Exploring Future Technology Trends - A Look Ahead with Avnet Silica

What's on your plate?

Thomas: Yeah, if we consider today's situation, I will split my answer into two directions. The first one is from a technology perspective and the second one, I would like to talk a little bit about vertical markets. So, let's jump into the first one needless to say that AI is a big driver at these days and we're going to see it a little bit more today during our discussion when we talk about the individual applications.

So needless to say, that AI is an accelerator in our industry these days, but on the other side, it's also an enabler, enabler to make things easier. But despite this big trend, which we see right now, there are some others like from a technology perspective, which is high power application. Let's touch the thing about gallium nitride.

So, let's incorporate where we see quite a lot of movement today. This is important area where we have especially electrification and energy management or smart factories participating quite a lot in, but I don't want to leave this room for technologies without any wireless discussions, because let's be honest, going back many years ago, IoT played a significant role.

And, this is still a very crucial element, especially when we talk about smart city environments and to surround this brings me right to the topic about verticals. Let's be honest, considering smart city applications is and will remain an important topic for our industry because we see quite a lot of trends where we get connected devices and all the dashboards, which are providing us information, which way we can do things even better and more efficient.

But although smart industry and need us to stay automotive, maybe not today because we see quite a market dynamic, which is still post COVID a challenging one, but on the other side, we will see over the next year's double-digit growth in this area, particularly driven out of semiconductor content.

Ruth: Impressive. Michael, do you, do you align with these trends?

Michael: If we take a look to AI, it's a big topic today, especially if it's related to generative AI, which I think we will cover also more in detail. But Thomas mentioned also about a smart industry where we have like autonomous robotics. What we see also as a trend is that humanoid robots will be put into manufactories.

And this is something that We will see more and more the upcoming years, for example, like a Tesla with their Optimus humanoid robot. There is a company figure AI. They started in 2022. They raised more than 750 million in the meantime; they are valued at 2. 6 billion. And what is quite interesting is that, so this is a completely new field where they put a humanoid robot into the production sites.

For example, they work with BMW as a test case. It's also in California. This is something that, yeah, we will see more and more that this kind of robotics will be implemented. So, besides the standard robotics applications for the industry.

Ruth: Interesting. Yeah. Circling back to the, the semiconductors and the market potential.

And Thomas, you mentioned that there's still basically a little slump after the pandemic years. How's the total market now evolving and how does this then also impact technology development?

Thomas: If we go back many years ago, and I would like to invite you for a journey going back 37 years ago. We have seen a trend in the semiconductor industry of a growth rate of 8.

7 percent component annual growth per year, which means every year 8. 7, obviously not every year in the same dimension. We have seen up and downs like a rollercoaster, but over the years been 8. 7. If we consider today, the trends, which we see from a semiconductor perspective, it will continue and even it will grow.

And we're talking here about the billions, which are already today. And by 2030, you're going to be even 1 trillion. We're going to see the market reaching a level famous people, like for instance, Elon Musk mentioned quite a while ago that the chip rush is even bigger than a gold rush. Needless to say, that we see from a geopolitical situation, although dependency on some of the territories, I just want to emphasize Taiwan.

As that country was the biggest production capacity worldwide, where we talk about 66 percent of all semiconductors, which are finally then produced or turned in their territory. But we have just touched the topic about markets, and you say, okay, Thomas, how do you see it from a perspective of growing?

Let's take a look now into some of the points Michael mentioned and myself before when we consider today, for instance, automotive. Taking into account that a semiconductor content of a midsize car with ADAS level two and combustion engine has $400 semiconductor content and the same car equipped with level three plus, which is one level above.

Plus, uh, electrification is 1200. Gives us already a clear indication about factor three of semiconductors in a car. It doesn't mean the market will tribble in, in a certain period of time, but it gives us a clear indication that the semiconductors equipped in car getting by far more. And we talk here about applications like ADAS I just mentioned, which is a radar lighter sensor or sensor fusion, but also about topics which are display oriented.

Once the car gets more and more autonomous and you can integrate more display and people are using a car in a similar way, like airplanes, which gives us a clear indication of where this comes from. Plus, and this all plays a significant role in conjunction to what Michael says before, when we talk about V2X.

Which is vehicle to everything, and in this particular case, you can use AI, and you can use large language models as an interface where you're going to see by far more communication in the future where data is required plus and also semiconductors.

Ruth: And how do you see regional markets such as Europe positioning themselves competitively?

You mentioned, yeah, the major role that Taiwan is playing.

Thomas: We consider this from an end market perspective, it obviously varies here from a European perspective. We are less consumer oriented in regards of production and regards of development as we have maybe in Asia. However, when we consider as big markets like smart city and smart industry, we are playing indeed a very big role globally and a very big role in the European territory.

And I would like to underline this with some examples. When we talk today about smart manufacturing, and we have just touched before with Michael. The topic around robots, cobots, predictive maintenance plays also a significant role here. We have a lot of momentum and still talking about the double-digit growth rate over the next year.

Once we have rebound, which we all expect by next year from the market perspective. But independently on that: A lot of the devices are meanwhile equipped with video analytics with sensors where you get additional data, which we haven't had in the past. And this gives us a far more opportunity than to optimize the things and even to make it better over the time in the same direction.

And let's take now a very easy example, which we all know it's about EV charging. In this particular field, we have wall boxes, we have AC to DC charger, we have DC to DC charger. So, you see already quite a variety. Where high power, what I mentioned before, plays a significant role. And in this case, it's not only about the cars, it's not only about the wall boxes and the charging stations.

It's also about the smart grid, which is behind. And smart grid plays indeed a significant role, especially when we talk about applications where people would finally charge their cars all at the same place at the same time, for instance, arriving at work. And it gives us a good indication that it would not work with smart grids, with grids which are.

centrally coordinated in a, in a way that people would be behind and then to adjust this because let's be honest, if everybody would arrive at the same time, the system would collapse. In this particular case, we talk, for instance, about bidirectional charging capability, which means you charge your car, you de-charge your car.

And these kinds of trends, we see quite a lot in Europe and will definitely come even more in the future.

Ruth: You already mentioned smart manufacturing and how that is transforming production lines worldwide. Can you share how companies are adopting these technologies? Do you have some use cases or examples?

Thomas: We can definitely, and we work with quite a lot of customers around that. But before we jump into some use cases, which we are facing a little bit explanation what is behind, because in the past we have spoken about industry automation. Meanwhile, it is factory automation or smart manufacturing, similar words which are used in this case.

If we go today into our warehouse or into an, uh, production facility, um, you, you have, for instance, industrial robots, like we all know from the past, like a KUKA robot. It's a very pragmatic example from a, from a central European perspective. But meanwhile, if you take a look into such kind of facilities, you find also smart robots.

Where you have collaborative robots, you have mobile robots, which are moving around from one place to another and trying to find the best way how you can move in that kind of facility. And if we go back many years ago, and to the point where you just asked, we had technologies like connectivity, where we have received in the past information, hey, is it running well, or is it not running well?

And do we need to do anything in a different way? Meanwhile, we have developed over the time that kind of system in that way that you have, for instance, predictive maintenance within systems tells you when, for instance, something would need to be adjusted or to change because of a certain cycle, which was planned or even unplanned.

But the same, for instance, in anomaly detection where we use for instance, video and image analytics. And this is quite an important point because for such kind of task, you get a quality control and you can, for instance, find detections quite easy. And this all comes into one element at the end of the day to become more.

Efficient to become more independent and, to run such kind of applications with the support of robots and cobots. And I would like to underline this with one example, Michael just mentioned before. And this was about figure AI and BMW. If we go back to such kind of facility in the past, we had people who, for instance, been at the production line to put in some equipment.

Finally, it's been, mounted, it been worked out to a different component and the person took it out again. Today, if we take such kind of robots into consideration, there are at the machine, you have a real time communication, you can even tell the robot what to do. And at the end of the day, this is.

It's done by this kind of machine. It will surely take a certain period of time until this will become a standard. However, seeing the trend, which we have seen on AI, on factory automation, and then the integration of both into one gives us already a clear indication where the trend of the future goes into and how we will enable today's work into the next level.

Ruth: What are some of the significant challenges companies might face then when starting this shift?

Thomas: Today's situation is clear that this is in a prototyping phase, which means we are testing, and we are seeing how this could be a use case. And we have just spoken about factory automation, but obviously in companies like Avnet, we are also trying to find out how copilot GPT could enable our day-to-day work to make us smarter, to make us more knowledgeable about things which we may have not seen.

And this is all about data lakes. At the end of the day, these are all unstructured data. Where we do not have at least at the first instance, a clear view, what kind of content may be behind. But at the end of the day, you get a structured answer, and you see what potential solution, or the potential idea could be behind.

But back to your point about such kind of systems, we need to face the reality that in kind of. Test environment. You have by far less obstacles and less challenges, which you may have finally in a production side. And one of the key elements here is definitely a security because at the end of the day, when we consider the situation with connectivity, we have the situation with the robots, we also need to consider that the data which are used and the content, which.

Drives finally the robots can all be taken out from the environment where it is and be even reused. And this is some of the elements, which is quite an important topic for today to find the right firewall in between and to make sure that you have the right security as a cyber resilience for the companies to protect your knowledge and to protect the ideas and the data behind.

Ruth: Yeah, I think it's quite an interesting evolution, isn't it? Because when we, as you said, even 30 years ago, everyone thinks of the robotic arms that were basically just limbs that did mechanical work. But now, Michael, it's also like when you focus on AI and machine learning and then tie vision into this, those technologies then make, make a whole new story and whole different use cases, right?

Michael: Absolutely. This is also if you zoom in a bit more on some of the applications, for example, if you have a cobots or collaborative robots that can support a human in doing some tasks. And in this case, you can imagine like if you have a welding system so that the robot need to support the human that even the human can ask to the robot, please pick up this part, help me with instead of really taking the actions himself as a human, you can really interact with, with the robot of doing this.

So, this is something that is a game changer. And of course, this is something that since the implementation of generative AI is a lot of the tasks for this implementation, but this is a trend that we see absolutely.

Ruth: We've talked about automotive. We talked about smart manufacturing, but what I also find so exciting about this podcast and all the guests we had on the show before, it's always so interesting that technology now has the power also to solve some global crisis we are facing, whether it's sustainability or supply chains or other tech challenges.

So that kind of brings me and try to segway us into the topic of clean energy management, which is crucial for sustainability topics we are discussing. What trends are you seeing in this area?

Thomas: I've mentioned before about smart city and smart industry from the element of industrial and so this topic route is absolutely crucial.

Today, and I covered before about EV charging, we have spoken about smart manufacturing, but green energy or energy generation You just ask about some subsegments in here. We talk all about E meter ring. We talk about energy storage or some of the applications fit quite well to the technologies, which I mentioned before, and when we talk about renewable energy, needless to say that solar wind and many other applications are fitting here quite well into the entire picture.

But in consideration to the point about clean green energy, we also need to consider, okay, where does it come from? And what is actually the main reason for rent? Let's also face the reality that CO2 or greenhouse gas footprint is a main topic. We know it, for instance, from the automotive industry, where we have a CO2.

Level, which needs to be below, otherwise, uh, the companies would need to pay a penalty just for everybody's information. We talk here about 93. 6 grams of CO2 per kilometre. And this is, by the way, the same for industries. And when we talk today, for instance, about semiconductor companies. We have some where we have a greenhouse gas emission, where they want to reduce by 70 percent by 2025, going back already from four or five years ago.

And they would like to be a carbon neutral by 2030, which gives already a clear indication that the contribution of this companies is not only on profit. It's also about ESG, which gives us a clear topic about the environmental aspect. In addition to that. We also need to say that this market grows quite significantly.

So, the global energy management system on the market itself is growing right now at roughly 13 to 3%. We need to consider that the market situation right now does not give us a tailwind. We need to consider that in the next years. We will definitely see by far more integration of these as we see today.

However, when we talk about these markets and subsegments, like you asked before, here, we have energy management in building. We have EV charging station, energy storage, like I said already before. And I gave her as an example, the smart grid one, where we have spoken before about the bidirectional charging capability.

So, all this in, in one combined, we see a lot of customer applications, and I would like to underline this with one final one, which is a kind of mix between smart lighting and energy management. Here, for instance, we have development with customers where you have, for instance, the light on the road.

Bicycle cross by, or even a pedestrian, then you have the situation that the light switched on. And then if the person is away, then it's switched off. And it gives you already another indication about green energy and how this trend moves into the right direction.

Ruth: We had a guest on the show also that he mentioned that actually there are counties in England that cannot afford to keep the streetlights on at night.

So, for listeners who are wondering why is this relevant to save lights when a bike rides by, that's a real use case, right? And then you wonder why. How's this all not coming together when we are, um, having smart grids and infrastructure projects and yeah, wind farms and solar parks. So that's, that's really interesting to me.

Autonomous robotics, another exciting area we would like to touch on, which I think is really exciting to get like an open discussion. overview of everything that's going on in the industry and how semiconductors are shaping this world. In this case now, how are robots changing the industrial landscape?

We have started with some touch points, but what are you also still very excited about that we haven't mentioned yet?

Michael: Well, there are a few points we discussed about cobots, humanoid robots. I think most important is what is the value or what is the benefit if we take a look to autonomous robotics. And then we also need to take a look to the impact or the efficiency you can have.

And there are a few things that of course, with robots, you can enhance like it can operate 24 hours a day, seven days a week. But mainly one of the key things is like safety or a robot can do work that are complicated for a human or too heavy. So, this will help in this process also like increased production, for example, or to have a greater flexibility or adaptability when you use a robot.

On the other hand, of course, if you take a look to the challenges, because it's not only the good behind this, but there is also the initial cost. If you take a look to the investment for a humanoid robot. This is huge. On the other hand, it's a humanoid robot because it looks similar to a human and it can be immediately put into a manufactory, for example.

But on the other hand, there are some parts and especially on return of investment that it can be a concern if the price is too high. Also, technical complexity can be an issue. But again, with how we see how Things are moving on, especially with AI and especially with generative AI. It helps a lot to increase, for example, the way how humanoid robots are built.

I gave the example of figure AI. This company, if they are only like two and a half years old, what they were able to establish in this short timeframe, this is tremendous. This was not possible a few years ago. So, this is something that we will see that the speed of implementation. Of course, bots, humanoid robots will go faster and faster.

This will benefit them in the manufacturing process. Many of our customers

Ruth: and Thomas, do you see these considerations impacting market adoption?

Thomas: Absolutely. And this goes back to the point what I said before about testing and, uh, although about the way that we see by far more customer considering this topic.

Just to give you one indication, when we consider the current market situation and all the trend about the demand creation and demand creation is in our world, the engineering time and an engineering opportunity, which we see considering the market situation, which we have right now, the biggest. Trend respectively, the biggest increase we see today on the technology side come from this environment.

This gives a clear indication where companies have started to work on it. And in addition to where we're going to see by far more traction in the future, is this already in that way that we could say we're going to see it in one year and two years having this crystal ball in front of us would make us rich.

But to be honest, I absolutely not because obviously many other things would need to be considered. But, uh, in regards of. Considering as a topic and seeing where the use case could fit for an individual company to optimize things, to make things leaner, or even, like Michael said before; to have some cost savings and to reduce environmental impact, we see indeed by far more momentum as we had in the past.

Ruth: And generative AI obviously has been the hype topic of the last two years. Um, what makes this area particularly significant for you?

Michael: Well, this is a very interesting topic. It's also one of my favourite topics, to be honest, but if we take a look to how we approach this, so first of all generative AI, so it's a type of artificial intelligence focused on creating new content that can be text, images, audio, video, typically based on, um, on prompts today or in many cases.

And the best case we can consider is like a chat GPT from Open AI, which is well known by everyone. So, this is running on a cloud environment on high end servers with a lot of memory and processing power. What for us is very interesting is that because we are focusing also with customers on embedded devices is that we have this kind of implementation of large language models, even on the edge.

This is what we call edge gen AI generative AI running on embedded devices. Of course, there are some constraints here because you have less processing power, you have less memory. But on the other hand, what we try to find out is the limit of how far you can go with this kind of systems to support our customers for, uh, for several applications.

So one of the key parts here is that we focus on conversational AIs, so that's what we can also call a chatbots because we see that For us, this is in particular, a very interesting field, uh, to support our customer and it's something completely new when we talk with customers, it is an added value because they have something as an upgrade of their products.

And especially if you take a look to the potential we have with the generative AI, there is an expectation that this will become 1. 3 trillion market, a dollar market by 2032 and about 600 million. It's based on hardware infrastructure and inference. So, infrastructure is, of course, what you see on the cloud environment or the server infrastructure.

But mainly, if you take a look to inference, so running it on an embedded device, this is something that is expected to be 100 billion within the next seven years. And especially with generative AI, this is very new. So, this is a completely new market with a lot of potential for us.

Ruth: And those numbers are incredible.

Are there any particular applications or use cases that have surprised even yourself with their impact?

Michael: Oh, absolutely. Well, initially we started with an investigation and to find out what is now the benefit for us, because of course, generative AI, is something that everyone takes a look at, but we started with some demonstration, and we found out when we went to customers that this generated even new kind of applications or use cases.

What we did is we started with a kind of implementation where you have a chat bot so that can understand what you say, but also can speak where you have a large language model in the middle. I would say and even running on nine billion parameters, which is a quite relatively quite small, a large language model, but that we can use it for a use cases related to conference systems or, for example, industrial human machine interfaces.

And if you consider when we discuss this also with customers, it creates really completely new ideas. And let's consider a conference system that the. A chatbot is, for example, the moderator in the conference, it can be the assistant, it can understand the conversations, you can even ask to translate something if someone in the room is speaking another language, it can take notes, it can summarize the meeting.

And even send it with an email to the participants. What is quite interesting for us is that of course you can do this with a chat GPT, but this is you send your data to a third party. You send it to someone that maybe you do not trust. What we do here is this running all on embedded devices. So, there is no internet connectivity needed.

So, it's a secure connection. So, you can imagine if you have confidential meetings, that this is a very, very important topic. In addition, of course, besides data privacy security and no need of connectivity, it consumes also less because it's just running on the device. You can imagine that this is like a kind of internet that you have on an embedded device and that it's that is capable of understanding and reacting on what you're saying.

So, it's, it's to have a conversation with. So, this is a kind of use case, but I mentioned also like industrial HMI. You can imagine like for transportation or like bus stops, for example, where you have a visually impaired people that instead of trying to find out what happens, that they can talk to a system and can ask when do I have the next bus in which direction will it go or, for example, for hospitality, where you have an automated check in.

So those kinds of chatbots can help you. In assisting and answer queries, provide you service room, for example, if this is related to a hotel from desk application, for example.

Ruth: Could you walk us through how this use case operates in a live environment? How do I, how do I have to imagine this?

Michael: Well, we did several tests.

We did several kinds of implementations. What we also did is we build out the use case to have it more in detail. We can hear it on how this is working. So, in fact, if you consider like a hotel from desk and you arrive late in the evening, for example, you do not have anyone on the reception. That this kind of conversational AI can support you and can do the check in.

What we also did is like connecting this to a database, so it knows how the hotel looks like, where the rooms are. It can guide you to your room, for example, you can ask questions about it. It knows how you booked. Uh, for example, if you forgot the breakfast in the morning, it could ask you, do you want to do that or not?

So maybe the best way is to listen to it and to have an idea how this is working.

Scene: Hello.

Michael: Hello. Good evening.

Scene: Evening. Welcome to our hotel. Could I have the name on the reservation, please? This will help me assist you more efficiently with your check in process.

Michael: My name is Michael.

Scene: Thank you, Michael. Let me just confirm a few details for you.

Was ISIC that you have a reservation for two nights under your name? That correct?

Michael: That is correct, yes. It's for two nights.

Scene: Excellent. You will be staying in room 223, which is located on the second floor. You can take the elevators just around the corner to reach your room also. I'd like to let you know that breakfast is not included with your reservation.

You'd like to add it for an additional 20 euros per day.

Michael: Yes, I would like to do that. You can add it for the, for the two days that I stay.

Scene: I've added the breakfast option for both nights of your stay. The total amount you will be displayed on your receipt at checkout now. Would you prefer to settle the payment for your stay and any extras now?

Or would you rather take care of it in the morning when you're checking out?

Michael: I prefer to do it in the morning if that is fine.

Scene: It's perfectly fine. Michael, I've updated your account accordingly. You can simply head up to room 223.

Michael: Like you heard, so this is something, this is even running on a local device, but it's, it's, the quality is, is good if you compare it with what was possible a few years ago.

So, this is something that is completely new but gives a lot of possibilities.

Ruth: And beyond generative AI, what other AI trends should we pay attention to?

Michael: Well, there are a few things. One of the key ones today, I would say it's like cyber security, as you can imagine, with the situation we have today worldwide, that's the implementation, like automated threat detection and response systems, also like analysis behaviour.

Of people and having proactive security. Those are one of the key trends that we see also in, uh, in generative AI and in AI in general. And I think Thomas, you also mentioned some additional ones.

Thomas: Yeah, I can cover one, which comes from, uh, from sports. If we consider today professional sports, where we have, for instance, data, which are captured during a game, uh, let's take for instance, football, soccer, you have the situation that a lot of these players are wearing a kind of bra and, uh, in the bra, you have sensors, uh, which are, Measuring the movements in three dimensions.

So, X, Y, Z. And while at the end of the day, you bring all this kind of data together and a kind of metrics, you can put this kind of data into a one data lake. Finally, you just take the data out and you can, for instance, put virtual reality glasses on, and you can bring yourself into the body of each individual player on the pitch to see each movement the player did during a game.

And while we are just talking about generative AI and considering the next step, just bring yourself into the position that you are playing next Saturday against another team, and you would like to. Train during the week already, some movements of the others, how they were moving into the centre or how they are moving out by putting, for instance, a virtual reality class on and then bring such kind of AI intelligence into the virtual reality world.

You have easily augmented reality situation, which you can train at the pitch with professional players and particularly the first point, which I have spoken about before. This is not the future. This is already a reality. There is a company in UK called Resil. This company captures data during the game, converts this data finally, and then you can use this in your training to optimize the players and to see how it moves from the level where they are to the next one.

And this is not only about the players from the perspective of the tactical can also be optimization of cognitive things. where you can bring themselves to the next level to combine the right and the left part of the brain and to optimize situations. And you see, this is already something where we see today in professional sports.

And for us, it's just a matter of time where we're going to see it perform more in our day to day, respectively, in a business environment.

Ruth: So, before we wrap up, is there anything else you'd like to share with our audience?

Thomas: From a semiconductor perspective, I would like to share that the future gives us definitely the right momentum from a technology perspective and from the innovation side to further growth.

It is not an advertisement for a semiconductor industry, but this is an advertisement for the technologies which we see today in our world. And I'm confident that with the integration of AI, thank you for having me. We're going to see the next level, uh, of, uh, learning of innovations and all the things which are going to get changed.

However, I'm all the neutral and cautious in the way of how this kind of data is used and how we're going to make sure that we have certain firewalls, security, and privacy in regards of the individual data to ensure that nothing will move into the wrong hands.

Michael: I think what is interesting for mainly than our customers and also as a kind of a message is that they do not need to be afraid to start doing investigation with this kind of new technologies.

It is quite tough to follow everything related to AI today on what is ongoing, but if you see what some of our customers already. have in place by doing this kind of investigation. This is huge. And I think for our customers to start with it or for customers that are new for investigating on AI, that this makes sense to start the investigation.

Ruth: Thomas, Michael, it has been great. Thank you so much for sharing all your valuable insights with me and our listeners today. It has been a terrific episode. Thank you so much.

Thomas: Thank you. Thanks a lot.