Hannes Niederhauser CEO Kontron Interview: Wenn nur zwei der fünf Zukunftstechnologien durchstarten

Good day, Mr. Has. Yes, my name is Hannes Niederhauser. I am a dinosaur among CEOs and 61 years old. How did I get my CEO job? Well, I was there at the IPO in 2000, I think 24 years ago, and I will soon be presenting my quarterly report.

Yes, it fits my introduction. A technology company must adapt to new trends. In technology, especially electronics and IT, data and technologies have multiplied a hundredfold, and Kontron has experienced all of that. What does Kontron do today? It does only one thing: we connect machines and devices to the internet and control them remotely. That simple. It's called IoT. Our products existed five years before that word was invented. We remotely control medical devices, for example. We were in almost all COVID respirators. We are in welding robots and smart factories, controlling all devices. We are in all kinds of vehicles—4,000 aircraft are controlled by us. We have over 50% market share in high-speed trains in Europe. We are involved in the first self-driving cars driving around the world, and we have self-driving dump trucks on construction sites. So many applications, but our technology is always the same: we connect a device or machine to a large server where decisions can be made about what the product should do. Why do we do this? A welding robot that communicates with other welding robots pays for itself in three months. It's a must in this industry. For trains, my vision is that today on the highway, 500 cars pass by in a short time, but on rails, a train passes every five minutes. That's a massive waste of tracks. With trains that communicate, I can run 50 times more trains on the same tracks. That's one of the goals. If we make the tracks faster, straighter, with tunnels, and run more trains, we can connect Europe's 20,000–30,000 km of tracks to all major cities. All of this is done with IoT.

We don't make chips at all, but we control the factory at Infineon, for example, and at Parmac. Our software, Susitech, is used to improve production processes in the semiconductor industry. We are 70% a software company. We have over 3,000 engineers, 2,000 of whom are software engineers. Most of it is software. In communications and protocols, software is more important, but hardware is essential. We use Qualcomm chips to connect trains, for example, but we are a customer like any other, though a strategic one for Qualcomm or Intel. So we use the technology as a customer, but I would describe Kontron more as a software company.

I see it differently. My house is not representative, but when you come in, there are over 300 IP addresses. There are 4 to 5 billion people connected to the internet, and already 30 billion machines connected that you don't see. The refrigerator is not one, I agree. But every smart meter is connected, for example. Our customers have machines costing €100,000 or more. You don't get an airplane for €5 like an Alexa. In the professional areas where we operate, it's absolutely state of the art. The big breakthrough will come with self-driving cars, of which we already have 2,000 as tests on the roads. There is no reason not to have them. Many more drunk drivers hit bridge pillars than self-driving cars. Self-driving cars are already 100 times safer. You can load all traffic signs into the system. You can stop without traffic lights. In trains, it's standard. 50 years ago, trains had traffic lights; now they use GPS and servers. That will come to roads too. Pedestrians will get airbags so they don't get run over. The IoT wave is already in full swing, even if not visible. When I walk through the city, I find at least 20 applications where Kontron is inside—trams, subways, trains, traffic lights, highway speed lanes, elevators, coffee machines. So it's already there, and you see it in the 30 billion figure.

Our biggest competitor is Advantech, also from China. In America, Mercury, for example. What does the customer pay for? This IoT know-how is very specific. A major part is security. We have our own hacker department to test our systems. You don't want an airplane, car, or train to be taken over by a malicious hacker. Security is crucial, so we developed our own operating system, Kontron OS. The customer pays for security and for the fact that they don't have the know-how in these protocols, which vary greatly, e.g., for defense technology. They are proprietary protocols. I always laugh at the movie Independence Day where they upload a Trojan to the aliens' computer. That's complete nonsense because they have a completely different operating system. A virus is always tailored to something, and it's the same with our defense and train systems. Different protocols, different languages—there is no virus that can attack them.

We already have dump trucks in use with Caterpillar—each vehicle is a customer worth 15 million, and it works excellently. What is the barrier? It's clearly politics; they haven't given approval yet. It's easier to build a skyscraper than to get a permit. The same problem exists here. I believe Tesla's system is also ready. We at Kontron have partnerships with Waymo, Lyft (the taxis), Bosch, and also BMW in certain areas. What do we do for self-driving cars? A very simple function. Kontron is a real-time specialist. If I talk to you and the data packet arrives a minute late, the world doesn't end. But if you stick your finger into a welding robot and it doesn't stop, your hand is gone. Real-time means commands and controls are given within a defined time, typically 10 milliseconds. In smart grid, it can be 15 minutes because it doesn't need to be that fast. A car must be real-time capable. If a child steps onto the street, the car must stop. That's exactly what our system does. We have a computing system integrated with Intel's six cameras that recognizes something special and applies the brakes, for example. It doesn't control the air conditioning; it handles safety-relevant functions. If a bridge pillar comes, it steers left. These are real-time central functions, programmed in assembly language for speed. That's edge computing because you can't upload to the cloud and decide there. The data comes from the OEMs, and we work with partners like Intel and Pioneer. Our function is absolutely local, with high computing power for image recognition. Real-time is the key. I want to mention a number: the first systems we brought to market two years ago cost €140,000. You can't put that in a €40,000 Golf. That's for testing. A dump truck costs about €25,000, which is advanced. The auto industry wants it for €3,000. Kontron won't achieve that for cars, but the dump truck will stay at €25,000. For cars, someone like Foxconn will build the electronics, but they will use our software, and that's what we want.

We have five key technologies, and we will develop superbly if two of them break through. It's like private equity buying several companies and two succeed. AI is the youngest; we only make about 3–4 million euros in revenue from it. What do we do? ChatGPT and similar systems are very expensive and know a lot. ChatGPT does 5,000 teraflops and has the knowledge of half the world. Our systems are much less complex. For €3,000–4,000, you get a system for parking lot management that remembers credit cards and links data. Or a major benefit for our customers: training a new worker on a robot takes three months. With an intelligent system that communicates and guides them, AI can reduce the human-machine interface to 3–4 days. That's what you want for a smartphone too. Our devices are mostly custom, customer-specific. A welding robot works differently than a respirator or high-speed train. The data protocols and security are almost the same, but the last 20% are always customer-specific.

Well, the application costs money, and it's the same for our competitors. We make money; Kontron is not in the red. We have double-digit EBIT. The costs are there and must be paid. I can't start the 400 high-speed trains with the self-driving car because they are different. The last 20% we develop ourselves, which gives us high customer loyalty. I can't remember ever losing a customer except through insolvency, and I can't remember a competitor taking one away except through their insolvency. Once a customer has our system inside, it's there for the entire lifespan. After that, there's a new tender, and we win because we already master the technology and their internal processes. Our entire battle in the market is for new applications, like when self-driving cars eventually come. But most customers stay.

Katek is 80% of the problem. The not-so-simple structure maybe 20%, but definitely Katek. Kontron lost 20% market cap this year while increasing profit by 30%. It doesn't fit. People say Katek was not a good acquisition—you are a software company buying a hardware company. I am convinced of Katek, otherwise we wouldn't have done it. It's our strategy: we buy hardware companies, put our software on them, and increase gross margins. We already raised Katek's gross margin from 30% to 36% in the first six months. It's the same thing we've done four or five times before. People are very pessimistic, especially about photovoltaics. Katek also has photovoltaic revenue, but in Kontron, photovoltaics are only 7% of total revenue. I don't know why the stock should fall 20% because of that. I had an investor who said, 'When I see the word photovoltaic, I see red. I've lost so much money with those systems, and I won't invest in any company that does that, whether it's 7% or 100%.' So they just don't like photovoltaics.

First, about wallboxes: they are running superbly. We are almost making money with them. Why? There are many wallboxes, some smart, but a connected smart wallbox is ours. It has a specific function: it can only be at home in your garage, and it is matched to the car. For each car, exactly one wallbox. If your wife has a different car, it's useless. We have customers like the German sports car manufacturer, who now sells his cars in the USA and China only with this wallbox. The largest German car manufacturer is also our customer. Our customers are only car manufacturers, not hotels. It has two big advantages: first, it is connected to the car manufacturer's server. They have a power contract and can buy electricity at cheap times on the exchange. If you go to the website of the largest German manufacturer, they tell you that with this wallbox, the VW ID.4 costs €2,000 in electricity per year instead of €4,000. The wallbox costs €1,000, so it pays for itself in six months. Second, and more importantly, the battery is not a simple vessel. Chemical processes change with charging cycles. With our system, during charging, we can detect the battery's condition and even heal or improve it. Intelligent connected charging can at least double battery life. It would allow car manufacturers to give 5 or 10 years of warranty on the battery. It's a no-brainer. We have about a year's market lead. For photovoltaics, it's not the case yet. We are not that far with intelligence. But we are making AI systems for home energy management. We can predictively say how the electricity price will develop tomorrow. The system can decide: send to the heat pump, to the grid, to a battery, or to the car. Next year, our system will make the car part of the house as a battery, with bidirectional charging. For the energy transition with volatile solar and wind, we need smart systems that charge everything when the wind blows. 20 million electric cars are good storage. They can also be used as uninterruptible power supplies. If lightning strikes and the power goes out, the car kicks in. That's the HEMS system we are working on, to be released next year. With software alone, I can achieve as much as doubling the solar area on the roof.

2024 had the big Katek acquisition, so we have enough to do. There will be nothing in the first half of 2025. By the second half of 2025, I think everything will be digested, and we will continue to acquire. Kontron wants to double every four to five years and has done so. When I started, we had 30 million in revenue; now we have 2 billion. That has always been maintained. It's like Moore's Law for Intel: every four years, double. The IoT market grows at 30% per year in connections, according to Gartner. But there is price degradation, so revenue grows about 15–16%. In a recession, maybe 10%; in good times, 20%. If I calculate 1.16^4, that's not yet double, so we need acquisitions. There are many companies that have missed connecting their products to the internet and making them smart, like Katek with photovoltaics and wallboxes. We can buy them cheaply and quickly generate added value. It's easier to make a bad company good than to make a good one even better.

Thank you for that sensible legal disclaimer. Kontron is a play or a gamble on growing new technologies. That chance is not reflected in the stock price. People say 'show me,' but Kontron has shown it for the last 12 years. We are not known for profit warnings. That's the opportunity. The downside: we have our hands full making Katek profitable as needed. That's a negative. Another topic: 85% of our revenue is in the EU, 30% in Germany, and Germany is currently the sick man of Europe. We are not a global player but a European player. However, our technology can compete worldwide, and the USA and China like us. If I were to sell Kontron in bits and pieces, like in the movie with Richard Gere, I think a much higher price would come, but that's not our strategy. But if someone offers a lot, we would do it.

I don't feel comfortable with the P/E, but that's how it is. We have no profit warning currently and don't see one. In January, we will present our agenda for 2028. Another thing investors often ask: we have a large shareholder, Foxconn, with 27%. Foxconn is the largest electronics manufacturer in the world with 2 million employees. In the Foxconn universe, there are thousands of companies like us where they are invested. Some investors turn up their noses at Foxconn, but on the other hand, we were just at the Foxconn Technology Days, which is like the Apple iPhone presentation. I think the idea that Chinese only copy is long over. Foxconn is Taiwanese, so it's not the typical China risk. But in America, having a large shareholder from the People's Republic can be a disadvantage when winning customers. However, there is another Austrian company in aviation with a large Chinese shareholder, and it hasn't hurt them.

This is our 5G module. It's an internet connection with a processor inside. We sell it to cars. If you drive into a tree, it calls for help. There are six manufacturers; we are the only one not in China. We are making a new version for trains, and we will be the only manufacturer worldwide. Every train manufacturer will have to buy it from us. It doesn't cost much, maybe €60. But inside, there are ten chips and a lot of software.

Yes, this controls an elevator. It's in the car as an example. Both systems have processors and a lot of software.

Appliance. Yes, hardware and software. Good that I showed it.

Thank you very much.