Transform: Siemens Innovation Day India 2026

You will also find nine of our partners about how they are actually using some of the technologies, how they are building on Siemens accelerator in order to make all of this work. So, moving forward, the role of Siemens that we want to play and that we're playing in all of this when it comes to industrial AI is where we can connect, as I said, the real world with the digital world. We can bring together the data together with the applications, and then all, of course, the competencies that we need to have. And you find many statistics here on there. What I always find most important is the following, that getting data is one thing, but actually the key thing is getting the right data and understanding the domain that you're operating in. And one of the things you're seeing here on the slide is that actually every third machine that's operated in the world is being actually controlled by a Siemens controller. Which means that gives us the data in order to infer and to make, of course, those models available in order to address some of those use case challenges when it comes down to quality, productivity, time to market, whatever the challenge you're looking at solving. And maybe one fact that many of you are not aware. Siemens is the number one company when it comes down to industrial software. We have more than 11 billion dollars of sales in the digital business, which makes us the 16th largest, 16, so 16th largest software company in the world. So, probably you think of Siemens as these are the people that electrify. And you're absolutely right. This is what we've done the last century, and this is going to continue to do so, and we bring more digitalization and electrification. Then we moved over to automation, and then we automate the world, and then we digitalize the world, and now with that digitalization finally comes the industrial AI all along. Let me say it just a few words more about industrial AI, because actually it really helps to understand why this is different. For industrial AI, as I said, you need data. And most of those large language models today, they're being trained on the internet. That does not work in industry. You cannot go on the internet and download production data, design data, engineering data. This does not exist. So, the only way, the only way how we can build industrial AI is by doing it together. I tell you, each of you in your operations, in your company, you probably sit on your own data, which is the good news. The bad news is that data that you're sitting on is never going to be enough and actually building these models that need to be so reliable and so accurate that they actually make a difference. Even Siemens, we sit on about 300 petabyte of data. We have 2 million CAD files, all of this. This is a lot of data, but still not good enough for building something in the industrial world to make those applications work, because they need to be heterogeneous, they need to come from different places. And this is why we have to work in ecosystems, in data alliances, and bring actually that data together to train a model on specific use cases, and then deploy it to all of us. So, therefore, we can work with all of you. We are serving today 30 different industries from cement, glass, semiconductor, pharma, life sciences to data centers, where we can provide that data and those solutions to make that work. So, nobody can do this alone when it comes to data, but also nobody can do it alone when it comes to technology. And this is why we're also partnering a lot with Nvidia. In early this year, at CES, we spoke about how Nvidia and us are bringing the latest technologies together, and this is going to be GPU-accelerated and going to make a significant difference in the industrial world. And you're going to see two use cases out there where we talk about something that we call the digital reality viewer, and also about the digital twin composer where help you to build this digital world in a much much faster way that then connects you to the real world in real time so that you can make better predictions. So, how does it look like? This is an example and you're going to see that video also out there. It's a simulation in the end where we are able now to scale it to an amount of complexity where you can build these ships, ammonia propelled ships of 7 million parts where you can actually simulate factories where we think about how can we build these GPUs in a much much faster and more efficient way on a daily basis where we can think about shop floors, how we change them in order to make cut out of times much faster and of course for India in particular important, how can we bring railway simulations also faster time to market. So, all of this can be only done if you do this together. And that's why we are really really happy to also have today two of our key partners there. So, please everyone welcome with me on stage Vishal who is the managing director of Nvidia and Sanjeev who is the CEO of Adverb Technologies. Welcome Vishal. Welcome Sanjeev. Great to have you here and thank you for joining me here on stage. This is fantastic. And Vishal, maybe I start with you. Obviously, we are building a lot of great latest technologies together. What's all this role for you in India? Maybe you can share a few words from your perspective.

Wow, okay. So, that's 20% faster for more productive, 90% capturing failure rates earlier. Virtual twins. Now, Sanjeev, tell us a little bit about Adverb, about what you have founded and what you created and how those numbers are reflecting back on you using Siemens and Nvidia technology.

Great, thank you Sanjeev. Fantastic obviously because fastest growing robotic company. Now, now Vishal, you mentioned that we want to unlock that power. So, we really want to contribute to the next generation of manufacturing. How do we unlock that here in India? What are the challenges and how can from your point of perspective, how can we overcome them?

Fantastic. Great examples there, and maybe to conclude then Sanjeev from your perspective. So, we want to bring technology from India to the world. So, you've built all the quadcopter pads and the mobile robots and so on. Tell us a little bit about where do you see those industries and which markets are really important as you go from India and then conquer, of course, all the markets that are out there.

Thank you so much, Sanjeev. Thank you so much, Vishal for joining us on stage. You can't go. We need to ask you a question, too. Okay. How can it be only one-way traffic? So, we got Peter, who's very fortunate. He gets to see everything across the globe. The mega scale when it comes into play, which India needs to play one day, Sangeet is already going through that experience. What is one or two things that we can learn from you that India should be sensitive about instead of learning on the spot. If that wisdom is available, we can overcome it. So, give us one or two lessons that you've seen across the world.

That's so good. Go beyond POCs to implementation. If our country became a shining star with less than 1% of developers from our total population, imagine each one of you now is a developer. You can speak to the computer in your respective language. You can mix your language. Computer understands it, program it, go from POC to production, and make India really smart. Fantastic. Great summary. Very good. Thank you so much. Thank you. Wonderful. Okay, you're okay, we should take a quick picture, I understand. Okay, good. Very good. Okay, thank you once again. And I hand it over to Indu, our master of ceremony.

So, I talked with Sunil. He had so many numbers for you today. I'm not sure how many numbers you're going to remember going out of here. And I'm not sure if I remember all the numbers that you had on there when you spoke about the Indian economy. But, there was one thing that I kept in my mind, and this were the arrows. And these arrows, they pointed all steeply upwards, and it's very true. No matter if you talk about the data center deployment, no matter renewable energy, no matter about airports, semiconductor plants, and so on. All such a steep increase, you rarely find this anywhere in the world. So, it's a super exciting place.

Yeah, thank you so much for the question. On ethics, I think it's a little bit different in industrial AI as opposed to, let's say, general AI where you talk about, of course, a lot of discrimination and harassment, everything that the AI can generate in language. In our case, it's not as much. If you think about what the AI is doing, it gives you recommendations of what you had Sanjeev talking about, what-if scenarios, should I produce this, should I produce that. So, therefore, you don't have that ethical question per se. What you do have, though, is the trustworthiness. So, what we always argue is how do we make AI trustworthy for an engineer to say, 'Actually, this is really working.' And so, it's slightly different in that regard because we don't have biases and all of these kind of things. But your second question is a very very important one, and that is, of course, the data security element of it. And there's different answers to this. One, of course, the model is going to be trained on a large data set which can come from all kinds of sources. It comes from us, comes from you, comes from synthetic data, actually. This is where also we work with video quite a bit to create, of course, all the synthetically generated simulations and pictures and images and everything in order to train the AI. But once these models are available, they are generalizable. And so, therefore, we don't see there too much of a specific risk for the individual company. It's getting interesting, though, once we deploy that industrial AI, very often you're going to have fine-tuning based on racks and everything. And this is where we need to make sure once it's deployed, of course, it cannot be hacked. So, as you go out there, you're going to see an entire section with regards to cybersecurity and where we have the usual elements of of course the zero trust and everything in order to safeguard the data to your very point. Yeah, so very good question. Thank you.

Yeah, I'd say domain know-how is quintessential for making the AI actually useful. Otherwise, this will not work. I give you the example there a train example where today we can do predictions on doors failure of a passenger train, and we can predict that based on traditional AI, but also on gen AI in order to make this work. And that's domain know-how. You need to understand that whole day the doors of a train go open and shut so that you have to really focus on the key parts that are going to fail. That's on a train, but the same applies to for factories. You need to understand which parts of the machine is going to break down, what kind of material flow that is there. So, you need to understand the domain specifically. So, you need to train that to the AI. So, you cannot separate it at all. What I think, though, what's important is that that you can fine-tune, this is for the earlier question, that can fine-tune that AI to your very specific manufacturing process because, as I said earlier, every industry is slightly different, every manufacturing site is slightly different, you probably know that. And so, therefore, it has to be adapted and fine-tuned to get to the level of accuracy that I mentioned. This last point, there's always going to be some human in the loop. In a way, I think for the foreseeable future, I know we all talk a lot about agentic, but even in the agentic space, eventually there's some human in the loops because we are not there yet in the completely autonomous systems, but we are certainly getting closer and closer. Thank you.

And let me just one quick sentence to add on this one is it's augmentation. And then so, the idea is repetitive tasks are being replaced by automation, but in this case, by of course, by AI, but you heard that answer before. And then then really it's getting get to the human nature that is about creativity, problem-solving, empathy, and communication, all of these elements that you can free up. And so, we believe there is a huge opportunity to become more productive in that regard. But yes, there will be there will be definitely some changes with regards to training and of course the level of skills that are required going forward in the Indian workforce.