw last week at CES, one of the CSP group winning voice activations designs in garbage cans and all over the city. There's lots of stuff where it's not connected to a cloud at all, so just localized processing for some sort of function. That's actually happening around. Question?
Yes. How much vocabulary can you currently support without being connected to the cloud? Do you have one, two, three commands, 10 commands? What is it?
The technology with the neural network we have developed, there's no limitation to support. In the area of the neural network, this is the area what is called data science. Meaning, you need to build a database for that purposes. When I said we move from one, two, three keywords to develop to a vocabulary, it means to building a bigger database, and this takes time, and it depends the use case. We can decide to go to this area or the other area. From neural network capability of the company, we can support a much bigger vocabulary.
I think if I may add to that. We see the sweet spot today between one or two keywords, up to maybe dozens, 20, 30 commands, in a certain context, whether it's a smart home or an audio player or an action camera. After 20, 30 commands, you stop remembering them. That's the sweet spot, and we can handle that. It goes beyond that, like Gideon said, large vocabulary and speech to intent, as in NLP at the edge, natural language processing. It just requires more computing and more memory, but we can handle that.
Yeah. Question on, I can see the attraction of the newcomers to the space to play with your sound portfolio. What gives you the confidence that you can crack the tier 1 chip makers? It has to be made in-house, it wasn't very important. People, companies, big companies started to do that. What gives you, or what makes you think that you can crack into that?
It's difficult to understand. You said what? Can you repeat the question?
Yeah. What gives you the confidence you can crack big chip makers for sound, versus previously they always did it themselves?
No, I don't think they did it by themselves, in the past. The area of voice, it's an area that is scarce. You don't have that many people that understand voice. Keep in mind that it's not just the neural network voice recognition. This is the area of noise cancellation, echo cancellation. You need to do it at the front of the voice recognition in order to get good recognition. There are not that many companies that do it, not in the semiconductors. There was few in the past, but not anymore area. We went into this space because this speciality is scarce. The algorithms are not open source, are all proprietary, and you need people and the DNA to do it.
Gideon, talk about the jump from the cloud to the edge, gives me that question.
When you do things in the cloud, like Amazon and Alexa, you don't have the restrictions in terms of performance. You can take very high-level algorithms and let the cloud does it. When you move to the edge, then you have to make a lot of trade-offs. That's something that requires not just software. It's a new type of algorithms, because what you have in the device itself, the condition, the environment around device is much difficult than a cloud.
Maybe a follow-on. The value-added software and ClearVox , do you have the ability to charge for that, or is that more of a lock-in?
No. We do it in order to charge, to get ASP, to get higher ASP. It's a value add to our offering. The deals that we signed, they are all included a component for the software. Royalty component for the software.
That percentage is the model?
Same model. All the same.
Previously, I remember we talked this before, where there's about $0.40 of royalty going into these types of devices, $0.10 that goes to hardware versus software. Now we're adding to the software royalty base as well. It remains at a lower price, there's a bigger market opportunity there. You still have Tier 1s and so on, who can do their own pieces, but there's so many customers looking for voice space now, pretty hot market. You do have some experts out there, but it's very limited, and not too many emerging companies who need voice for the first time. There's not many options out there.
On the connectivity side, historically, Wi-Fi has not been the connectivity of choice behind IoT device . What is the three big things changing now that are going to allow Wi-Fi to come in where Zigbee and Z-Wave and Bluetooth have kind of dominated up until you get to the IoT gateway? Why is Wi-Fi now going to the devices themselves?
I think what enabled the Wi-Fi is the smart home. Smart speaker, smart TV, they all will be enabled with Wi-Fi, because whenever you want to have a longer range and stream video, you need to have the Wi-Fi. That's one area, the smart home, that we're going to do. I think when it comes to residential gateway, we used to have gateways that we get from either buying in an electronic store, or you would buy it or get it from telecom. Today, with the smart home, you're going to get to have a gateway that are different than just data. You mentioned Zigbee, you mentioned Bluetooth, it's part of the whole connected devices in the home. This will be provided by other vendors that will get into this market because it could be connected to 5G, to the internet, or stuff like this.
We're going to see new devices offered by new guys like Amazon or others that will try to get the service out of it, they will need these kind of technologies like Wi-Fi and others. Overall, on top of the movement to AX and the natural things that happen in the Wi-Fi, the evolution, there are new devices that are led by the smart home.
Do you see Wi-Fi extending to battery-operated devices as opposed to plug-in-the-wall devices?
Yes. A good example are surveillance camera in the home. We see more and more opportunities there for Wi-Fi because it can connect directly to your box at home, and it's getting lower and lower power. There are already some products on the market today with just a battery, which are quite good, with a few months battery life, six months, something like that, which is decent, but that will be improving over the time.
You think about as opposed to Bluetooth, which is peer-to-peer, also it's changing now also, but Wi-Fi, it could be much longer range. We had a customer in the past, I don't know the exact status of it now, but they use it for medical. Things are changing, and there is no new obligation from the start standpoint.
Okay. If there's no more questions, we'll have Issachar go and present some insights into the China market now.
My name is Issachar Ohana. I'm the EVP Worldwide Sales at CEVA. China is the biggest market for us, biggest region in terms of revenues, and we decided to give you some insight into China, the way we see, the way it affects our business. I'll talk about the semiconductor market in China, and derived from that, some market opportunities, give you some insight into customers in China, some of which you may have never heard of, and a quick summary. China in the past was disregarded as a market follower, and it's no more the case. We see a lot of innovation today in China. Starting with the semiconductor. People like to look at the market in four segments: IDM, fabless
Foundry and assembly and testing. If you look at percentage of revenues in 2015, China was not anywhere close to IDM activity. Only 10% in fabless activity, 7% in foundry, 12% in packaging and testing, bigger than others, but still small in terms of the worldwide market. On the other hand, China was the driver in the last two decades in the semiconductor industry. China was the largest customer for those devices in the worldwide market. This created imbalance between the supply and demand in China, when you look at the domestic market. Also, those activities of manufacturing and packaging are considered low tech in semiconductor industry. In the last two decades or 10 years, China has been growing in where the sophistication is in design. They have made tremendous increase of about 30% annually and increasing their capability in design.
This is where we see them, when they are designing. Made in China 2025, I'm sure everyone heard about it, Made in China 2025. It's a strategic plan by the government. It's a very aggressive plan. They would like to move from $67 billion of semiconductors in 2017 to $305 billion by 2030. They would like to go from 33% of domestic supply that they are meeting today to 80%. Basically, the aim is to close the gap between the supply and demand, and to reduce dependency on other. This is creating opportunities for us because we are here to help people do their own products, do their own design. I'll go and talk about five segments of the market that affecting us. There are many more, but those are big and correlating with what you have seen today.
For example, if we talk about 5G, in mid of last year, there were 1.1 billion mobile subscribers of 4G LTE. More than the population of U.S., Indonesia, Russia, Japan, Germany combined. However, the technology for those products, for LTE, was mostly made and developed outside of China. In 5G, China is planning to get out of it. They see it as a chance, since it's a completely different approach to wireless communication. They would like to invest, and they are investing heavily to get in front and change their position in the worldwide market. When it comes to 5G, given this plan, and as part of the Made in China 2025, they are actually pushing it and see it as a way, one, to meet the target of reaching 82% of the population connected to broadband internet.
This relates to things like what we've seen before, fixed wireless that allows them to get to more homes. They would like to close the gap and get 40% of the domestic mobile phone IC supplied to those industries. They would like even to export and become a world-leading maker of telecom equipment. This also going to help them with other markets like IoT, self-driving car, industrial automation, et cetera. As part of this, if in 4G they had 7% of the IPR, here they are investing heavily and would like to be more independent, have some more power in IPR. By early 2017, they had 10% IPR of 5G. I guess that now, early 2019, they have more than that.
They invested also in IPR, not just in the. When you look at the deployment, China has 14 sites for every 10,000 people, compared to 4.7 in the U.S. As you have seen earlier today, 5G, by definition, needs more towers, more base stations. This will grow. China Tower added approximately 460 sites per day. They recently raised close to $7 billion in one of the largest IPO in the last two years. Most of it is going for 5G deployment. It's a huge power that is being put into this. They added more towers in the last three months than what U.S. added in the last three years, just to give you an idea. 350 sites in China versus 300,000 in U.S., which is considered a big market. Again, just to give you a comparison, it's about $24 billion more invested. They are taking it seriously.
Another market, AI. We talked today about AI. China government here as well, everyone knows about it, they are planning to lead the world in AI. I'd like to quote Google's Eric Schmidt. He said, "It's pretty simple. By 2020, they will have caught up. By 2025, they'll be better than us. By 2030, they will dominate the industry for AI." This is coming from U.S. executive, they are very aggressive. You can see in July 2017, the government have put together a plan and committed $150 billion. Within a year, Chinese VC has been pouring money, some of this probably coming from government, and they are already investing more than anyone else. 48% of the global funding for AI is in China. There are 168 unicorn in this field as of April 2018. There are project like smart cities.
One of them is close to Beijing, Xiong'an, and everything is based on AI, autonomous shutters, autonomous driving. Again, those project, Elephant, are attracting a lot of funding, a lot of investment. Another third market, surveillance. China is expected to reach $24 billion in 2020, growing at pace of 15%. Again, there are some big plans such as road infrastructure, safe cities or smart cities, and other rural areas that they will start covering that are driving the growth. You can see the market as a whole, we are here to benefit from all of it. If you look at the market, about 45% is in China alone compared to the other part of the world. Each of this camera need technologies that we have in the form of analysis, AI, as you have seen earlier today. It's a big market. Automotive.
Automotive is also a big market. They are already more than U.S., Japan, and India combined. Big market. They are expected to reach $37 million by 2025, 25% of which will be level 2 and level 3, and they'll start having level 4 and 5 entering the market. If you look at all the big telcos or big companies like Baidu, JD.com, Alibaba, they're all investing in autonomous cars. China is likely to become the world's largest market for autonomous vehicles and mobility service. The last market that I will be covering today is IoT. IoT is heavily backed by the government as well, and China is betting big on IoT. They are already a major supplier of components worldwide.
By 2020, there will be $200 billion IoT-connected components in the world, China, according to the forecast that I've been reading, will probably supply 95% of all of it. The market for IoT in China is growing today at a pace of 25% per annum. All of this very impressive and creating a lot of opportunities for CEVA. I'll go to the next part of my presentation and give you some insights about some of our customer. I couldn't cover in the time given all of them, but try to go through some of them. First is UNISOC. UNISOC is the leading fabless semiconductors for mobile communication and IoT, covering all the way from 2G to 5G.
It's a subsidiary of Tsinghua group, Tsinghua Unigroup, with more than 4,500 staff, 14 design centers worldwide. It was created as a combination of Spreadtrum and RDA, both used to be traded in Nasdaq. RDA was acquired for close to $1 billion, Spreadtrum for about $1.7 billion. Later, Intel in 2014 invested $1.5 billion in this company, just to give you a sense of the value created. They are at the mid-range of the market. They are forecasting 20% growth in India in 2019 alone. They're also focusing in other third world. They are strong in Africa, South America, and China. There were some changes recently. A new CEO was appointed in November 2018, I expect to see some changes. ZTE, a leader in communication and IT, publicly traded, with customers in 160 countries around the world.
Tier 1 cellular infrastructure, one out of the 5 that we mentioned today. CEVA is powering their cellular base station. I can say exclusively powering the cellular base station. CEVA is also powering their IoT. You probably all wonder what's going on after the ban was lifted. They are back to business today. As of today, they get to the same level as before the shutdown. They were completely shut down. They resumed operation to the same level. They are a key contributor for 5G standard values. DJI, a world leader in commercial and civilian drones. 70% market share. Another company that shows amazing innovation in creating a new segment. They're using CEVA for the proprietary communication from the drone, from the remote control to the drone, getting, by the way, amazing range and quality.
They're also using CEVA computer vision and artificial intelligence for collision avoidance. They are getting recognition worldwide, some of which is the Emmy Award. Artosyn. I'm not sure anyone heard about them. We have demo here from Artosyn. They are using CEVA for AI and computer vision, high-performance intelligence vision, located in Shanghai, using CEVA for object detection, classification, tracking, targeting robotics, drones, VR, surveillance, ADAS, and they're already deployed in multiple products, robots, home robots, unmanned retail, face, and object detection. This is the, how they call it, autonomous attendance or whatever. They replace a receptionist over there in many companies, and you can see the logos their customer. Another company that is impressive, founded only in 2015, they already have product, Bestechnic, excellent low power, ultra-low power RF powered by CEVA Bluetooth.
They are in wireless speakers, wireless stereo, wireless AirPods, one of which is here, the Huawei FreeBuds. Apple deserve the credit for the product innovation, but they took it to the next level with voice ID for e-commerce, Alipay, WeChat, et cetera, voice assistant, very advanced technology. Just recently, they won another tier 1 company in the U.S. I cannot disclose the name, but you all know this company. We'll be able to publish it probably soon. Beken. IC solution company for wireless application using CEVA DSP, Wi-Fi, Bluetooth. Just approved for IPO in China. They'll go public soon. Low power Bluetooth devices, low power Bluetooth audio, low power IoT, and more around those things. Espressif. I'm not sure if anyone heard about them. They present themselves as multinational. They are exported in China, they have offices in India and Europe. They present themselves as multinational.
They would like to go globally. Low power IoT, Wi-Fi, and Bluetooth using CEVA Bluetooth. They were getting some recognition by Gartner, partnering with the big names in the industry, Baidu, Huawei, Haier, Microsoft, Amazon, Xiaomi, Alibaba, JD.com, Apple, and recently got investment from Intel Capital. This is about the companies that I cover. There are more, as I said, but we had to squeeze everything to short time. Summary or to summarize, we have an excellent fit for what's happening now in China in terms of technology. We are enabling companies around the world. China is certainly an opportunity, and I believe we are well-positioned to enjoy from this drive toward homegrown products that they would like to achieve. We have deployed team in Shanghai, Beijing, Shenzhen. From there, we cover the whole country.
The team in China is actually quite independent. With the help of the R&D, we pretty much cover all what we need over there. That's it. Thank you.
Good afternoon, everyone. Thanks for staying this long for the last presentation of the day. It's exciting. It's something that we've talked about for the last couple of years. Finally, we get together and spend enough time to understand CEVA's businesses and product lines much more in-depth than we've ever had done. The nice thing about it is that it gives us the opportunity to talk a little bit
Further longer term and not short quarter, what's this announcement all about? How does that impact you? Really to look at CEVA in a more, a lot of people use it here, holistic way, and what do we do exactly and where do we want to go? I'll try to summarize the many different presentations that you got today and see how we build and now add a little bit more numbers, flavors, and understanding of how all this could work together. As you realize, we have a potential for revenue growth type of business with many, many different products. This diversification strategy that you've heard of throughout the day has really paid off in different aspects of us winning design wins, increasing market shares, and building a business in all the different aspects that we talked about, which is pretty much identical.
It's an upfront license fee and then a royalty scheme as part of the success of our company as soon as they go into production. That does not change throughout any of the business lines and any of the different metrics that we talked today. What also makes sense is still when we look historically and analyze the business and try to forecast a few years out, is that the licensing activity and the dollars that our semiconductor customers or OEMs have put to play in licensing technologies from us, is a precursor to a royalty stream. Yes, it could take time. Not all markets have the same lead time until we see those royalty streams. Sometimes you lose a little bit hair on the way, waiting for those royalty stream, that's all part of the game.
The margin expansion happens, and we've seen it again in prior cycles at CEVA when the royalties kick in, and to what extent the royalties kick in. If we look on the other side of not money coming in, but money being spent and the capital allocation of the company for the last couple of years, a lot of it, I would say the biggest chunk of it, is R&D investments. M&A, which we heard about, and I'll cover in a second, and stock buyback. Eventually, we're trying to build a model and to build a business case around this model, which is focused on delivering long-term shareholder value.
If I summarize all the different markets, all the different product lines, step back a few years, how part of you guys knew us back in 2013, dealing mainly with modems here, a little bit more communication DSPs and multimedia. In a very short period of time, when we understood that we need to diversify and we want to diversify because this is the right thing to do, especially in the technology world, we managed to increase our TAM 3x from 3 to 10 billion units. This is our addressable market today. When I talk about 2022 in a few more slides, you'll see that that number even further increases when we summarize all the different parts that you've seen in the presentations today. For every one of these green circles today, we have discussed in length the market opportunities, our technologies.
I think you've realized quite well that CEVA is really a very detailed-oriented technology company. Our DNA is technology. It's technology-centric. All the presenters today, whether it's the BU managers, whether it's the marketing guys, whether it's Issachar from sales or Gideon, the CEO, maybe I'm the only one that have to step aside, is all coming from an engineering background. This is the depth of understanding of technology and how we manage to build all these nice different markets. It's almost a company or multiple different companies from a lot of different angles, but tied together with the same IP model and the same concept and the same idea to grow and succeed there. If I focus a little bit more on R&D, and these are questions that we've been asked over the years. What is this R&D all about? Are you giving services?
Has the business model changed a bit from licensing and getting royalties? There's not a single deal that you've heard about or customer that we have discussed today that in their business model with us, do not need to pay royalties at the end of the day. If they are successful, we'll see the royalty stream. The R&D dollars that we have spent, and these are big amounts, $250 million in the past decade, was really to grow our markets, to get into new markets, and to be able to present you all these new technologies that maybe a decade ago, if we would have an analyst day and only talk about 2G and 3G modems, we probably within two hours could wrap up and get back to the office.
You've seen how diversified, how different each market is, and even in some markets, it's not enough to have one product. You need to come up with a portfolio of products. It's not enough to have a hardware solution, but you need to add the softer piece on top of that in order to have a real offering and win a market. R&D personnel, which is the bulk of the R&D investments and really the bulk of the essence of CEVA, is doubled when we start this diversification. End of 2013, just before ZeroWave, 125 to close to 280 people, engineers, all doing different designs, hardware, software, AI, Bluetooth, Wi-Fi, all over, all the different aspects sound that we have discussed today. We've managed to grow the licensing revenue.
We've managed to grow the royalties, not to the maximum extent that we think they could grow for now, but this is how those R&D dollars were put to place and to work. On top of that, this M&A strategy that we've been doing it small, but with the two interesting small acquisitions or investments, Bluetooth and Wi-Fi, was, as Aviv said, a huge success story for us. As you heard from Michael, the Narrowband IoT investment that we did with RF and the full solution of being not just a modem company, but the whole solution, really enables us to get 10 deals in the last year just on that. Small, very strategic investments. The idea is that this technology could be cross-sold to other customers of ours or taken into new markets and new customers and expand our offering.
Usually, we try to find and look for mature technologies, that we're not going to start some new concept and see if it works or not. We want to have something of essence or maturity. Part of you know us for many years, we are very bottom line-oriented, and if we do an acquisition down the road like we have done before, we would like to see a very nice ROI or a quick ROI and try to build something accretive in a very short period of time. We haven't finished the year yet, I am showing you the first nine quarters of 2018. A few weeks from now, we'll have the full picture here.
If we compare it to the minute before the diversification versus where we are today, throughout this diversification, we talked about the TAM increase, but also the revenue and the breakdown of our overall revenues, 34% a few years ago, 56% or higher even, when we look at it today. The idea is to be able to grow both segments. There are cycles in the industry. We are aware of that. Handsets were not a strong 2018 driver for us in royalties. Overall, if we keep both of them and nourish both of them and deal with the non-baseband, the diversification strategy, that will continue to work out for us on the longer-term model.
Here are some few different data points that we've gathered as part of the presentation today to try to show and then see for ourselves how CEVA was before this diversified portfolio that you've heard about in very detailed today and how we were before, to summarize that 10-year anniversary. From a licensing perspective, before 2014, it was about flat, mainly dealing with modems, as we've discussed before. Much higher, step function higher when we added all the new markets year after year. Every year, something new came out and was added to our portfolio. Over a 10-year period of time, we generate $280 million of licensing revenue. $250 was put to work in R&D. That's the manufacturing plant. $280 million of licensing deals.
When we look at the deals signed over those 10 years and look at the colors, the baseband, which was the dominant player historically, has become a much shorter and smaller. This is the traditional handset, just because it takes a lot of time for one generation to the other, from 4G to 5G. It takes a few years. We did discuss the opportunity of getting that light blue darker over time or bigger slide, the big advantage that we brought in and companies licensed our technology came from the non-baseband, 300 non-baseband, non-traditional baseband deals in the last decade, the bulk of them in the last five years. After licensing, come the royalties. Here again, as I mentioned, some take more time, some haven't yet started.
A lot of the markets we have talked about initial steps in licensing. For sure, we're not seeing any royalties yet. Let's look again, start with the units and move to the dollars. If we look at the unit count, today we're celebrating 10 billion devices. The big driver still from a unit and dollar perspective are the traditional modems, 2G, 3G, and 4G. When we look at the dark blue, both in units and in dollars, you realize that in units, every single year from 2014, this is the start of the diversification, including 2018, for the last five years, the volume in our non-handset business has increased. Not all of that is yet reflected in dollars. In dollars, some of the ASPs are very high, like base stations.
Some are very high volumes, like Bluetooth are very low-cost devices that could be $1 or less than $1. This is the market price when you implement some of these devices. We still need more volume. We still need some better spread of diversification within that non-handset business between vision and base stations and Wi-Fi devices and the like. When we summarize everything, in the last decade, 10 billion units and $300 million of royalties. If we invested $250 million, in the last 10 years, we managed to generate $580 million from this business line. If I step again back one slide, the opportunity really for us is the licensing of these new technologies, new products, new markets, that we've discussed about today.
We really don't think yet that we have utilized or got to the full extent of the royalty contribution that we hope to see over the next couple of years. With that said, licensing revenue, how this business model works. We are also trying to maximize and look from the shareholder perspective, whether it's existing shareholders or new ones, and try to see how we build this model right. We're penetrating new markets. We're growing and leveraging our expertise in different segments of the technologies to win significant portions of markets like the base station. You heard today from Michael and from Issachar, some of our customers, we're the sole provider of our technology into base stations. Yes, when we look at a few years from now and we could have a much more significant market share there, that will also show in dollars in the royalty line.
We continue to invest. You've seen that over the years, either small M&As or a few million dollars of increased R&D investments. We'll continue to do that. We think that we have a leading position in many of these markets and could offer the semiconductor space very exotic and interesting technologies, usable technology with talented R&D teams. Last but not least, shareholder buyback activity and return over that same decade. We talked about the revenue, we talked about the investments. We also returned $107 million under different buyback schemes. We did a little bit of an analysis to see how those buybacks and when they took place. You could see about three different time frames, overall, north of 6 million shares. Average price was in the teens, went up to the 30-ish. Overall average of the last 10 years, still lower than market price.
We don't do it with our eyes closed. We do it also from economic point of view. If you also analyze the cash positions over the years, and every time the cash position increased, we started to do a buyback. You could see that over time, we've tried to keep our cash position in the $160, $170 million, more or less, after the acquisitions and after buybacks. This is something that we've been doing for the last couple of years, and no reason not to think that this will continue in the future. This is something that we have proven to do consistently over a relatively long period of time. With that said, we do need the equity to retain engineers.
You've seen in all of these presentations that these are resources that are not easy to get a hold of, and you want to keep them and retain them. Equity is part of the tools that we have to when we compete with Nvidias or Apples or any one of the big players today in the industry that are trying to do AI or exotic new technologies. With that said, we have reduced significantly on burn rates. It's about 2% today. It was much higher in the past. Again, we believe that this is a healthy level to be in, and we don't have any plans to increase it back to 4% that we've had for few years.
I think the ones of you that know us for the last couple of years or that will know us in the next couple of years, the transparency is something that we have always believed in and try to share and work with you. Every quarter, whether it's earnings call, and hopefully we'll have more of these types of events in the future, we try to give very detailed technology and market information, not just things will be great, but really to look for the sources, to understand where the market is heading. You understood from our presentation today that when we design a new product or a new core, it's probably about sometimes five to seven years before the product actually makes it to market. We need to get it right.
We need to get the definition right, the specs, how we want the product to work, how do we sell it, and then, of course, hopefully it's successful and it's a long, long period of time. With that said, we have, and we have always looked on the longer term business and technology bases and build the model this way. Always happy to get feedback, analysts, investors. You guys meet Richard, myself, Gideon, quite a lot, and any other feedback is something that we always take into consideration and we'll continue happily to do so. Where do we go from now? If we look at our licensing business, and again, analyze it historically, it's pretty nice. It's easy to do it retrospectively, a little bit tougher to do it going forward. We had three areas, and we're about to start the fourth.
First one was the traditional 2G, 3G, maybe a little bit 4G modems. 3G, mainly. $20 million a year. For the next couple of years, we started the diversification, started with Wi-Fi and Bluetooth in 2014. Was very successful. Increased our licensing by 50%, around $30 million for a few years. Last two years, we are at the $40 million level when we added Narrowband IoT, very successful product line in 2018. It's a new product line. Sound, we added throughout this year. AI, we added throughout this year. Last two years, managed to grow from $30 to $40. When we look from 2020, because Richard said we're not talking about this year, towards 2022, we could be seeing, if everything works well and the products that we have discussed today will flourish, we could see 10%-20% growth from that $40 million type of step function.
We're adding now, this is something new, Moshe talked about this, voice software. This is a new add-on to our offering with licensing and royalty aspect of it. Hopefully, we have more plans. Not hopefully, we have much more than we could do, plans to increase R&D in products and the licensing activity. As we said before, historically, those $250 million generate more licensing, and of course, generate much more royalties further down the road. From a licensing perspective, this is our plan: continue to work and increase and invest in R&D, maybe find here M&A add-ons that we could add to our offering, and then that step function, of course, will be easier, with the risk of any additional new IP that you buy. This is something we're looking at, and we are seeing a new step function if all this works well.
From a royalty perspective, 2018, $10 billion is the TAM of all the presentations that we have started and heard today. If we step forward three years, and again, I'm not sure if this is the runway of 2022, ending the year, beginning of the year, let's see when we get closer. We're looking at all the different markets that we've discussed today, another 70% growth in TAM. These are some new markets that will involve, some new use cases that we could be a significant player in.
If we did our own homework, look at the customer, looking at the design wins that we have today, looking at what potential markets and who are the players in each of these markets, whether it's our traditional handset to keep our 30% market share, whether it's the new market of the base station stuff that we have excellent idea of how the design cycle is and how the licensing activity, who are the players, we need to see all this in royalties, because this is translating to royalties as well. The bottom line of looking at all these markets, Bluetooth, Wi-Fi, Franz and Aviv talked today, this is the fastest-growing segment in the semiconductor space in the next couple of years, anywhere between 15%-30% growth. This is huge.
Although maybe lower ASPs for us, that will add a lot of volume, and at the end of the day, also dollars. If this all works well, our royalty could double from where we are today, and that will be also mean that instead of one and change billion units that we have powered for the last two years, we will go up to about 3 billion devices from some of these new markets. This is our second step function of the business, licensing first, royalties coming in later, some or mainly from the existing customer base. Of course, some will happen further down the road with the new licensing activity that we have added and plan to add over the next couple of years. Summarizing our goals, targets from a financial perspective.
If we were anywhere between 40-45 deals a year in the last couple of years in this diversification age of 2014-2018, we want to go higher, 50-60 deals, new customer, new markets. That, as we mentioned, will generate 10%-20% growth in licensing starting in 2020. That will double royalties from the $40 million-$45 million level, more or less, to, of course, a higher level. That means also much higher volume in non-handset baseband chips. Operating margin, if we put that to place and continue to invest in R&D, that's what we put in our model, that we look for new markets. We do the same analyst day three years from now, we'll talk about new things and not the existing ones.
North of 30% operating margins, which is again, this year is maybe 2018 is maybe not the best comparison, but it's 2x or north of 2x from where we are today if things work well. That's also from the 2018 leverage, it's 3x EPS opportunity to get into these sockets markets and mainly royalty contributions. Overall, no changes in the positive cash flows that we've always generated over the last couple of years or decade. This will continue. Buyback, M&A, as we mentioned, and if we find add-on components that we think that we could add to the business. Overall, that summarizes products, our R&D efforts, some of the success stories that we talked about today with customers, and how we see the business progressing a few years from now.
I want to take, before the Q&A, the opportunity to thank you guys for spending quite a few hours a day with us. It's not obvious, it's not simple, especially not in today's markets, we really appreciate the efforts. I thank the team, our team, that really, for the first time, took the day-to-day activity that they do and manage in R&D and sales, try to build it in a format that it's user-friendly for others to understand and to understand better, or even to learn. That's part of our goals for this day. Last but not least, Richard, without his help and guidance, we wouldn't have been here, really appreciate all the hard work and effort that got us to the first Analyst Day.
We'll see when the next one is, I think it's a great platform, really, to share much more detailed information than we usually do in conferences or post-earning calls and Q&As. I'm really excited, as I said earlier, from this event, and think it's a great use of time. Last Q&A session of the day.
Sure thing. Thanks for putting out the share targets. If I draw a line from now to, say, 40% in LTE, 5G in 2022, I can imagine there being a linear argument. I can imagine there being a front-end load argument, because you already have three customers, you said two of which are exclusive. What's the shape of that share prediction? I know it's hard to call it exact, but is this more appropriate to give it a front-end loaded getting to that end share market, or is this something that happens steadily out to 2022?
This is why you're a smart analyst, I'm just counting the beans, understanding what's going on, getting the royalty reports and reporting to you, not knowing if there are going to be any ZTE shutdowns or any Nokia delays from time to time. I think your answer is as good as mine. We have the customers, we know where the designs are, we need the patience sometimes we even don't have, but hopefully we'll get there fast. As Issachar said, ZTE is in full scale. They're back. Now, new questions arise. How to divide the new markets, East versus West. It's a whole new world out there. On the other hand, Nokia is ready. A lot of commercials. Their CEO is very enthusiastic about ReefShark.
It's a big gross margin improvement for them on their own financials, which they need to go on and exchange and change. It's all good questions that we'll see. As we always say, we'll see from quarter-to-quarter how things evolve. It could be very fast, like a hockey stick, it could be linear, it could have few holes in the way.
A separate question on royalties. I think going one click down from unit ASP, you talked about unit growth and royalties. You have developing versus current royalties. Can you please give us the differences there? What is the ASP profile of the guys that are in the pipeline that are going to come into royalties versus today? Should we think of it as the same royalties as today because you have some that are a little bit higher? Or is there a potential for the royalties per unit to go up because of what's coming online?
We've seen that in the past as well. It's the same. Depends on what products kick in faster and to what magnitude. This year, we saw an ASP increase in handsets, the traditional business. Why? Because there were less, not great answer, but we had less low-power LTE devices from some of our key Chinese players. They had a weaker year. We saw some high-end devices that we got into with higher ASP. Every quarter, every year, really depends on what is the key driver. If Ilan's group will do well and we'll have a lot of DJI and drones and 44 billion cameras, you said, in a few years, if we get a small percent, what did we use? 15, I think, here. Computer vision, 15%, we'll be very happy with both the ASP story and the dollars for royalties.
Really, a lot of moving pieces in this puzzle. It's not bad. It's better than to rely on only one product. Every quarter, we'll need to see what comes to the market and who's successful, and how does that drive the ASP for royalties.
Just add that all these new products we bring to market, you see how we're going vertical, bringing more software, bringing profiles and so on. These are all generating incremental licensing and royalty opportunities for us. You're right, if you go into the SAM market, it will be just like DSP if you're going to use that. You add on some software for voice processing, voice recognition, and so on, you're adding on more SAM on top of that. Computer vision and AI, if you're licensing SDKs for SLAM or you're licensing computer vision algorithms, you can charge more. There is a concerted effort within the company to make sure that every deal we sign can maximize the incremental revenue we get from that customer by giving themThere's an answer there.
Like Yaniv says, when it comes to the mix, if we could ship to 10 billion devices and we're getting a $1 condenser or a $1 chip, that's going to have a big influence on the overall data feed. We try not to focus on that blend of A of C, just how the market itself is so divergent dynamically. It could happen with the mix of everything, but certainly, we're targeting as much as we can in the sort of new exciting things the market kind of moves more towards.
Again, in regards to the 2018 roadmap from several years ago, how confident then are you that you can reach these units shipped targets of 20 billion? How confident are you that the data confidence will be able to-
Yeah, it's a good question. I think the first time we tried to build a longer-term story, unit-wise, the non-handset base band we got wrong. We talked about 800 million units. We're not that far out, but we haven't reached that. Bluetooth out of that 800 was 300, as you saw, more or less. It's a nice chunk. When we started that model, it was just in 2014 when we started diversification, we had almost no volume back then in non-handset. If you had the little bit vision and a little bit some of the sound and that we talked about, we are off from 800. The main reason is, I think you heard that in some of the presentations today, especially around the vision or the AI, that we were too early. Too early for the market.
Everybody talked about needing those technologies, the use cases didn't go into production. The technology was ready. Some of our customers even had a chip ready to go. Then the last piece of where do you put that in the product? How do you get it to work? How do the OEM itself say, "Okay, I have computer vision, but what do I do with it?" Ilan talked about so many different use cases on computer vision. Not all the OEMs knew how to deal with it, and that pushed out the royalty scheme, things that we haven't seen in the modem business, because that business we know for 20 years. In the newer businesses, we realized that it wasn't a technology fault, it was just that the market was not mature enough.
Look at how fast drones, for example, how quick. It took a few years. You suddenly have not just camera's improvement and the autofocus and lots of different features in the drones in the last two years that have improved significantly, and suddenly you see the market moving because of that use case. I think that's part of the reason for the last miss of the 800 for 2018. This time, we have more moving parts. It's not an easy type of exercise, but we looked at the customer base, we looked at the market shares, we looked at who dominates today and who we think will be around three or four years from now. This is how we have done the math. Hopefully, maybe we are wrong with the units, but hopefully we're not wrong with dollars. It could be a combination.
It's not a model, it's not a crystal ball. This is just putting few of these moving pieces together and seeing what could come out of it.
You mentioned your three offices in China right now. How has that footprint evolved over time? Or in other words, how much incremental investment in terms of having boots on the ground in that market?
In China?
Yeah.
China, our biggest sales team is in China today, in three different offices, I think shy of 20 people. All the R&D is done either in Israel or in France. France, Bluetooth, Wi-Fi, Israel is the rest of the technologies that we've talked about. We don't have any local R&D presence today in China. It's all done in the regular manner. It's just a big market because of what it is and the investments there.
That [sales of software for people], how many was it three years ago or five years ago?
I used to remember numbers. Sales in China five years ago, how many people? Probably shy of 10. I think we more than doubled it.
Got it.
I would think.
I also just, to clarify my question, the 10%-20% licensing growth, is that on figure or is that absolute?
No, absolute. From the $40, either anywhere between $44 or $48, something like the next step. Same question about the royalties. It's not that we know today that licensing is going to grow by $5 million or $6 million next year. It's just hoping that if we do the right things, then we come up with the right products like we have seen for so many years, we'll manage to also license more to new markets and new customers. That's the basic case of it. You have to make sure that you know what you're talking about. You need to make sure that you know that you think at least that you have the right product roadmap.
So far, I think we have a good history that this did work for us, and we have managed to decide where to go and what markets to tackle, and done it quite successfully, at least on the licensing side, and royalties will follow.
With these kind of $40 million or $45 million of royalty revenue, do you think ranked between your peers and how much the base station customers you're looking at versus something that has a larger bucket?
I think it's a bit maybe premature. I would say that half of the growth, at least, should come from base station. The size of that, as we saw today, the higher ASPs that we have, because these are very expensive type of chips, should contribute about half. This is why we have more confidence in that jump of $40 million, because a big portion of it should come from base station. All the other parts, Bluetooth, Wi-Fi, computer vision, AI, sound, software for sound, that is a bit more difficult. We have the data. It's part of the rest of the $20 million.
Are there one or two within that group that you rank as?
In vision, we hope that vision will kick in. We have the design win, we have the automotive stuff we talked about. We have the sensor business that we should see volumes kicking in on the automotive space. Bluetooth and Wi-Fi. I think we did discuss that this is the fastest-growing market. It grew last year 50% in units, and we don't think that should stop. If we get that right, that's a nice solid add-on over the next couple of years. Gideon, anything else?
Maybe another one on China. Of the $3 billion, what kind of portion of that, if you have any idea, is going to come from China or China-based companies? If you also talk about any trade tension currently, what you're seeing, and if so, if you see it in China, maybe then ability to protect your revenues?
It's hard to say from the $3 billion, because even today, if you ask us what exactly sold in China, we know. What is, for example, UNISOC, Spreadtrum, what portion of their chips or phones are actually sold in China versus India and the rest of the world or emerging economies? They don't give us that information. If we get reports from Intel about Apple sales, we don't know if that Apple iPhone is sold in the U.S., in Mexico, or in China. You don't have really the end product and the way that goes, that's too complicated. We need to get that information from other reports and try to extrapolate. Today, China overall is 50% of our revenues. I would believe that that will continue to be a big portion of our business. China has been doing excellent for us.
I think last quarter, out of 13 deals, nine deals were signed in China. We didn't see any slowdown. Issachar talked about huge investments. They want to do it on their own. We are a great partner to do so. For the time being, we don't have any issues with export control rules, to the best of our knowledge, because the technology is developed, as we mentioned earlier, Israel or France. It's not in the U.S. For now, we could do business, and we have been doing excellent business in China.
[audio distortion] , I mean, macro, how do you see?
Sell the same. If you have ink in your pen and you could sign the licensing agreement, it's all good. No, nothing has changed from a macro environment on the licensing front. Ladies and gentlemen, thank you very much, and Happy New Year. Successful 2019.