Future Back Drinking • Roland Busch

Cheers.

Yeah, I'm a physicist by education and I'm not the first head of Siemens who has a technical background. There was maybe 30 years ago another one, also a physicist. But my way was basically, I started off just during my PhD, I was supported by Siemens. I started off in corporate technology. I moved on to automotive business. Had a chance to run our automotive business in Asia. Was living for three years in China, in Shanghai. And then when I came back, I got a chance to run the strategy of Siemens. So a huge opportunity to get to know all these businesses. And from that onwards, after three years I was promoted to the managing board, running different businesses. Most of these businesses were in bad shape, so I didn't have the blessing of running a business which is running well. I have it now, we are running very well, but I had to fix a lot of businesses, which is okay. It's more my style to say we start at some point and try to improve it. And then four years ago I was selected by our supervisory board as the CEO of Siemens.

That's a very good one. I mean, surviving for more than 170 years is amazing, and it's not just surviving. I think we are stronger than ever, but it's a completely different company than the company which you would have experienced maybe five, ten years or even fifty years ago. So we made a lot of changes, but there's one constant in our history, which is transformation. We started off with the telegraph, which is what you would call communication infrastructure, with electrification, with power plants. And now, we sold power plants, we were market leader in communication technology, we sold it off to Nokia. And at the same time we started acquiring businesses. We didn't have software except embedded software for our hardware, and since the last fourteen years we spent twenty-one, twenty-two billion in acquiring software, including the latest announced acquisition for ten billion. So think about it, this is a massive transformation and it's going over and over and it doesn't stop. We keep on going. We are just about to make the next transformation to our company and it's about the speed. This will be the fastest transformation ever in the history of Siemens because technology is changing so fast. Think about the speed of semiconductor development, of software development, AI development. And this is so massive that we have to stay tuned and have to be extremely agile in the way how we, number one, shape our portfolio, but also the way how we work. We have to come to agile way of working. We have to come from a less hierarchical organization into a more empowered organization where you make decisions on the lowest possible level. We call it one tech company program. The one is the specific one because we have to act more in concert to really unleash the power of Siemens.

Yeah. So one thing is clear, if you go maybe five years down the road, you will see there will be a difference between companies who are embracing AI and using it and who don't. The train is out of the station and it's just about how you're going to use it. And you're right, we are working since decades on AI. We have, I think, fifteen hundred AI experts in our company working on AI technologies. We even have research on it. People are writing papers on knowledge graphs and all that stuff. So we have also patents. From the patent statistics, we are one of the leading companies in AI. So the key question is how can we make AI usable in a trustful way on the shop floor without hallucination. You need to be able to rely on what an industrial co-pilot does on the shop floor in running your machines, and you cannot have a robot sitting on the shop floor without a fence. This is not going to work. So making this technology hard, make it industrial-specific for certain applications and making it useful, though it really drives a business case on the shop floor. That's the name of the game. Managing grids, managing trains, driving trains autonomously, but also running robots with a natural language on the shop floor. This is what we want to do together with our partners. Then you talk about the impact on people. We're living in aging societies. If you take maybe Africa or India aside, most economies are aging societies and the labor shortage, in particular the shortage on skilled labor, is so immense and it's just the start. So for me, it's not the question about jobs and endangering jobs with AI. It's more an answer to our problem of having a huge shortage. So driving productivity with technology, with AI technology, is something what I believe is a must. And we have the right platform, the right technology, hardware and software, and the right ecosystem, our technology partners, to make it work for our customers.

So this is a reason why we are spending more than four hundred million in education and training of our own people. We have a dedicated platform where we are training how to prompt, which is a very interesting element, how to use AI technology. We have it for software developers but also for everybody, the way how you can use and should use AI technology in order to use the large language models with whatever space they're working on and whatever function they're working on. That's important. So to lower the barrier to use it on the one side, and to encourage people and of course to train them. So there's a lot of angst and anxieties behind, that's clear, but that fades away once these people start understanding that the way how to make their targets and how to achieve higher productivity is obviously in using technology. However, there's one common, regardless what function you're working in, it will change your job. And this is the whole thing about transformation. People very easily talk about transformation but it's more in the sense, yes, we have to transform and maybe the others start first. No, no, no, this starts today and it starts here and it starts with me. And staying in this comfort zone to say I know what I do, I learned my stuff in university and I deploy it for the next forty years in my job, this is not going to work anymore. I rather go for a model where people study maybe a little bit shorter, they go to work and they come back to university, study again, and they go to work and they study again. So you constantly have to update yourself and stay tuned. I, by the way, I tried to do that at the same time. I had a coding lesson. I tried to code an app with large language models. Without my experts I would have badly failed, but at least I got a glimpse on how they work and what they do. And so I tried to stay tuned as well.

If you talk about the partnership with Microsoft or with NVIDIA, we are working with them on this industrial metaverse. There are many different ways and I just sketched two or three of them which I think are important. Let's start with the Microsoft agreement, talking about the industrial co-pilot, which is in short basically programming a robot on the shop floor with your natural language. So a shop worker, blue collar worker, can talk to a robot and say do that job or I want you to go from here to there and do another one, or you have a chat where you say I have a problem here and the system comes back and says in the last twenty cases of this problem we did this and that and it worked, and then you can go on, or they even fix it or make a proposal how to fix it. The other dimension when you talk about cities, and the best example is Siemensstadt Square, which is a new precinct which we built. And we built it twice. We built it once completely digital and not only the buildings but also what's life above and below the ground, the heating, the piping, you name it, the mobility space. So anything that goes around on energy, on mobility, on housing, including of course photovoltaics, renewables, reusing energy, reusing water and whatnot. And once we were satisfied with the way how you run a basically carbon-neutral precinct, then we start building and excavating. And once you start simulating, it's not only building in a 3D design but it's how you simulate how it works, how it operates, then you figured out that we still are not there with carbon neutrality, but people found out in a digital drawing that there's a sewage pipe which brings stuff away and it's warm. So if you connect it with a huge heat pump then you can basically run this whole precinct in a completely different manner how you manage energy. This is what you can get once you start simulating. And the third element which we are working on, and this goes more in the direction of NVIDIA and AI enabling technologies, is the industrial metaverse, which is basically a real-time, physics-based digital twin of a factory. And it's real-time at any point in time, so it behaves like the real thing. It's a massive driver of productivity but also of innovation if you really think it through.

So you will see a difference of the level of adoption and the speed of adoption depending on the countries or regions you are talking about. And let me give you three different views. Let me start in the western United States. Limited regulations on the way how you use AI. Extremely high IT maturity. Less OT maturity. So operation technology is more the hardware stuff, it is more the IT stuff. Tech companies, they make their living from data. So they know how to make money out of data. Therefore their readiness to adapt, to use data and to drive technology extremely fast is very high. Yet it's not really so close on the shop floor yet, which is not their strength. It will come and we try to bring that to the United States market. Moving on to Europe. I know I can speak a little bit open. I do believe we overregulate. We regulate before we innovate. It's just too heavy of regulation. So that some companies like Meta, they even don't offer their AI functionality on their glasses because they don't know how the liabilities are. It's unclear regulation. So you cannot even be sure that whatever you do that you're fulfilling regulations. So that's very, very bad. That holds us back. And it's not only the way how we regulate AI but also the way how we are using data or allowed to use data. Yet on the other side we have an extreme strong base on OT technologies. We have super strong machine builders in Germany, in Italy and other places. Strong car makers. We have strong, I mean we talk about medical equipment, I mean Siemens is one of the leading technologies there. So therefore I believe if Europe doesn't make a change here in using the regulations on the one side and on the other side leveraging the OT capabilities and bring them into contact with AI, then we would slow down or fall behind. Now I'm shifting to China, a very different place. As you know, there are technology restrictions from the United States. They don't allow the latest and greatest GPUs to go to China. So their compute capabilities in China are lacking behind. But what I see there is they make a workaround. They still have some capabilities, they upgrade them. They are still maybe one node behind, maybe two nodes behind in terms of performance, but they will develop smarter algorithms, how can you use less compute power because you just curate your data first. So therefore they are extremely smart and innovative and fast in order to make another way of using it. Coming to data, if the government wants it, and this is an example for the healthcare system, they open their data, the patient data, to startups in order to make a productivity reading of the MR scans in order to provide more healthcare service to their people. They have 1.3 billion people, aging society. They don't have enough doctors. They can't buy enough equipment, but they don't have the doctors to read it. So they give the data, they release it for the benefit of the customer at the end of the day. But on the other side, for some data you cannot access them. They have a clear regulation. So they're sitting somewhere in between, but it's a completely different animal. So therefore, depending on where you go, you see different ways of development and speed.

Since we are such a company with a broad portfolio, I give you a couple of ideas what I hope the legacy would bring. Let me start with our mobility business. One is signaling in the cloud. So signaling systems are, city level four, highest kind of or one of the highest levels of cybersecurity because it's about running trains. You don't want to have that hacked. And we are the first company bringing signaling in the cloud. What you basically can do, you can run a whole country out of one cloud system. You don't need signals anymore. You can provide twenty percent higher capacity on the very same line, saving money and even have a higher security level. This is mind-blowing technology which we are already bringing to the market and I believe this is the future. Another one would be the industrial metaverse. I hope that we're coming as close as possible from a digital twin to this really comprehensive view of physics-based, real-time photorealistic representations of a whole plant. We can do that for cells, but maybe for plants. And maybe the last one on healthcare. Two innovations which I'm really proud of is a photon-counting computed tomograph. Currently you use a certain sensing technology which translates X-rays into optical signals. And these sensors, they count each individual X-ray signal, which is higher resolution and lower dose. And the last one is the MAGNETOM Free.Max. Normally you need thousands of liters of helium to cool down a magnet for a 1.5 Tesla MRI. We brought it down to a liter and having the same picture quality using AI technologies. So we can run a 1.5 Tesla magnet which provides pictures which are as good as a three, three-and-a-half Tesla magnet. So using AI technology in a complete innovator. So if that's legacy is what I leave behind in terms of innovation and scaling in the market, I would be proud.