Transcript.

Erik: Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.

Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE. And our guest today is Dima Feldman, VP of Product Management and Marketing at Sony Semiconductor. Sony Semiconductor is a leading provider of cellular IoT chipsets for wearables, vehicle telematics, smart utility meters and other devices. And in this talk, we discussed the evolution of the cellular chipsets from 1G through the IoT chipsets, Cat M and NB IoT that are enabling devices today. We also explored the rollout of 5G and the implications of low power, low cost and high reliability connectivity solutions for IoT business model innovation.

If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Thank you. Dima, thank you so much for making time to join us today.

Dima: Hey, Erik, thanks a lot for the invitation.

Erik: So Dima, I'm really looking forward to this conversation because it's on the topic, which is related to a technology I use every day, really unfortunately, probably 16 hours a day. But that is kind of embedded in all the solutions I'm using, so I never really touch it. I never really understand it, like I do some other technologies. But before we get into Sony semiconductor and the work that you're doing there, I want to just understand your background a bit. Actually, you've been working at Altair before Sony purchased this since 2007. Can you just walk us through a bit your path to join in Sony and then some of the work that you've been doing at Altair over the past, what is it 13 years now?

Dima: Yeah, of course. I've actually never thought I was going to work for so long for one company. And actually we are based in Israel. Altair semiconductor was based in Israel as a headquarters and obviously Sony as well. Like many other Israelis, you go to the university, some do army after that, Altair was pretty much my first job after completing this Israeli path, I would say, of university and a few years in army.

And I've started Altair as the RFA or RFIC lead, we started developing the new RFIC. The company actually started as Ymax those days. We had a great digital technology, good software engineers. We pretty quickly realized that if you want to be successful, you have to own the entire solution and all the key components of the solution. And the first task for us was to bring on RFIC technology. And we started without source, we build a team. And I was really in charge of that and actually go with the company as the company grow. Later, I've spent a few years in Canada for relocation with my family working a lot with North American customers.

Coming back to Israel, I've established the solution architecture team and later the product team and recently, I've been appointed to VP Product Management and Marketing at the company. So pretty diverse route within the company and I think this is one of the reasons people like working for this company, because you can start at one position and really do different things, both vertically and also be promoted, so different management ranks.

And as a company may be, which is probably more interesting, so we started with Ymax. 2008, we have realized that Ymax is not going to happen or not in a massive scale. And we have shifted to LT, our first solution power all kinds of data devices like tablets and portable routers. However, in 2016 or slightly prior to acquisition by Sony, we have realized that the market is probably much bigger and really looking on the thing that communicate to other people on one hand and smartphone businesses very crowded and basically dominated by huge players who are very focused on that and spend teams of multiple thousands of people.

We made the decision, we focus on IoT. And this is when Sony acquired us in 2016. Within a Sony, we’re mostly part of the IT business solution unit, so we do the cellular IoT, the other groups are doing communication one type called [inaudible 05:31], which is a homegrown in Japan. And this is mostly happening type of communication where a device sends a few messages a day, and Sony provides a full service.

The other team is the Genesis team, which is very successful and well recognized for extremely low power consumption devices. And they power probably most of the wearable market today to the environment and other companies. So as a team, we work really on communication technology, the location technology, the results of some processing capabilities within the group. Those capabilities, both across the company, and also at least our business unit or cellular it is led from Israel.

As a business unit, we can utilize the technologies within Sony and being fully independent with a huge backup of Sony and, of course, the market. And maybe the other face we do in Israel and this is less known because it's part of Sony, we also develop AI calculation engine for Sony semiconductor. And you probably know that Sony semiconductor and solution is one of the biggest sensor providers globally. We do AI computation there.

Erik: And you've mentioned developing IoT chipsets and kind of segment in that from the mobile market. I'm really interested in understanding more in detail how you segment the IoT market. Because what I find is every company has their own unique perspective on what the market looks like, and because it's such a fragmented market, and it can be defined by industry or by asset type or by requirement like low power, etc, there's no kind of standard definition for what the market actually looks like. How does Sony segment the IoT market from a chipset perspective?

Dima: So I think, probably the formal definition would be where the things communicate and personal or people are not involved, or mostly not involved, so this probably would be their fastest way to distinguish. And for me, things could be literally anything. Give me an object that now find a way why should communicate with the cloud. So, it could be a speaker phone, it could be mic, could be my mouse, or I now use a speaker for this conversation. And it would be nice if my team manager will know where speaker is because it's pretty expensive. Or I'll get a notification if it's battery's about to die in the morning. So there are so many use cases for devices to communicate.

Saying that there would be several segmentation within the market, I'm not sure it's a formal or not. This is from technology perspective. One could be high bandwidth applications, like cameras, camera is a thing and it transmits video, so it's definitely within the market. But from a technology perspective, it's, of course, totally different from a temperature sensor I might have or electricity meter I have outside my door. And once that you report electricity uses, so it's different use cases, but it's all things.

So I would say there are devices that are high throughput, low throughput. I can say there are devices which are very time critical, like V2X. My vehicle communicates to the next vehicle or to the infrastructure around it. So it's very low latency, high availability, communication versus again, low latency or we call it usually on 3GPP we call it massive IoT is the huge number of devices with a small amount of communication.

I will say also, the other way to classify it and say some devices are connected globally whenever you go and they use a cellular connectivity or wherever they are located, meters never roam, but they use cellular connectivity. While some other devices, especially within our houses, you often use WiFi or Bluetooth connectivity. So I would say indoor devices, which would be typically consumer electronic devices versus B2C or B2B devices, which typically would use a cellular network because of their higher quality of service. So, very different ways to describe it. Does that make any sense to you?

Erik: Yeah, that does make sense. So do you describe it based on the functional requirements of different types of solutions? And does that map to your portfolio? On the one hand, as you mentioned, you've got probably thousands really different categories of devices that could be connected. And each one has unique requirements, but then you can't build 1,000 different products for each many categories. So what does your portfolio look like? Is it that you have five different chipsets for different situations? I guess you do still do a fair amount of customization for your larger customers. But what does this look like from a portfolio perspective?

Dima: I think I'll start with the end. I think IoT market, unlike smartphone market which is maybe cellular as well, and I often think about cellular as the Holy Grail, but there's so many other technologies as well. So IoT is very different from a cellular smartphones, because they are so different. And this is where I can easily say we don't play in a smartphone market and we definitely have this low power wireless array IoT. In terms of portfolio, we have those small sensors, extremely low bar with a battery life of 10 or 20 years in some cases. We have devices with a little bit higher throughput, for example, for cameras upload. But I think this is as a Sony semiconductor Israel, but they all focused on connecting things, not people.

Erik: And then where would you sit in the supply chain? So I mean, obviously, Sony, the larger corporate, they sit in many locations in the supply chain. But your business unit, specifically, are you selling directly to OEMs? Are you being integrated into larger modules? Where would you sit? Or who would be your direct customers?

Dima: I think IoT and you touched a little bit customization for the larger customers. And I think interesting thing about IoT is that the market is so diverse that you don't have many huge customers. It's not like you have Apple and Samsung who are 20% of the market share. There are probably thousands of thousands of device makers.

And therefore, this would lead me to two conclusions. One is you are hardly customizing the solution for your customers because this way you are not scalable to serve most of the market. This also brings us to your question is how do we arrest the market. And if you go and look on the traditional cellular market, and I'm looking this more from the cellular perspective than IoT, perhaps if you look on the smartphone, people usually provide chipset to the smartphone builders, and we call the chip on board, and it can only work for very large volume devices.

Because the overall infrastructure or the ecosystem of the cellular business is quite complicated with the amount of certification and regulatory body we have to work with and therefore does not necessarily make sense for smaller volumes. And so for really big devices we will go directly will be maybe I would say as an example, big wearable guy or very big smart meter vendor. For them, it could in some cases could make sense to chip on board, while for others we’ll go through channels, we call them our module partners.

And it's interesting because each one of them has a different specialization and different support body and sales organization and some could go for really small customers doing really devices in magnitude of hundreds, or thousands of units. If you can think of a company who want to do smart garbage collection for your municipality, it's just not always a huge companies. And we will address those companies through one model maker, which often will bundle our support with maybe connectivity services or even cloud services to enable really quick time to market. While for other markets we’ll typically work with, actually, all of them are very high quality module makers who can address the bigger markets or maybe being a little bit more cost optimized. So I think there is a really big, big range of the ways we can address the market starting with direct customer support for strategic accounts. And going through one or in some cases, even two module makers they’re like staggered to address really small customers.

Erik: Is that the market of modular makers? Is it also quite fragmented? Or are there a handful of companies there that are composing 40-50% of the total market size?

Dima: I would say about 10 companies who compose 80% of the market. It's a traditional market. Maybe five or six years ago, it was called machine-to-machine communication, and this is where most of the module makers came, because at that time if look historically, maybe there was machine-to-machine market or was too small to be addressed by chipset companies. And what happened is module makers will take chipset design for a smartphone and modified or stripped down or do some software customization to serve the machine-to-machine market.

And there’re a few very established module majors in this space like CR wireless is one of the biggest. And they're really experts in this market. And they have channels and they understand the actual challenges of their customers and device manufacturers. Or they sometimes even come to the service manufacturer, somebody wants to do smart mousetrap and they will help him to build the mousetrap because guys know mousetraps, he doesn't know cellular. So those are the more traditional and provide a lot of services.

Looking among our channels, we also have more IoT guys which doing more like I would say hardware-oriented modules, companies like Zericom, Moraco, Whinstone, global market we also have few more in Japan which specialize on the Japanese market would specialize on global market and some would really specialize on the local markets.

What modules for me means and maybe this is why they're valuing the ecosystem is so important. When we work with the module maker, he will make the design, he will go and certify with regulatory agencies like FCC and [inaudible 18:16] in Europe or Pelican in Japan, and he will also go and achieve a carrier approvals like from AT&T, Verizon, Vodafone, so many actually, I had to say a name because I will definitely miss somebody. But all the big carriers have certifications.

By doing so, when I am a small company doing mousetraps, I don't have to approach all certification bodies and I just use the module plug it into my bulk, some cases even bring connectivity with the module. And I think for me, it's maybe one thing about the IT business is when I'm in a discussion with device maker, or maybe end customer, he will typically know his business quite well.

So it could be mousetraps, it could be smart meters, gas meter, water meters or logistic tracker. They're really experts in what value they bring to their customers, what are the relevant use cases. And what are the challenges they have? And if case of materials, what are the materials that water meter has to be built from so it can withstand for 20 years in very humid environment? [inaudible 19:40]. And therefore it is important for me and for module maker partners is to offload all the difficulties of the cellular ecosystem for those guys and really let them focus on their core business and let us and the module makers solve the challenges of the cellular networks.

Erik: And then companies like Arms or design companies, and then the cellular networks themselves, I suppose you'd also be in close contact with them. How does this work with arm? I mean, are they a competitor to any extent or they’re purely a partner, or other similar companies?

Dima: Well, Arm is such a diversified company. They do processors. They do some silicon IP. And they now do some cloud services. And we also touch with them actually in a very close cooperation for a technology we call iSIM, Integrated Sim. This is another piece which build to simplify the overall complexity of the solution.

If you think of Sim technology, actually, it was introduced in late 80s, back in 90s when we started to use the phones, even before the smartphones, we would use the sim card to store our contact list, the text messages we send because it was so difficult to type the name, and our credential, my phone number. So if I would swap a phone, I would take a sim card from one phone and plug it into another and my phone number will go with this action and also my contacts, which was a great deal at the time.

And SIM involves in that a few more functions it can do. But essentially, it's there to serve the way you identify yourself versus the network in a secure way. And even today, if I buy one smartphone and more I buy next one, I move the sim from one to another, and I keep my subscription and I use now cloud services to back up my contacts.

But if you think of an IoT device, it actually has a phone number. You will be surprised, but this is how devices work on the network, but nobody is calling my smart meter anymore or sending text messages. Also, it doesn't need to have a contact list because he's not going to call, text anybody. It's all IP on data communication those days. But for the cellular communication, you need to have the phone numbers and identity and the security credentials.

And we are no more flipping those between meters of different IoT devices. So therefore, there is absolutely no need anymore to move a physical card or sim card from one IoT device to another IoT device. Actually, I'm really happy to be absolutely the first one to commercialize it that say, okay, the physical sim card is not needed, the soldats sim card, which is another version is no more needed. We can have a secure sim credentials on the model chip.

And this is where ARM with their ecosystem also come into the game and we provide the hardware. In this specific partnership, they can provide a software or sim credential working with the carriers. And together, we have a smart piece within our chip, which is now your plastic sim card. Actually, the chip itself is half of the size of the smallest sim card anyhow, so we really don't need those sim cards anymore. And this is where our partnership with ARM comes into the game.

Erik: Yeah, I suppose the sim card is so large, just because our fingers have to be able to pick it up, so a lot of that space is just to handle basically.

Dima: Yeah, but you definitely don't need it in the IoT. One thing is maybe the functional, or the logical piece of the sim card which we embedded. But the other one, many of them has to have very strong ingress protection, because they live in a humid environment. Nobody wants people to fly with them. And therefore leaving any physical port for a sim card, it could jeopardize device functionality in field and also create a security risk.

I can have electricity meter, I just pull out the sim card and does not report the meter readings anymore. So it's really, the use case has changed and I think the industry has to change with it. To be honest, we are not doing it only without, we have additional partners in the specifics space because it's all about the ecosystem. I think specifically for IoT, the overall solution is still, I would say, a little bit complicated because there are many parties involved, and many standards. And if you build the right ecosystem around you because maybe some partners want to prefer one specific vendor over the other, and therefore, you typically have to build a pretty rich ecosystem around you, so when you go to the customer, you can always choose the right one players from your ecosystem.

Erik: So let's move now into a bit of a deeper conversation around the technology. So cellular is a set of technologies that I think many people have a very high level understanding of, and few have a deep understanding. So maybe you can give us a 101 course on what are the different cellular technologies, LTE, CAT-M, NB-IoT, we have 5G coming around the corner. What would be the key cellular technologies today or you're expecting to be relevant in the near future? And then how do they differ from each other in terms of the functionality that they provide?

Dima: But let's first define, maybe for the audience what cellular technology is. And it comes from what cellular is and cellular means that we have many cells within our city or surrounding. And our devices communicate with cells. And so maybe different way from my WiFi which will communicate typically with the only access point I have at home. In a cellular technology, the entire idea is that operator which is typically a big company who buys spectrum, which is a very expensive asset could be billions of dollars, they buy spectrum, and they put many cell towers that will use that spectrum to communicate with the devices.

So this is, in essence, what cellular network means versus maybe WiFi, which is short range networks in our homes, or Bluetooth, where we typically have a device-to-device communication and each device can talk with each device. In cellular network, the idea is that my device talk with the cellular tower which is going to talk with the cellular network, and the other way around.

And between the technology, though there was long time ago it's all started analog, then 2G which is started at 80s and they still exist and deployed and slowing down only now. But it's 2G, it started for a voice conversation. And maybe it's important to say that cellular technology started to replace the landline. So, started with a voice conversation, text messages was actually a next service on top of that, and very naturally the people understood or we realized that we don't want to have only voice, now we want to also transmit data and videos.

And today, invoice is actually very small portion, the entire traffic on the network. And most traffic is our rich content like videos and photos and music. So you can see that there was a shift from the voice technology to the voice payload to the data payload, and networks had to be changed accordingly. And this is where 3G can forestall just pay networks to replace the 2G networks because they could provide data traffic in a better way.

Then next level and maybe one bit technology. But when network was built to bring voice from one site to another, those networks were not based on IP technology, it was based on a packet switching technology where we basically got a specific payload or specific time where it could transmit your data in advanced protocols like IP which are used today everywhere.

And the first IP only the network was LTE network, which deployment started in 2008. What we call the LTE, some people call it generation 4 or 4+. And LTE networks started with category three, which the base side was 100 megabit per second, downlink 50 megabits per second, uplink, which is pretty decent, even today, it's pretty nice and then started about 10 years ago. From that direction, cellular communication globally adopted the LTE technology and this was a big change because almost no more original specification like different one in China or US pretty much aligned globally, which was an LP and this was a very big change.

And if you read the specification of the smartphones, you'll often find Category 4, 6, 12, 16, it's all means higher throughput. And the higher category you have, the more throughput you have. So this will happen to the LTE network. At some point, this is where it was actually enlightened for us. We said, well, you don't have to run after the throughput because most of the IoT devices don't need that. Also, [inaudible 30:25] realize that and instead of doing higher categories, like 16, or 20, etc, there was a state.

Some devices going to achieve higher categories and higher throughput actually by using more and more spectrum, while other devices say, well, we need to slow it down, we need to generate really low cost and extremely power efficient devices to meet the IoT market needs. And this is where CAT and technology came to the game really starting probably 2005 or 2004, and only now it's starting, it's a massive ramp up.

So this is why CAT-M was invented. CAT-M was invented roughly the same time. More a political, I would say, CAT-M was driven by 3GPP and US based company, Chinese companies wanted to have again their own proprietary technology. And why we went through the acquisition of startup in UK, and this is how NB came into the world. I think so it was initially totally competing technology serving very similar needs.

However, I think one of the good thing that happened is that it was also adopted by certification by the core 3GPP which is the certification body for cellular. So one, we have two slightly competing technology within the same standard, maybe it's not optimal, but the good thing that we have all the competent technology under the same body. And once we have a rich ecosystem, we can build and scale up. And I'm really happy that all those standards are under the same organization body. And this also means that same cellular network and the same cellular towers support all the categories.

Actually, CAT-M and NB is also part of the 5G. But do you have more time to talk about it, by the way?

Erik: Yeah, absolutely.

Dima: So 5G known for one thing, but it's actually totally different pillars. 5G is a kind of a marketing name for a new generation of technology and it came to me three different pillars or three different target application. This is what most of the people call 5G, it's called enhanced mobile broadband. This is where our devices transmit more and more data and receive more and more data. I would say it's native or logical continuation of the higher categories. So we need to have more data because all of us use much more data, gives us terrible amount of data as well those days.

And also, this is the enhanced broadband. So it's with a new radio on the same cellular frequencies, but it also brought additional spectrum available because allows us physics still in place and we need more spectrum to transmit more data. And therefore 5G uses much more spectrum, also its type [inaudible 33:41] space and even in millimeter wave. So, this technology, cellular technology is expanded to into more spectrum.

And this is what most people call 5G. But the other two pillars are one is called ultra-reliable communication or ultra-low latency communication. And this is a different segment. It's intended for V2X, vehicle-to-infrastructure or vehicle-to-vehicle communication, where you don't send videos with the vehicles, but you want to send information in an extremely reliable way and it's extremely diagnostic. And same goes for the industrial IoT where you can have any factory one machine with the camera and another one with a knife. And you want the camera to detect that the piece of shit came and you need to cut it then you have to give a command work quickly to the to the knife, maybe doing a bit of simplification, but this is what you need for the industrial IoT were very quick communication.

Erik: So you've just introduced two of the pillars of 5G, is it that these pillars can be thought of as functionalities in any 5G network or any 5G enabled device would have all three?

Dima: No.

Erik: 5G is a larger set of technology device by device, they might have one or more of these?

Dima: Yes, you could imagine that while for the mobile broadband you will communicate with a tower, a tower with a network and it comes with certain latency. When we have a V2X communication, then my vehicle wants to communicate to your vehicle. It doesn't have to go all the way to the tower and the core network and back because it will either delayed.

And the third pillar is actually more similar or uses exactly the same infrastructure. The first one and the third pillar I was waiting for is the massive IoT pillar where the target for the 5G was our cellular network has to support over a million devices per square kilometer. And this is the third pillar. So, for us in the IoT, this is actually I would say 90 or 95% of the things we do is covered by this pillar. And the interesting thing that CAT-M and NB you mentioned before are actually meet the requirements.

So whatever we do today is already 5G massive IoT. So the more devices you bring, and the more core network capacity you provide, the devices stays the same. And this is, actually, a great deal and helps us to scale up today. The CAT-M technology and NB technology is the 5G IoT technology.

Erik: So I have a question about the potential impact of 5G on IoT deployments. Well, we've been speaking with different people and have gotten a number of different opinions as well. The question is what will be the impact on edge computing in particular? So if we look at 5G technology, or basically, one group of people says that when 5G is deployed, it's going to allow devices, more bandwidth, higher latency, etc. It's going to basically make it easier to do edge computing because you'll have this very robust network that connects maybe devices around the factory or within the city, and can then enable more compute on the edge.

And then another group of people are saying, actually, it's making edge computing less necessary because these devices can then be connected to the cloud with very high bandwidth, low latency solutions. So it's actually not necessary to do edge computing. You can just move everything to the cloud, obviously, there's costs involved there.

Is 5G going to be supporting edge computing or cloud computing more or less? Or do you think that's just maybe not the right way to think about it, it's just more on a case by case basis?

Dima: Yeah, I don't want to be categorical, I think it's really case by case. Even today, we have devices with extremely small bandwidth that can do a lot of edge computing. For example, let's think of camera [inaudible 38:17], maybe in colonial days it's really important. Can count how many people came into the shopping mall. So we can have a camera with a lot of processing on the edge that can count people and it will only transmit the number how many people are in the mall.

So we do an edge computing here. Theoretically, we could dream up all this video to the cloud with a 5G and do everything on the cloud, but then maybe the battery power consumption of edge device would be much higher because it has to transmit all the data and maybe the overall system is less efficient. So, this actually, while the second camp may be right, but I can also think in another way that you have an edge computer which receives a lot of data from the edge itself but it can also augment it from data from other edges or from the cloud.

And also looking in perspective into the computing market, the more bandwidth we have, the more flexibility we have to spread the computing load. And we can say okay, if I can transmit everything to everywhere, I could actually put the load in the right place, and far from cost optimizing performance optimize the solution. While in my first example that I gave you, if I have only a very narrow pipe and I cannot upload the video, then I am somehow limited with what I can do on the edge

So for example, I have two cameras, and I want to augment it into one or one data processing unit and I certainly can't because I don't have enough bandwidth between the two. I think your assessment of case by case is pretty much correct.

Erik: So if we look at the big categories today, so I suppose we can list a few of these, we might be looking at things like smart sensors, cameras, etc, that you've mentioned. We have CAT then, we have NB-IoT, we have 5G, we also have non cellular solutions, so LPWAN, etc. What trends are you seeing in terms of adoption of technology by the device manufacturer?

I don't know, if you want to compare a CAT-M versus NB-IoT, or maybe we can bucket those together? But there's also this open question around when we'll actually start seeing devices ready for 5G? I think mobile phones, it's a little bit easier, because there's very large companies that are pushing that forward, devices is much more fragmented. How do you see the adoption trends developing among the ecosystem right now?

Dima: I'm a little bit biased because we do cellular. I can make a very strong case why I think cellular is better than other technologies, which are LPWAN as well, you all themselves LPWAN. And I would say for maybe three main reasons. And not saying that there is adoption of only one side, but I see benefits. And I think one benefit is cellular exist everywhere. So if I go to my house, or if a utility company wants to install a smart water meter next to my house, they know that cellular networks is there. If they are about to deploy their private network, or some kind of LPWAN technology, quality of coverage is not as good.

The other advantage of cellular is the spectrum is licensed and this means that somebody paid a lot of money to use it. This is unlike unlicensed spectrum where some technology can use it, but I can bring many other devices or totally unrelated devices transmitting on same frequency and suddenly the quality of service will be highly degraded till not available. I would say the third point is security. And as we deploy devices that have to live in the field for, we have no customer who wants to play for 20 years. So it's amazing, the devices has to work till 2040, and they have to be secured because this is a critical infrastructure.

I think many think that working with a technology which is more mature, and the standards are being created by multiple people, and I would say maybe many thousands of many years has been invested into right them in made of standards truly secure and interoperable, this brings some people to think that cellular technologies much safer to use in terms of cyberattack or denial of service attack.

And this is why I think cellular technology is a better fit for most, not all actually. I do see some use cases like I would say agricultural farms, where having a local network could make a lot of sense. And as I said before, CAT-M and NB is already 5G, but you can easily get the guarantee.

Erik: If I have an NB-IoT chip installed in a device today, does that mean that when a 5G network comes on I'll be able to use that? Or is it that maybe I need a newer generation of NB-IoT, or if I have something that was maybe deployed two years ago?

Dima: It's super important point. Because people who deploy now devices that will stay in field for 20 years have to guarantee that they will work for 20 years. And I think it would be fair to assume that in 10 years or 15 years from now, the LTE networks that we know from smartphones will probably shut down most of them and we are going to have 5G. And therefore, it is critical and this is super important and this is how standouts are built that both CAT-M and NB will work with 5G base stations.

Erik: I was thinking that each device as it goes through product development would have to be modified to enable 5G and it sounds like that's already the case for any CAT-M or NB-IoT?

Dima: There is some testings I don't want to say whatever works out no issues 5 years or 10 years from now, maybe there will be some firmware upgrades to fix some bugs. But the way the standard is built, it's fully interoperable. And those are testing we do now in the background with the infrastructure vendors to make sure that this is actually the case.

Erik: Is that a trend of shifting the functionality away from hardware towards software so that it's more flexible over time? Or how do you do this?

Dima: I would say yes. Well, for chipset, a certain thing you have to do in hardware. But software gives you more flexibility to develop and adjust and optimize. I know that maybe it's not my core expertise. But I do know that for the network infrastructure, there was a significant movement from hardware base stations to more and more software, and cloud based solutions also for the infrastructure. So yes, definitely, there's a significant shift towards software.

Erik: So Dima, we've talked about the status quo, and then also about 5G, anything else on your radar that you're really excited about in the cellular world?

Dima: I think what I'm excited of is I would say in the last 12 months, we see a big role of cellular IoT. By itself, it's really exciting to see that something you work on and dream on is actually happening and ships in volume. One thing that we started to see is more and more use cases coming into the game. First, it take to mature the use case and understand that this is what you want from a business perspective and then you need all the technology pieces. If in the past for logistics space, you had to track a container or maybe expensive pallet, silently, you can track a parcel or a bag of seeds. So I think one thing that I really see happening is more diverse kinds of application and use cases.

Erik: Then from a marketing perspective, is there any type of 80/20 rule where 20% of the use cases are providing 80% of your volume? Or is the market so fragmented that you're just responding? Are you able to kind of target it and say, these are the X number of categories that we're going to prioritize for the next year?

Dima: I would say one steady market, which is really, really steady is a smart metering market. It's been determined by a number of households and ways of replacement. So it's really steady and pretty big. I would say it take significant portion of our volume. I see logistics market growing I would say exponentially. So I don't know when it will be bigger than the other markets.

Erik: And the connected vehicle market, is this something that you would serve, or is this more the traditional kind of high performance cellular type chipset that would serve autonomous vehicle?

Dima: I would say for vehicles, there are many, many different needs. Definitely one of the needs, which is V2X, this was the second pillar of the 5G and somebody else to serve it. But there are other segments within vehicles that are related to tracking the vehicles, monitoring the engine, the battery, doing all kinds of preventive maintenance.

And the different trends, once there was an idea that we can unify all into a single radio, but now we see a trend where actually different manufacturers say well, I don't want to rely on anybody else, I want to have my own car telematics. So you could think of a vendor who will say I will monitor the engine with one radio and I'll do the multimedia with other radio and the smart driving autonomous driving with a third radio, totally different. So, vehicles are extremely complicated, and super fragmented. So I would say we will probably cover only a portion of that vertical. Still in every vehicle, it's a big volume, but you cannot do all.

Erik: Of course. It's funny you mentioned that probably every other week or so I have a conversation with a component manufacturer who's trying to figure out we know our components are very important but they produce a lot of high value data, but we're only components. So how can we access that data? How can we build a business model around that? But yeah, you can see that there's a strong incentive to have control of the data that your components are producing.

Dima: Exactly. And maybe one thing to mention that really enables the latest development is we spoke slightly about cost when I said that one of the reasons for Cat-M and NB is to reduce cost. I also see dramatical cost reduction in IoT related data plans. Some of the public data says, for specific plans, like $10 for 10 years, and you can imagine that in private conversation, numbers are even better.

So if I only need to collect a small amount of data from the vehicle, the connectivity base, which is once was tens of dollars for months, now could become a cent. And I can imagine a device which can monitor your vehicle behavior which could cost between $10-20 including the connectivity. You mentioned earlier components generate value, but it takes money to move this data from one point to another. The lower the cost of moving the data from the edge to the cloud, more and more people will find creative ways what they can do with their data. And this is why it's moving. We see vehicles and extremely strong in the logistic industry.

Erik: This is very, very interesting market Development. The business search for business models is really driving innovation on the cost structure. I think it's very hard to build a business model for a low cost device if the connectivity is going to…

Dima: I think, if you're not talking as end customers, but for B2B and this is where most of the IoT devices are going to today. I think a good business model is extremely important to actually justify the investment and build an end-to-end solution. Because even today, we actually would like to have much quicker design and deployment time, and it still takes probably 12-18 months, and we actually do a lot of doing that. But it still takes some time, you come with an idea, you can deploy significant number of devices. And you need to justify that to your management and have a funds invested in that activity you have. You really have to come with a very good business case.

Erik: So then maybe last question from my side would be, especially with maybe your larger customers, where you're working with them but more closely, how involved you get in these discussions of the business case or helping them to understand? Beyond the core technical functionality, how deep do you get involved in these conversations?

Dima: I think this is probably one of the most exciting parts of my job that I get to meet many of the end customers are not only the direct customers. And I think one of the reasons we really want to be so involved, of course, so we’re all human and attracted to interesting things, but we also open the eyes for some of the innovation manager of those companies about what he can do with the cellular technology. The most advanced ones know everything and they know the business case, they know the technology, they can teach me many things. And this typically happens when the company doing their second or third project.

But when they do their first project, you would imagine that their innovation manager within the organization will be highly skilled expert within his area of interest. For example, within agriculture company, they will know everything about the seeds, and what problem their customers have, and when they use them properly and how to store seeds properly or improperly. But this is where we come to the game and we help them together to shape the product they need. And for a small amount of really strategic relationship we are exposed and we can build those great solutions together.

Erik: Well, Dima, I really learned a lot here. As I said, this is a critical technology but because it's a bit hidden under the hood, it's something that me and probably many other people in this area know too little about. How can people reach out to you or your team or to Sony Semiconductor in general to learn more about what you're doing and also if there's an opportunity to collaborate?

Dima: Well, there are so many ways. One is social media, LinkedIn, either for myself, which is Dima Feldman at LinkedIn, or today we are still Sony Semiconductor which we are in position to change the name. So this is one of the easiest ways. Second, of course, website Altair_semi.com. And actually, I would really appreciate if we can publish those links with the podcast so people would have a good way to reach us.

Erik: Yeah, very happy to I'll include those in the show notes. Well, Dima again, thank you. Appreciate your time today.

Dima: Okay, great. Erik, it was very interesting and great questions. Thanks.

Erik: Thanks for tuning in to another edition of the industrial IoT spotlight. Don't forget to follow us on Twitter at IotoneHQ, and to check out our database of case studies on IoTONE.com. If you have unique insight or a project deployment story to share, we'd love to feature you on a future edition. Write us at erik.walenza@IoTone.com.