Good afternoon. You know, I've been with Siemens ever since the software movement started years ago with the acquisition of UGS. I'm now just starting my 10th year as CEO of Digital Industries Software. I'm excited to be here. I'm excited because I get a chance to talk to you about what we've done to build the progress we've made in building the world's leading industrial software portfolio. We did this with a very clear game plan of organic growth and inorganic growth. From that, we've built a larger addressable market for what we can go after with our software, but also we've set ourselves up very, very well to be the leader for industrial automation. Over the next few minutes, I want to talk about the value and the benefit that creates for our customers.
If you think about our customers today, they really have relentless pressure in the product development process. You've got new products, they've got electronics and software being added at a rapid pace. They've got to deal with the features in those products that come from that addition, as well as sustainability requirements, regulatory requirements, they've got supply chain concerns. All of these things are making it very, very difficult for our customers. And some would argue, they would say, let's try to simplify that complexity. But that's not realistic. That complexity is increasing. In fact, if anything, it's growing very, very rapidly in what we're doing. So we take a different approach. If the complexity is there, why can't we help our customers use that complexity as a competitive advantage? And the way they do that is by using Siemens Xcelerator. We help them create a digital enterprise.
Let me talk about what we mean by that. Siemens Xcelerator, it's not just a brand, it's the backbone of our strategy. One of the first components of this is the comprehensive digital twin. You heard Roland talk about this earlier today. I just want to say, if you've been around this for a while, the digital twin is not a new concept. It's been around for a long, long time. It really was 3D models that went together to make sure parts go together properly. The holes line up, there's no interferences. And while that was necessary, it wasn't sufficient. What's really happened over the last five to 10 years is the idea of digital twins that are more based on functional characteristics of the product. Is it going to operate the way I expect it to operate?
Is it going to work in the field the way I want it to? To do that, you can't do that then with just mechanical components. You need software, electronics. You need manufacturing engineering, automation, how you're going to build this in the factory. You need all those pieces. Why do you need those pieces? Because the value that we create is the value of linking the real to the digital. The closer I can make that relationship for the customer, the more value it brings. Because think about it, now they can move very quickly and make decisions in confidence. All that complexity we spoke about, think how fast they can move through that when they're using Siemens Xcelerator and their competition is not. So truly a competitive advantage in the way they would work.
The next part of this is what we talk about with lifecycle intelligence. For AI to really work, its power depends on trusted data. This data has got to be accurate, it's got to be available, it's got to be well managed. Teamcenter, our proven PLM product, leading product in this space, is the place our customers go for that trusted data. That's where they go. And it's also kind of the core of the data fabric of what we talk about and the investment we're making here, which Vasi will talk about later and Roland referenced earlier today. Vasi, sorry, we'll reference later and what Roland talked about earlier. We also have adaptive operations. What do I mean by this?
Oftentimes software companies will create a solution for a small, medium enterprise, and when that enterprise grows, they make them switch to a different tool because it's not scalable, can't grow. Or what they have on the cloud is different than the product that they have on the desktop. We took a much different approach here. We did a lot of investment because we don't want our customers to have to go through some revolution every time they want to move from small to large, cloud to the desktop. We have one seamless transition we use. You can start with us any level and we have the data compatibility all the way through whether you're in the cloud, the desktop, small to large. And I'll talk about that a little bit later as what that means to our customers. So how did we do this?
For the last 15 years, we've had a clear game plan, like a series of mergers and acquisitions to kind of build this out. Now this is overall Siemens, all of software that we've done. We did this to really create this leadership in industrial software. But I'd like to take a few minutes just to talk about what some of those expansion areas mean for our customer. As I mentioned before, electronic systems are growing rapidly with our customer. You've got this convergence of mechanical design, electronics, software all coming together. And as AI is being put into more and more products, it's driving that process even faster in the way our customers work and the way they think. And so how do we help in this transformation? Teamcenter, the most proven, most scalable PDM tool in the world, is now expanding.
Because we're taking it from discrete manufacturing where we've had that presence for decades with Teamcenter and now moving it into the EDA space. And so now what's happening is you have companies like Micron, Intel, new startups like Rapidus in Japan that are using Teamcenter for semiconductor lifecycle data management. So the same value we provided for years in the discrete space, we're taking into the EDA space and having really good success with this. What it's helping our customers do is move a lot faster through their R&D process and cut their R&D costs. They've got a managed process now, just like we did with discrete manufacturing over the last couple of decades. Another key growth area is purpose-built semiconductors. You know, if you really look at most of the world's phones, watches, smartwatches, the vast majority are built with our software.
Design, manufacture, built with our software. But these customers, as well as many other industries right now, are having to really optimize their products. They optimize this by building their own semiconductors. So what that does for us, that drives usage of our software. In our EDA space, we have two products, Tessent and Calibre. They are the gold standard of doing the validation of what you do in integrated circuit manufacturing and production. So that's helping us grow that business because you now have customers advancing and growing that market by building their own integrated circuits. Another key growth area for us is the idea of hardware-software compatibility. This is a big challenge for our customers. When I talk to them, it's an integration nightmare for them. And what's happening is you can constantly keep changing the software very late in the product development cycle.
The problem with that is then you assume the software is the issue. Well, it's not the software generally, in most cases. It's the software, but it's how it interacts with the processor, the network, the rest of the digital twin. How do you make sure this thing's going to work and really get it right? And so our playbook that we've executed was specifically built to address these kinds of complex problems. Again, moving through that complexity a lot easier than anyone else. And you know, when others couldn't see the potential, they couldn't see the potential of bringing together EDA and PLM, we did. And now the rest of the competition is trying to catch up with what we've built years ago. Another place we're leading is industrial AI. We intend to define this market and what it really means for our customers.
We're applying it to design and simulation tasks. And if you look at these turbine blades, for example, or this turbine blade, this thing operates in extreme conditions. It's operating in an environment that's above its melting point. So think about it, it's operating in a place above its melting point. So you've got to cool this thing very efficiently. And for years, customers have used our product to do simulation of these turbine blades. The problem is you do one, it's very timely, very costly to build that out. Using AI now with Simcenter, our CAE solutions for simulation, we can do this much more rapidly. We can do a thousand blades at the time you were doing one previously. Now think about that. If your competition is doing a thousand blades in the time you do one, how long before you're disrupted?
So you can see why this is such an attractive market for us. Customers are all over this wanting to know how they can move faster in this space. The other thing we did was we combined the CAE tools from Altair with our existing tools. We're expanding the market. Here's how we're expanding the market. Most times, those kinds of problems were reserved for the PhDs, the experts in the company. With the tools we have now, we can take them and apply them to every engineer. And they're able to do this simulation and move much faster with what they're doing. So again, expanding our market as we go. Another example is Rolls-Royce. They've been a customer of ours for years designing their aircraft engines with us. But like many large companies, they've got data all over the place.
Data in silos, it was hard to find the data, hard to leverage it. So we worked with them. We used Teamcenter Xcelerator to set up the environment for them, the digital enterprise, to be able to manage this data. So now we can look at the engineering functional data, the manufacturing data. We can look at the performance engineering data. We put that in a closed-loop process. So now all of that information can feed back to the design and the manufacturing process for continuous improvement. But here's what's really cool about that now. If you've got the data that we provide through all of our PLM tools, my colleagues in automation run all their factories as well. So we've got the information around the PLCs, the sensors, the edge devices, and we can take all that data and bring it together.
Once we have that data, we can start doing things with agentic AI, where I can say, look, I need you to summarize the status of the turbine blade production that's going on right now. It analyzes what's going on, sees the issues that are there, automatically identifies the issues, maps them to maintenance records, all the other information we have in Teamcenter, for example, and it comes back with a proposed fix for you, and then it can actually create the maintenance ticket, drive the whole process all the way through. This is where it really gets exciting for our customers when we talk about what could happen with AI, and we're doing this already in many of our accounts, and we're really just at the start of where we need to be.
Now if we combine this information with Altair and their RapidMiner capability for data science, you can see how we can start leading the way in industrial automation, in industrial AI, all the work we can do with our software, this is how we lead the way. So these are just a few examples of some of the things we're doing with our customers and the advantage they can use by using complexity as a competitive advantage. So four years ago, I stood up in front of you and I explained that we're making the SaaS transformation. I'm proud to talk about where we are right now, although I think we've covered it between Roland and Ralf. They talked a lot about this, so I'll go through fairly quickly. But look, we said we're going to be at a 10% ARR through the period.
You can see we're well above that. We continue to do very well with the process. As Ralf said, we're committed to 40% cloud ARR. We'll be very close to 50%, so we're one year ahead of schedule on that. But the one thing that wasn't talked about, I think is also extremely important, is we had to go to every single customer and open up the contract to redo the contracts to deal in a SaaS environment. And any of you knowing the sales process, when you open up that contract, you start all kinds of new discussions. We've done that. And so we talk about being 80% or whatever through this transition. A big part of that was getting our contracts set up with our customers.
Now that we've done that, it really sets us up for that high-performing growth that was referenced earlier in the day about some of the things we see in our transformation. We also had a transformation that occurs when we talk about recurring revenue. To be fair, even going into this, we had a large amount of our revenue that was subscription-based. And for those of you that understand revenue recognition and subscription, you can still recognize a big portion, a fair amount upfront, and then the rest is ratable. But we keep working on this. What we've done is to make sure that we can make more of our revenue recurring. We did that in two ways. One is we reduced and eliminated most of the perpetual license software contracts to go to these SaaS recurring revenue models.
The second thing we did was steadily reduce our services as a percentage of revenue. We want to make sure that we don't keep growing the service business, and we've kept that very flat, even though all of our other growth, as you saw, is well above in the double-digit range. So we expect our recurring revenue to remain above 80% as we go forward. We also set a key metric for us was to grow our small, medium enterprise business. 87% of our new SaaS customers are coming through small, medium enterprise. We've been very, very good for years putting Teamcenter into large accounts. Now we can take Teamcenter X into the smallest of accounts and be able to have a path for them to grow as they go.
Companies like Cobot, that's one of the reasons they select us is we see this growth path that you've got, and I can't stress this enough, as what I said earlier, cloud to desktop, small to large, seamless growth path. That's a unique capability in this industry, and we spent a lot of time and effort to make sure we did that. We didn't go buy a bunch of cloud companies just to have a cloud solution and then not have any compatibility with the rest of our portfolio. We worked very hard to make sure we've got that aligned, and then as far as our overall revenue, you can see where we were in FY20. We have good growth to get to where we are in FY25. However, you have to realize we did that in the middle of this SaaS transformation.
That had a little bit of a negative effect, obviously, on the overall growth. Really, we're coming out of that because, as most of you know, it's really just a mathematical model, really, as you come out of this when you get through the transformation in SaaS. But with that, you can see that we just crossed over the EUR 6 billion revenue mark in software. Still roughly a third of our business is EDA. The other thing that's quite interesting, well, first of all, this also only includes six months of Altair and three months of Dotmatics in these numbers. So you can see that'll advance as we go forward. But the simulation portfolio that we have now together with what we had before with Altair and our own pieces is generating over EUR 1 billion annually for us.
You can see it really underscores the impact of the moves that we've made in a very, very attractive market. Speaking of these acquisitions, since I've been in this job, I think we're close to 40 acquisitions that I've done in the job. Now, some of these are extremely small, tuck-ins, few people, whatever, to build out that portfolio. But others are larger. Overall, the vast majority are green, which is almost unheard of in the software space. We've done a really good progress in these acquisitions, and the same thing with Altair. Altair right now, we're right on the game plan that we have to live up to and how we measure ourselves on this acquisition. We're actually ahead of the plan on cost synergies, well ahead in time in what we've done in execution here.
But the more important part of this is our customers are excited about what we're doing. They see this integration. We're winning new business in our existing accounts, and we're winning new accounts because they see the vision of what we put together. Same thing with Dotmatics. We're very early, but so far right on the plan that we said we would do. And the same thing here. It's really this idea of this comprehensive digital twin, as Roland mentioned earlier, going from the initial drug discovery design to production. Our customers are excited about that. And we have no intention of slowing down with the growth that we've talked about here. We want to take more than our fair share of the digital market expansion that Roland referenced earlier.
The reason I say it's so important with Dotmatics, just like the market and our customers underestimated what it meant to bring PLM and EDA together, we think there's the same opportunity with Dotmatics. And why do we say that? First of all, this isn't new to us, this space. We've done medical device design with my software for years. We run the manufacturing execution software in many of these companies. My colleagues run the automation for these factories as well. And so when we see that, we see a market that looks a lot like discrete manufacturing did 20 years ago, meaning unintegrated applications, not an enterprise layer like we have with Teamcenter. And that's exactly our opportunity, is to integrate, keep building this, and build that enterprise layer like we did so successfully with Teamcenter. And we can bring that forward.
So we see a lot of opportunity not only for organic growth, but also, as Roland mentioned earlier, the idea of additional acquisitions that could help build out this space for us. Attractive market, we're excited about it. And I'm going to let my colleagues, Axel and Suzanne, come up and tell us a little bit more. So thank you.
Before we start the presentation, I would like to start with a question to all of you. How many doses of medicine are consumed every day globally? The answer is about 3.8 billion. And this number is only rising. And the Food and Drug Administration reports a 40% increase in novel therapies. And that drives rapid innovation and investment.
That is one of the reasons why we are today excited to share the insights in this dynamic market and this world of life science with you and what it means for us. As Tony pointed out, life science is not a new territory for us. It is where pharma works hand in hand with biotech, with diagnostics, and also with medical devices. We have a major footprint in this market as all of those major players work with our technology. Building a bridge on the presentation of Roland, nine out of 12 top players in that industry work with both Smart Infrastructure and Digital Industries. We have partnerships, for example, with Merck and Darmstadt. We have just extended our partnership of innovation with them. This partnership is an evolution that goes back for more than 100 years in the cooperation of the two companies.
Our reach goes beyond big pharma. We work with leading labs, with academia, with small and medium enterprises, and with research institutes, and all of them are connected and enabled through the Siemens Xcelerator platform, so what do we want to share with you today? What this attractive market means for us? We want to present our unique selling proposition and how we combine our solutions with electrification, automation, and digitalization, and we want to discuss how we enable data and AI, how this enables also the growth in this industry. Let us start with the market and give us some insights, Suzanne. Thanks a lot. Yeah, so the life sciences market is growing fast and innovating globally. If you look at the market, it's made up of two portions.
One of them is around the electrification automation piece that contributes with about 5% growth, while the overall market is growing about 9% CAGR over the next five years. With the acquisition of Dotmatics, we've been able to significantly increase our addressable market in the industrial software space, and that portion we see growing with about 12% CAGR over the next five years. Now, what's driving that growth? One, of course, globally, the aging population, which relies more and more on medical treatments, but also the living standards in developing countries that also demand access to reliable therapies. Biotech is becoming more relevant, and this is also why you will see some of the large pharmaceutical companies either enter strategic partnerships with these companies or actually acquire them, especially ones with promising technologies and promising pipelines. However, we've also seen some geopolitical uncertainties coming up.
One of them, the recently imposed tariffs that you've seen, but also because of the overall supply chain issues that we've seen and disruptions that drive for local value chains, we see large investments coming in the U.S. and also in India and some other countries in Asia as well. On top of this, digital transformation is playing an increasingly important role where companies are harnessing now the amount of data that you just saw before and AI to gain a competitive advantage. So overall, we're in a very interesting growing market. Yes, Suzanne, we are. It is an attractive market.
But if we would talk to our customers, they would describe it as the race against time because the journey from finding the next molecule to research and then development and putting the next product through the qualification and then doing the tech transfer and ramping up the production of that new product can easily take up to 10 years. Due to the very high innovation rate that we just see, the ROI of the R&D has been declining from over six to 1.2% lately. And the development of those assets is very, very costly. The average cost of a pharma asset is $2.2 billion. And it is risky because just a single day's delay of a product launch of a new product can cause a loss in turnover for up to $1.4 million. And the value of a single batch that those customers produce is up to $50 million.
It is crucial that we support our customers with those challenges, that we help them to overcome, to optimize their processes, to be faster in finding the next molecule and bringing it into clinical testing in the R&D, helping them in the tech transfer and scale up in the production to help them to become more agile, more adaptable, and master the challenges and take advantage of the opportunity provided by data, AI, and software. Suzanne, how are we going to support our customers to overcome those challenges?
Well, first of all, by combining the real and digital worlds. It's been a theme today, hasn't it? We integrate hardware, software, and services. Our portfolio spans electrification, automation, and digitalization. Let me share two real customers' examples with you that illustrate this quite nicely.
For Pfizer, we did on-site integration of power distribution, building automation, HVAC, so very much the real world. On the management layer, we combined the information coming from buildings, from infrastructure, and from production so that the data could be shared between the production and visualization systems. This resulted in faster, more flexible production, and in the end, saved 40% of energy in comparison to a traditional plant. With Merck, the plug-and-produce principle, MTP, helped us standardize so-called modules that afterwards, when you set up a new production line, you basically put them together and you gain flexibility and speed, and overall, it takes us 80% less engineering effort doing this new setup, but we don't only serve these large pharma customers. Actually, our offering, our technology offering combined with financing solutions is ideal for addressing a large part of the market, including also small and medium enterprises.
With that possibility, they can then, first of all, grow their business, but also gain better access to Siemens technology. Now, what unites all customers in this whole market is the race against time and also the drive for operational efficiency. Our customers need to accelerate in all the phases. You heard before the race against time. And to achieve this, data is key. Just imagine being in one of these companies. All the sensors, all the actuators, all the controllers, all the machineries, all of them generate data by the minute, by the second, by the millisecond. This is the foundation for the data fabric we spoke about today. The data fabric is the foundation for building AI and simulations to actually support our customers and unlock significant value. Now, let's dive deeper into these three phases, as we call them.
One of them, design and development, then smart production, and afterwards also equipment engineering that you better understand how these actually benefit these three phases. Let me start off with design and development, where experiments play an important role. And experiments take place in labs. And what do labs need to be? They need to be safe. They need to be reliable, and they need to fulfill regulatory compliance. Otherwise, you can throw out a whole batch or a whole test series if you fail there. So what's important is controlling humidity, airflow, temperature, power quality. That's all key for this compliance adherence. And by the way, all of that is also linked to energy consumption.
Thanks to digital twins and AI-enabled simulations of fluid dynamics, we have been able to optimize our ventilation efficiency by 45% and also, by the way, take out contaminants a lot quicker than other systems. Labs also need to be flexible because all of a sudden they want to test something else, and then you need to configure this. We can also do this now 80% faster by combining our room automation with our Siemens Xcelerator partner that is specialized in modular labs. The number of experiments and raw material usage are also key. They can be reduced by about 30% by leveraging our digital twin and AI simulation. We've actually proven this with our customer, J&J Innovative Medicine, by building a digital process twin based on gPROMS.
And last but not least, Dotmatics Luma is actually targeted toward scientists because you want to make them also more efficient when they're in this discovery phase, so Dotmatics Luma is an R&D platform, actually a data platform that collects data from various sources. It cleanses it. It also analyzes it, and it makes it available to a larger population there who actually works on this, so it helps scientists become a lot quicker in that discovery phase.
Thanks a lot, Suzanne. What I really liked about your presentation was the combination of the real and digital world and the example of that bioreactor because actually we have an example of that bioreactor. We have a small bioreactor in our demo out there, and we have a small real-world example with automation and instrumentation around it, and then we also have it digitalized in the digital world.
With that, we can show all the use cases that we are presenting today over there as well. Please use the breaks later on to take a look at that. Taking those digital twins from the design and the development to accelerate the transformation to smart production and manufacturing is important. Siemens supports that with our leading solutions in electrification, automation, and digitalization as well. From designing a process and the plant to run that process, we use those digital twins for the virtual commissioning to speed up the ramp-up and the tech transfer. We use those twins for the operational simulation and optimizations of the plant to enable our customers' efficiency, transformation, increase the resilience, and the flexible production. To give you an example of what happens after the plant is running in the engineering, that is the digital recipe transformation.
Just imagine a recipe, a process completely digitalized, easy to adopt and manage the production recipes entirely digitally, making it easy to scale them, to update them, and to transfer the processes across sites and across regions. Let us take a look at how that shows in production. We are using those digital twin solutions already today, and we combine them with AI technology and helping to streamline workflows and reduce time to market by up to 50%. We have done that with our customer, Novo Nordisk. Many of our customers are already today using our enterprise batch recording solutions that enables them efficient paperless production that leads to faster deployment of the recipes, lower operational cost, and greater responsiveness to changes. We're not doing that alone. We're also doing that together with our partners.
One of the examples is how together with Capgemini, we have been helping Sanofi to standardize their production process and accelerating the rollout of 52 manufacturing systems worldwide. What we did in that example is there were a lot of documentation of recipes and work instructions in paper. We were replacing paper-based batch records with digital ones, reducing review times by 70% and cutting deviations by 80%. This is setting a new standard benchmark for efficiency, quality, and compliance for Sanofi, but also for the industry. Now, I've been speaking a lot about the primary process, how we run the recipes and how to produce the medicine. We have to look also at the secondary part because in the secondary part, the life science industry needs machines and production lines for tableting, for blisters, for filling of doses. This is provided by machine builders.
Those machine builders use digital twins also to help accelerate equipment engineering because they are important partners of this ecosystem. Let's take a look at how we do that. We are also leveraging digital twins to reduce project times by 15%. We're simulating and digitalizing consumables and equipment. Smart data processing with industrial edge running on the shop floor can help those customers to reduce downtimes by 30%. The digital twins we are creating can also be used for realistic training environments and accelerate the engineering process even for non-experts. This increases the efficiency and the resilience of the production and manufacturing process. AI is already boosting the flexibility and enabling adaptive and autonomous production. This brings us to the next big thing, AI, in just a moment. Let us summarize.
We have shown so far how we enable our customers to become digital enterprises, supporting the research and development phase, the tech transfer, and the ramp-up of the production, and the optimization of the production and manufacturing itself over the entire life cycle. Now, how can we use data and AI to even further improve that and bring that to the next level? And Siemens is already offering today AI-based solutions based, for example, chatbots to capture and analyze and contextualize complex content with the help of GenAI-powered assistants. For example, we are offering industrial copilots and operations that empower the shop floor teams with intelligent diagnostics by integrating the IoT sensor data with machine documentation into an intuitive chatbot interface, minimizing downtimes through rapid root cause analysis and real-time resource optimization. Another example is the recipe builder.
GenAI streamlines MES recipes, creating and analyzing existing documentation and generating preliminary drafts, enabling non-experts to create electronic work instructions intuitively. This accelerates the MES adoption by up to 85% faster instruction generation and 25% faster deployment across global sites. Robust AI starts with clear contextualized data and strong focus on cybersecurity. We support our customers in laying the secure data foundation for AI and enablement and the data fabric. For example, mobile work options are growing, requiring secure remote access to data and systems, and zero trust principles ensure this access without altering network infrastructure using zero trust OT access services, so data and AI and cybersecurity offerings are part of our DNA already.
Now, with Dotmatics, we're entering even in a new market segment in the early process of design, helping our customers to make better choices and become faster to the molecules that they want to follow up with. This is expanding our leading position by extending AI-powered PLM portfolio into life sciences. It is building on Tony's presentation about the game plan and about the playbook that he presented. We will create an end-to-end digital thread for life sciences too. We already started that. At the beginning of the presentation, I was elaborating about the innovation partnership with Merck. This is exactly where we extend our partnership, building on the Xcelerator platform with Dotmatics Luma for our customers like Merck. Life sciences is a highly transformative market with a lot of opportunities.
Data and AI are the game changers for our customers to win the race against time, to shorten the time from patent to patient. With Dotmatics in the mix, here is how it all comes together for Siemens in life science. Suzanne, let's sum it up.
Sure, let's do that. We are in the sweet spot of a very attractive market, growing about 9% CAGR over the next five years, with the software portion growing double digit. We are unique in that we combine the real and digital worlds, and we have the most complete portfolio spanning electrification, automation, and digitalization. We have deep domain expertise. Again, we have been in this market for more than a century. We are really well positioned to help our customers accelerate innovation across all the life cycle. Siemens' offering is powered by AI and by data.
Also here, we're uniquely positioned. And this results in faster time to market and quicker innovation cycles, less energy consumption for our customers. These are all very important ingredients to drive accelerated growth in this market going forward. Thank you very much.
We'll now have a short 10-minute Q&A session, and therefore we would request to limit yourself to one question if that's possible. I see your hand in the first row. Just push the button.
Hi, it's Ben from Bank of America. Can I ask on software? We had the guidance of 15% digital growth over the next five years. Where does the software business fit within that? All right. Like I said, we expect to be right there with the digital growth and take our fair share of that. So we feel like we're well into the double-digit growth phase.
You saw it already with where we're at 13% ARR, which at the moment, ARR is probably the best metric to use because, as you know, when you're going through that SaaS transformation, the revenue thing is a little bit more difficult. As you come out of that, then revenue and ARR start to line up a little bit better for you where you're at. So we still feel very comfortable with being in that double-digit growth range.
And do you think you can do that ex-M&A? Is that an organic? Is that how we should think about that? Yeah. I mean, if you really look at what we've done in ARR growth, that was exclusively. I mean, we didn't have a lot of M&A through that period, right?
I mean, we had small tuck-ins and things that I referenced, but a very small portion of that was what was happening with Altair and with Dotmatics. So we feel good about that with our full portfolio.
Maybe in that row, second row, Andre?
Hey, it's Andre from UBS. I just wondered if you could talk about the positioning in life sciences and biopharma specifically now that you've acquired Dotmatics and what you've already developed with Teamcenter. Have you got the kind of full stack covered, or are there any white spaces there above or in between? And can you just help us with the lay of the land there? Where are you in this market? Are you a top three player now in the software for these applications? Maybe who are the key peers who we should track to kind of understand the space a bit better?
Yeah, it's a little bit. I mean, you got to define the market, first of all, where you're at, right? And so we've talked about being on the drug discovery on the front end. And really what Dotmatics did prior to the acquisition was a series of small acquisitions that built up their portfolio. And they would call these products resume builders. What they mean by that is every scientist in the world wants those products on their resume. And so those products are selling extremely well for them and have been selling very well for them. The part that's being developed is the enterprise level, which we call the reference Luma before.
That's still early, but we feel like our knowledge and the expertise we have of what we do with Teamcenter at an enterprise level can be applied to what we need to know about what we do with Luma. So if part of your first question is, have we got the full stack? On drug discovery, there's still other pieces you can plug in there. And that's what I referenced earlier, saying that if we think about inorganic growth, it's one of the areas where we're not really locked. There's lots of different places we can still go there, right?
And then, on the back end, when we talk about manufacturing, our colleagues with what Axel's done and so forth, we've done a lot of work there, but I think there's more we can do there as well when you think about some of the things in operational software and how you run those factories. Now, there's some stuff in the middle, like medical records and things like this where maybe some of our other competitors are gone. Not so sure that I need that. We need that for the digital twin approach that we're talking about. We're more in the engineering aspects of this and the manufacturing aspects of it. And so there are still some pieces we can build out.
We've got a lot of upside potential on the enterprise level because at enterprise level, if you get that right, that's where you can start applying the AI capabilities that are so important as well. And so we feel like two areas. One is still building out some of the science drug discovery side of this, and then also what do we do to build out that enterprise level as we go forward.
See your hand over there. Max?
Thank you, Max from Morgan Stanley. My question is, how would you rate your EDA business relative to some of the large competitors like Synopsys and Cadence? What are your strengths? What are your weaknesses? And how do you really kind of push the business to get in with that kind of really top tier of semis developers?
Yeah, it's a good question.
I think, look, when we did the acquisition of Mentor Graphics, they were always a number three player in this space, right? And sometimes I'm asked, are you losing market share? Look, the products that we have that I mentioned, Calibre and Tessent, these are the gold standard. If TSMC is going to go into production, you've got to be validated through those products. And so that we continue to do extremely well in. The difference is that market's a little bit smaller than where our two competitors have their strength, which is on the actual front-end design of the integrated circuit. That's a bigger market. So therefore, if they're both growing, you might have a little bit more growth absolute dollars on that space, right? Having said that, they're going to try to come after us on our strengths, and we're going after them on their strengths.
In this place and route, the design side, we've done a series of tuck-ins and other things we've done, and we're doing very well there right now with a number of companies evaluating our tool to see could this be part of what they put in their portfolio. We're still early in that, but we expect to make some progress there, just like our competition is going to try to come after us in Calibre and Tessent and where our strengths are. We're very, very good on the validation, the prove-out, is this thing going to be able to be manufactured, all those types of things. We do that, and we've always been strong there. They're a little bit better on the design side at the moment. We hope to come after that.
And then, as you've seen what they both have done now, the competitors are trying to copy what we've done in the EDA to PLM integration. So while what they've done is acquiring CAE capabilities, Ansys or Beta, whatever you need to do to fit Hexagon in the case of one of the competitors, is they're trying to fill out that digital twin portfolio. I was joking the other day that when I started first talking about digital twin with the EDA space, customers looked at us and said, "What are you talking about?" Had no idea. And now everybody's talking about it. And when we first said that we see the idea of the integrated circuit being influenced by the product, people said, "What are you talking about?" We build the integrated circuit, then we look for the product.
One of our largest customers in the United States that does a lot of stuff with phones and smartwatches and all these things, one of their executives a year or so stood up and he said, "You know, I thought this was interesting. Design the integrated circuit and you look for a product. Why are we doing that? Why aren't we doing them both at the same time? This is why these acquisitions are being made, right, to bring us together." Our advantage in that is they don't have a Teamcenter. They don't have a 3D modeling capability. They don't have the manufacturing simulation capability, and they sure as heck don't have my colleagues in the automation side as well to bring that together. That's why we think it's so important that you can't just pick and choose. You've got to have somewhat of a comprehensive digital twin.
You can see their strengths today, our strengths today, and where we see going forward,
and maybe just a very quick follow-up. Obviously, you've been through the SaaS transition in your PLM business. Why would you not need to do that in your EDA business as well, and what makes that different to not have to do that?
That's a good question. I think we made the decision four years ago with the board and everything to do it. And part of it was the cloud penetration on the EDA side wasn't as strong at the time because when you think about the thread, the flow they go through, it's a lot of homegrown pieces plugged into this as well. It gets very specific to their environment. And so it wasn't the same push. Today, we're seeing a lot more cloud. We're doing a lot.
And we think we can do this in maybe more of a step-by-step approach than a big swallowing of this fish that we went through. So what we're trying to do is find ways to gradually bring more of these capabilities into our accounts and start gradually changing some of the contracts that are a little bit more, I would say, less lumpier as what Ralph's talking about. So we're not there yet, but we're trying maybe a different approach that allows us to do this gradually rather than a big approach.
Any questions from the press? Just asking. Good. Then we continue with the analysts. Gail, third row.
Yes, good afternoon. Thanks for the time here. You've talked a lot about the benefits from AI.
Could you talk a bit more about the potential risks, the potential disruption that AI could do to the SaaS business model with AI agents potentially replacing some of the software users, with coding becoming more and more automated, and maybe with software becoming more commoditized to a degree, leading to some potential price deflation?
Yeah, I think, look, there's a benefit if we can write software faster and do all those things like any company, and we surely are taking advantage of that already, and we'll continue to take advantage of that. I would say on the risk side, depending on what parts of our software, operational software, where you're running the factory and you're doing the things, it's got to be right all the time. You can't have hallucination effects, right? So you got to make sure you've got that right.
On the other side, though, how we leverage the data for what we do, like you saw the examples with simulation and these types of things, that's proving out to be very, very beneficial for our customers. And there's some that argue, they say, "Well, we can just use data in a data lake, and we solve every problem known to mankind with this." And there's some value in that.
For example, in a data lake, if I've got a component that is failing or has an issue because of certain environmental conditions, like it only fails in certain parts of the world at this temperature condition, or I've got handling equipment on a plane that I know that particular group that maintains that plane is more rough or difficult than the other group over here, they can take that kind of data and they can analyze and have some results out of it. But when you have to know a subtle manufacturing defect that's tied to a revision of a particular component and how it interacts with all the other things, you need physics-based models on the back of this to be able to show and evaluate. So I think there's a way to bring those worlds together.
I don't know the environmental conditions of a plane running around the globe because I don't track that data, but I can take that data aligned with all the data that we have where we know all the engineering context of what's there. And I think that's sometimes missing a lot in these discussions is it's not about just looking at data by itself. You got to be able to simulate and understand any minor changes can have a major impact in how you think about that design, right? And so I think that's missing sometimes when people talk about that. So we see the real approach is to bring these worlds together, to use data from both sides to help our customers move faster. And then there's areas like we talk about where you got to be really rock solid.
If you're actually running something, you can't have any variations. It's got to be pretty rock solid. But we have a lot of interest. I mean, every customer I go to, it's one of the first things we talk about right now is what can you do with our tools, with our data, how do we take advantage of it? And there's low-hanging things you can do right away, and then there's other things that get more complex, but we're looking at all those right now. Along with V, I mean, that's part of what he'll talk about in this presentation, how we're going to build out this Data Fabric continuously to help us with our customers to solve those kind of problems.
Hello, everyone. I'm the Chief Operating Officer for the automation business. Any question? Are you in touch with automation?
Probably not always, but probably the shirt you're wearing was produced with automation. And probably, I don't know how many of you have a coffee cup. The coffee cup the ladies just brought have been produced with automation. And if you have a coffee, probably the coffee beans in here have been roasted with automation. And there's a very high probability that this automation was delivered by Siemens. Every third machine, every third process plant, every third line is automated by Siemens. What I want to show you today is how we are expanding from this strong market position, number one, how we are using new technology to take first place in the next wave of automation. And at the end, how we are using the data which are generated inside our controllers as part of the Siemens Data Fabric in order to generate value with AI.
Automation is a very important aspect. In the past, you automated because you need a higher productivity and higher quality, faster time to market. In the meanwhile, automation is very fundamental to solve really significant topics of our customers. A lack of labor is hindering our customers to expand, or if they have labor, a customer told me, "I have a fluctuation of 20% every year. So I need to hire labor. They are unskilled. So how I can maybe use AI to empower those workers working in a plant?" Secondly, you have geopolitical tension. You have supply chain interruption, so a lot of production moves more closer to the consumer. Now, if you do this, you want to build state-of-the-art factories which are fully digitalized, fully automated.
We believe these factories need to be even more flexible to react on market demand, or maybe in the future, even more autonomous. Last but not least, without automation and electrification, there's not a big contribution to sustainability. Electrification will be a major lever to move sustainability further. At the end, we will not think only linear. We think circular. With that, we are delivering to our customers, for example, a digital product passport where they can have a life cycle tracking of their product. We also repair our product. So we need to think much more circular. The market is attractive. It's about EUR 80 billion. It's growing 4%-5% on average, but there are growth buckets. We have talked about them already. So Axel and Suzanne talked about the topic of life cycle.
If you've seen Tony on his presentation at Rolls-Royce, that was aerospace and defense. We see semiconductor electronics as a growing vertical, and as data centers will become more an AI fabric, probably you need automation to run them, and Ruth will talk about that. If you think about AI and you think that further, we believe there will be a big trend to unattended operations or maybe light-out factories, and in our opinion, that would even boost that market growth, which is already on that slide, so how we do that now, how are we expanding from a strong position, so first of all, what is our strong position? I mentioned it. We have many customers. We are by far market leader in factory automation, in motion control, a very strong position in the process industry, and we generate a lot of data.
We calculated that number for that event, and it was impressive. Even the number of data which is generated inside our controllers was a 10x compared to 10 years ago. Expanding on a strong position, I want to go on one country explicitly because it was also an ask this morning at Roland. What about China? China is number one manufacturing market by far. The question is now, how is Siemens doing in China? First of all, we are number one in China. And in China, you need to see two market segments. The performance segment, where we are by far number one in this performance segment, the customers are looking for an end-to-end solution around digital threads, combining the real and the digital world.
And we see a segment we call. Roland called it value for money, where customers say, "I want Siemens, and I'm even willing to pay a little bit more, but you need to be somehow in the price range of local competitors," which we're seeing rising up in China. That segment is a big market segment, and we are addressing this segment. We are addressing this segment with local value-add in our factories, in our complete R&D. So we have built it up significantly. We had already significant R&D in China. We built it up even more R&D. And one outcome was when Roland was showing the A-Team product we just launched in March this year. Others were following now in September. And I'll give you one world premiere now. In this segment, we are number one at the PLC side.
That segment of 40% value for money, we sell our; it's called an S7-200 Smart PLC. And in four weeks, we will launch this product. And this is the market leader on the market. This one will be the successor. Now, this one has a double performance, has 40% footprint, and this relates to the question of Roland this morning. So are you losing margin there? It also has a significantly better cost position than that one. So we are in China. We are having the right products. We are market leaders in that segment, and we are expanding our market position in that one. And what we built in here, for example, is very high-speed motion capabilities. And why we're doing this?
Because on the servo drive side, we are not number one in China in that segment, but in the combination of that product with our new servo drives, which, by the way, also now were brought on the market, we are very confident to expand into that market segment even further, and maybe something in the hand. By the way, it's even a little edgy, I know. And it's even modular compared to the other one, so you could do that. But anyway, I'm very excited about that product. Now, there is for sure now this value-for-money segment, and there are the big customers acting globally. And you know that customer, BYD. I don't want to talk a lot about that customer, but that customer sees Siemens as a strategic partner, as a strategic partner to expand their operation in China and globally.
They are all in with us. They're buying the entire suite from Tony. With that, they reduced their time to market by 25%. And they use our entire automation stack. With that company, only last fiscal year in the automation side, we did more than EUR 150 million R&D directly with the customer. The interesting thing is now this customer has a lot of line builders and machine builders. Give them another data point. The biggest press manufacturer in China, Jier, delivered 28 press lines into BYD last year. That was a volume for us, again, of EUR 160 million only with that one machine builder. So those customers are buying into us, and they're buying into us not because we have only singular good products, but because we deliver the end-to-end stack, combining the real and the digital world. And for the value for money, yes, we have also products.
And this is a machine builder which never used Siemens before. It's a glass machine builder, also a very important customer in China. They were with the competitors, but now having those kind of smart products, they decided they go with us. We sell our PLC, our drive, and on top, we sold some digitalization where we still have a very good differentiation also in China. So those customers go with us if we are meeting a certain price expectation. And we're seeing that growing more and more. So secondly now, so we're expanding on a strong position, and we want to, as a market leader, trailblaze into a new market. And this is kind of the technology stack, how you automate today. It's called Totally Integrated Automation. Siemens introduced that about 30 years ago, and this is how probably 99% of our customers are automating today.
What we are doing, we are expanding this TIA concept to software-defined automation. Software-defined automation where you, for example, have a virtual PLC. We have been the first one bringing this on the market and the first one getting a TÜV certificate for safety. What we're doing here as well, we're using data and AI of our industrial edge system, which is coming out of the field, and use them for more analytics as part of a data fabric. Those products are market-leading, and you can see a lot of analysts' report, technical analysts' report. Here, for example, we just received in September from UL. UL is kind of the German TÜV. It's a very important certification body in the US. We received as the first company and the only company a Platinum certificate for our VPLC and for our industrial edge.
The PAC report sees us clearly leading in industrial edge. They are recommending us under the aspect of cybersecurity, of usability, of functionality, and of the topic around connectivity. Now, what we do with these new products, we want to trailblaze into new markets, and the second biggest market in the world is the U.S. Yes, we are not number one in the U.S., and I think we have a fair chance to grow our market share. Why are we doing this? Because we are combining the real and digital world. Tony is by far number one with the software in the U.S. We are combining this, and we are bringing new technology in the market, and we see customers very interested in this new technology nobody else, even at a local U.S. competitor, can deliver, so we want to grow in the U.S. with a new technology.
Ben, we also know having a great PLC, no customer in the U.S. will switch because the inertia to switch is too high. But offering differentiating use cases, showing the customer value, combining the real and the digital world around digital threads, that makes a difference, and that makes us unique, and that makes a customer to decide for Siemens. We're doing that, again, on certain verticals with a certain focus. But we're also doing that as part of Xcelerator for small and medium enterprises, where we build bundles which we sell via the marketplace. And last but not least, you've heard on the Hannover Fair agreement with Accenture. Just two weeks ago, there was an MOU together with Capgemini. We're working very closely with Deloitte because those global system integrators, they're very, very interested to work with Siemens to enter into the factory space.
Why they work with us? Because they use Tony's software already as an integration partner, and they see a great opportunity to expand their scope, expand their market. There are no other companies which can offer the right technology, combining that, having the best starting point. That's the reason why they want to work with us. Very dedicated accounts and being a frequent discussion currently. One big account, you see here is consumer packaged goods in the US, the biggest consumer packaged goods customer in the US. They looked at all the technologies. They were not a Siemens powerhouse, and they said, "What will Siemens offer is really the best, and I want to go with Siemens." They started with connectivity on edge.
They said, "I want to have for my recipe of a detergent, for example, to get it in production, that need to be faster." So they followed the digital thread of recipe transformation. Now they said, "I want to do now having edge installed high inline quality inspection. And you need to see that machine. A machine is producing 6,000 pieces per minute, 6,000 pieces per minute." And they do inline visual quality inspection using AI based on industrial edge. And they did it in one line, now in nine factories, and they roll it out globally. We increased our revenue at that customer last fiscal year by 50%. And now they talk with us, "How can you maybe help us moving into unattended operation? Because I want to have over three shift operation, one shift, the night shift being unattended.
Nobody is in the factory, and we control it remotely." And they work with us to go to that vision. That vision is very much based on AI because running unattended operations, you cannot do in a planned way because things happen, and you need to have intelligence to operate on unknown situations. And now AI comes into play, and that is ARC, which is our advisory group also from the U.S. And they have a champions quadrant. They just published this two weeks ago. And you see here where they see solution capability and experience around industrial AI on the automation space. And they see here a clear leading position for Siemens. And we are leading. We get this feedback from customer. We get this feedback by bringing product on the market. If you go around the corner here, you have the bioreactor here.
On the other side, you see the Copilot, how you generate code for a PLC. We are the first one on the market. We received in April the Hermes Award. It's a German innovation award for that product. And we are moving further. And Vasi will talk about that, how we move more into Agentic AI. And then the question is, what's next? Will we move from AI, from the digital world, also in the physical world? And that will be a game changer for automation. And then you can do something like this. Maybe you can produce in the future a very personalized product at the cost of mass production because you don't program a system anymore. You tell the system what needs to be done, and the system finds a way.
The automation system finds a way how to produce something completely automatic, connecting automation cells, very flexible, producing things which you couldn't produce in the past, automated, only manual. So the cost of mass production. And it's not only a dream. This is a video from the Hanover Fair. We showed how physical AI will look in reality. And with that, maybe the next time on a capital market day, that cup might be not a white cup. It might be a cup where you have an individual name printed on there with the same cost this cup was produced as a mass product. Thank you very much.
Yeah, welcome to the world of mobility. My name is Michael Peter. I started in 1991 with mobility, and for 10 years, I've witnessed how slowly that industry was innovating.
Not because we had worse engineers, but just because the safety requirement, environmental requirements are so high that once you have an homologation that costs you many millions, you just don't touch it anymore for the next 30 years. And then in 2000, I changed to what was then called Siemens Industrial Solutions. So I was basically the in-house internal system integrator to DI using their products to build factories in different areas of the world. And of course, the big buzzwords then were IoT technology, bringing IP into the industrial world. And we were so far behind in mobility because of the reasons I said. In 2010, I rejoined mobility.
Last time I talked to some of you probably was Capital Market Day four years ago, and I said, "These are the most exciting times in the market because we now bring in the latest technology into our portfolio," and when it used to be completely proprietary, now we're opening it up. And I think what I said exactly became true, and I would actually dare to say now are the more exciting times than the most exciting times, and let's have a look at the market, what that means. The market did exactly what we predicted. Global trends like population growth, like urbanization, give us a very clear prediction that travel volumes will double by 2050, and that was a time, of course, five years back when CO2 emission reduction was already a big item. Some cities were closed because of the diesel emission discussions.
So it was clear that cities and governments were going to look for solutions. And you see that here on the market, a very stable growth also predicted in the future. And you see the gray area, which is not the classic growth, but it's investment of cities like Frankfurt today replacing everything, or like countries, Egypt, looking how to transport the next generation. And when that happens, the second big consequence of that is that these customers will not look for the cheapest product. They will look for a product when we talk about infrastructure that gives you more capacity on the rail without drilling a new tunnel or putting new rail in a city where that is impossible or would be really, really expensive. So they will look to invest reasonably, but with focus on yield. The same on the rolling stock side.
When you build a fleet or you buy a fleet for a whole country, then you will not look at the cheapest purchasing price. It needs to be in the right dimension, but you look for energy consumption. You will look for low maintenance costs. You will look for the train being on the track as much time as possible. You will look for a digital experience, drawing people into that train because you can't force people by law to use a train. That's exactly what happened, and that's where our technology leadership strategy was built on. That's where we went through the foundational technology, as I will explain. The results you will see here, we gained 4% in market share, and that means from 13% to 17%, which is, to me, unbelievable in a slow-moving industry of mega projects or large projects.
It led to our backlog growing from EUR 24 billion to EUR 52 billion, and Ralf Thomas mentioned it was an increased profit, gross profit margin within that backlog because we sold technology. We didn't sell low spec. Revenue grew by 8% per year over the last five years, and we have a very stable prediction going forward because, of course, much of the upcoming revenue is already in our books today, and last but not least, we converted into profit and into cash. And that's really remarkable in an industry where the customer has all rights many times to withhold payments because these contracts are complex and give a lot of rights to the customers. But when the customer feels you have delivered what you promised, they do pay, and we have a cash conversion rate of 1.06 over the last years. Let's look at that strategy that I referred to.
And I think it all starts with the middle one on the lowest level, focus portfolio. We decided to play where technology can make a difference. Technology that we call internally foundational technologies. These are Siemens technologies coming from the broad knowledge pool and research pool of Siemens. And we said where this technology can make a difference in those four levers, when it can provide you more network capacity on infrastructure, where it can give you better booking rates for your seat booking service as an operator, where it can bring you more availability, then this is where we want to play. And these foundational technologies that are listed or made the difference in these levers, these go really from simulating the product itself, simulating how we manufacture it, about the product itself, the project being very digital. Many times we talk about a computer on wheels.
An ICE high-speed train has about 10,000 sensors. What can we do with this data if we have the right architecture? They go into, of course, IoT, as I mentioned before. But at the end of the day, what I didn't know four years ago is that AI will probably change everything. And what was our strategy then will be the foundation going forward to actually use AI in a way that our competitors cannot do it because they don't have the same access. If you do this right, you have a second big advantage. You can scale your factories. We are really proud that we have the least factories in the world from all of our competitors, substantially lower numbers. And that's because we specialize them. In Munich, we produce locomotives and only locomotives and the coach cars that belong to them.
And we produce them in a very, very high number so that we can invest in welding robots, that we can invest in building up a digital twin of what we manufacture, like a digital twin passport that then goes into the service of the train once we sold it. Yeah, and on the top, of course, I talked about the importance of data, of data architecture, our accelerator strategy that we launched about three years ago. When you do all of this, you will realize that proprietary projects with individual data architectures in every project really limit you in learning what your products do. So for us, it was very important, and that's when Accelerator came around to define architectures and make sure that we have access to the data that we produce, that our software is modular so we can innovate small modules.
We don't have these huge monolithic software things anymore. And that, in the end of the day, our target will be to bring that software into the cloud. And when you do that right, actually, one of the big effects that you have is that you stay in contact with the product. So when I look at what does this focus on portfolio mean, I've talked about rolling stock and infrastructure, but the beneficiaries, of course, customer service and software because they can now use that data. They can use that data to stay in touch with the product and to recommend customers what to do with the product. Or we have a much higher booking rate, for instance, on trains. And here, I want to tell you a little bit about why we're investing so heavily into our software business.
I think this is by far the underestimated lever in the transportation industry. An average long-distance train in Europe has a booking rate of seats of 50%-60%. The train leaves the station. The cost is exactly identical if you have full booking. Imagine a company that with the same cost can add 40% revenue, 30% income. What that means for your equation. Generally speaking, in Europe, it would mean you wouldn't have to subsidize tickets anymore because roughly every dollar on a metro ticket is subsidized by a dollar or by a euro. It would become a self-financing system. We achieve booking rates of 97% with our software because we know what's going on in the train. We can facilitate seat reservations. We can allow seats to be changed just before the train leaves, but we have transparency. For rolling stock, it had quite some big impacts.
We focus on rolling stock where we can sell what we call a platform to many customers with all the advantages that come with spare parts, identical, and so on and so forth, and I'll explain to you what a platform is later, but that means we can spend EUR 150 million on developing a new train type that is completely digitized that our competitors cannot do in the same way. It also meant that you will not see us bidding very much anymore on light rail vehicles or metro vehicles. If you check, almost nothing won by Siemens. What we do there is propulsion systems, the electronic part, because that's where we can still scale underneath.
But if the train has to look different in every city and it's every time units 20, you cannot spend €100 million on making it a digital product over the lifecycle and then actually build up a business model that we want to actually have an ecosystem for that. Yeah, and revenue on rail infrastructure is also very simple. We are the first ones and the biggest ones contributing to the development of global standards, be it ETCS on the mainline, be it CBTC on the cities. And we do that in a way that we can later on again develop modular software on it. And we focus on these having in mind already. We do want to bring them in the cloud, and I'll show you later where we are on that. And of course, we love combining all of this in a turnkey project.
A very good example is the Vectron, where we really flipped the market. Locomotives used to be a custom-designed thing for every customer. We developed it into a product and then actually into an ecosystem. What do I mean? We start by developing a locomotive with the idea it has to run in 20, 30 countries, so it becomes a modular locomotive. We already look at how can we produce it when we do this. If you go to Allach, you'll see the roof goes on last on our locomotives. What goes in is modules for the different voltage systems in Europe. In the future, we will have a battery module that can go in. When we're done with all of this, then we put the roof on. So it goes from designing with digital tools over manufacturing to be a modular product.
When that's the case, we can manufacture it in a really almost automotive-like style. Today, we have almost 2.5 locomotives being manufactured every day. And that, of course, allows us to use then DI technologies during the manufacturing process in a way that you usually cannot do when you build unique little solutions. Then everything becomes manual labor, just like in manufacture. Now, when you've done all of this, and this is where I said it starts flipping, the system starts flipping, you have a product that has a residual value. If a customer buys that product and he doesn't want to use it anymore after three years in this country, he can use the same product somewhere else. He can even change the bogey. The bogey is modular.
It can be for freight with a lower speed, higher torque, or it can be high speed for passenger service like FlixTrain is tending to do. So when that happened, actually, we had investors, leasing companies interested in buying our locomotives. We have sold today 2,800 locomotives. 2,200 have been delivered. 80% of them were sold to leasing companies. And we believe we are in an absolutely unique position that we sell these locomotives to leasing companies because then at the end, they can use them in many different applications. And what that means, again, we can finance it. We have sold over 1,000 of these locomotives with financing included from SFS. The word ecosystem that I use is particularly important because we stay in touch with this locomotive all the time.
We collect all the data of the locomotive, and we can tell the operator how to change its operational mode to reduce wear, to use it on different tracks also that the wear is even. We can tell them, "You should go to service." And by the way, the next service station is very close to you because when you have sold 2,800, you're the only one in Europe that can have service stations where the locomotive runs. They don't have to come to Allach. They can stay south of the Alps where all these freight corridors meet, and they can do it there. And what that means, in effect, actually at the end of the day, is maybe best explained by saying that this locomotive has a sales price, but the service cost over the lifecycle is over one time the sales price.
Two examples of what a digital product means, one from rolling stock and one from infrastructure. Rolling stock, the new design, your newly designed Velaro Novo. Siemens invented the locomotive, electric locomotive, 1876, over 105 years ago. So when I go to my engineers and say, "Let's do a new high-speed train," I would expect 5% improvement. But what's happening right now is that we have a real revolution using Siemens technologies. And again, that's not everybody in the market has that easy access tool and that we can use here that actually lead to 30% energy savings. I want to give you a little bit of an impression of how this is possible. A high-speed train has zero friction, basically. The bearings are so good. The steel contact on the wheel to the rail is almost zero. When you shut power off, it'll roll 150 km without stopping.
It's about weight reduction. It's about semiconductors, and it's about airflow, airflow being the biggest. All three of them are Siemens foundational technologies. And you see here, of course, the typical airflow simulation you would do. But I want to point you to this. We have developed a new propulsion system that builds so small that we now can actually close the underneath completely. So all you see is four wheels. So the air can flow so smoothly that in the end of the day, we have 30% energy reduction. And by the way, when you buy a high-speed train, one-third is purchasing price. Today, one-third is energy costs over the lifecycle, and one-third is maintenance. So if you save 30% energy, you should be allowed to be 20% higher in the purchasing price, which we are not.
Our cost position is very reasonable here, but it's a huge leverage for our operators. We still don't allow us to call this a platform until actually we know we can use it for many, many customers. And this is where this comes into play here. A platform of our trains means it's an empty tube. The customer, the operator can define what they want inside and what they want to operate. We have offered trains currently in a little bit of a wide body that have five seats. So they have more seats because there's no stairs and nothing than the double deck are competing against with us, even though it's a single deck. They have more seats than them. But you can also put a lounge car in here or a party car or a bar system so you can do whatever you want to do.
And that's when we start calling it a platform because you can use it for really low-cost competition. You can use it for high-end services, and it can be reused many, many times. Infrastructure. I also want to go a little bit in technology. What we have today in Europe is roughly 15,000 of these houses, each one filled up with relays and electronics because every 15 km you have an interlocking. And when a train drives from Munich to Berlin, you probably cross 30 times a border of this interlocking. You have a handover procedure and. So what we did in Norway, and I wouldn't call it evolutionary. I would call it half a revolution, is we do an IoT architecture. So our point machines 800 km away from Oslo are controlled from an interlocking that is actually in Oslo.
So now when that point machine has a problem because it puts snow together, we will realize that in Oslo. We will see that the power consumption goes up, and we can control it and service it from there. What it means for operations is that this train from now on can be scheduled as one train between Hamburg and Berlin if we have this technology. I read just a week ago in a newspaper that the operator in Norway has said they will in the future only need one-tenth of their operators. That's really the tip of the iceberg because this is not about saving the operators, but it shows you how much more efficient the scheduling will be.
It will increase the throughput on the line, and it will bring delay minutes down because every time you have a handover, of course, you can make a mistake. And on the right, this is the second part of what really is a revolution, removing that to the cloud, where you have 15,000 houses that are half the size of this room here, full with electronics. In the future, we will have one server in the cloud. And we have outside an example for you that actually is being used for a demonstration that we did in Singapore that Roland also mentioned. Now, where are we with this? On mainline, we have the interlocking in the cloud as a technology. Interlocking is simply avoiding green on green. We have the ETCS in the cloud, which is actually how do I schedule my trains?
When do I request this line must be reserved for me, which is both safety relevant? And we have the same on the mass transit, the interlocking and what there is called CBTC. And what we did in Singapore is we put all of it on one cloud server, and that's what we call Signaling X. Now you can schedule the whole line in one computer. You have access to all of this data. You have the CBTC on the same computer, and we have the OCS on the computer. And when you go out there, it's about this big, and you could probably run the whole city of Singapore, all the metro lines on that hardware. Where we are, we have sold it on mainline in passenger service in Aachen. We've sold it also in countrywide rollouts in Europe already. In the mass transit, we're introducing it right now.
Currently, we have almost 60% of our portfolio transferred, so it could run in the cloud tomorrow. Many times we're approaching our customers now instead of buying spare parts in Norway. Just give us a license for the DS3, and then you can buy your server wherever you want to buy your servers. That makes much more sense to the customer because they have increased functionality and efficiency of the system. Yeah, summarizing it real quick, it's all about focus for us. We play where our Siemens technologies give us an advantage over the competition, over the lifecycle, over energy consumptions, over throughput, capacity, and all that. That leads us to a manufacturing footprint that is unique in the industry. And when we grow and build a new factory, it will never be a copy.
It will be more modern, and it will take portfolio out of one factory and allocate it exclusively here. So also the old factory will be more specialized and better in the future. Maybe one thing I haven't talked about at all, sustainability that I want to end with because that is why our customers are investing. That's why the market is what it is. We calculated the carbon reduction that the portfolio that Siemens Mobility brings into the market over one year. And it's an unbelievable 100 million tons of CO2. It's one-sixth of the carbon emissions of Germany. So I calculated we have 43,000 employees times six. If we had 250,000 employees in Germany working on a portfolio as impacting on carbon reduction as Siemens Mobility, we would basically compensate the footprint of Germany.
This is what our industry can contribute to the targets of governments, of cities, and that's how our market works. Thank you very much.
We have 10 minutes of Q&A, and it would be great if you limit yourselves to one question. Let's start here. Ben, first row.
Questions for Rainer, and it's the never-ending issue and challenge of China competition. I completely take on board this moving down into the mid-market or the value segment. But what we're hearing and what I think is fair is we're seeing some of the Chinese companies, we all know the names, actually test upwards. So trying to move into large process PLCs, for example. First of all, is that something that you can confirm from your point of view? Yes, we are aware that XYZ companies are beginning to move up into this domain.
And if so, how do we think about it? We've seen it so many times, but I completely accept that the automation industry is not the same as the robotics industry. Every industry has got its nuance. So is this something that you're thinking about, and how should we think about it?
Yeah, first of all, I can confirm they are showing products which go in this direction. They haven't released it, by the way. And it's all that different if you show something and then you need to release it. And I know if you develop a redundant PLC, what effort needs to go in that it really runs smoothly because it needs to run redundant all the time because the customer is relying on that.
So first of all, yes, they're moving this direction, and they will also, because they're also moving with China's speed, they will be somewhere there, number one. Number two, I strongly believe we have a very competitive portfolio around that one. So I'm not fearing any competition as long as it's fair, number one. Number two, I mentioned that on our BYD example. The customer is not buying here one single product. The customer wants to buy an end-to-end solution, which is combining the real and the virtual build. And that's our value proposition. And if our product fits into that digital thread, we create more value than you have a single product which might be some percentage point cheaper. But yes, we are very, very diligently looking at what they're doing and in which direction they're going.
And we are willing, and we do, as market leader in China, to counteract accordingly.
Delphine, second row.
Yes, thank you. It's a question for Michael on mobility. You said that the gross margin in the backlog has increased, but you guide again for a margin between 8%-10% in 2026. So what drivers will allow you to reach a sustainable double-digit margin in mobility, and what are the headwinds?
Yeah, so you're right. Our guidance is 8%-10%. Our target margin is 10%-13%, and we're sticking to that because we think that is absolutely possible. Of course, going back, Corona was a hard hit in a project business like ours. We have logistics and so on and so forth. Our exit of Russia cost us quite some points. Those are being compensated right now.
You saw our results, and you heard our guidance for next year. The one I want to point out is the product mix that Ralf talked about. The market in the rolling stock area is extremely active, and we have decided to participate where we have good platforms and grow over proportionally because this will enable us to have a profit pool of the future in service. When this neutralizes out a little bit and rolling stock in proportion to service becomes more normal, then I think we are there. These are the reasons for what we showed today.
Take second row, John.
Thank you. Question again for Rainer. Again, I'm afraid it's on China, or at least Chinese competition, maybe not actually in China. Speaking to Innovance, I think last year their revenue ex-China was like 6%, but talking to them, they're aiming for 20%-30%.
If you delve down into that, I think they see Europe as pretty attractive, certainly against the U.S. and its current protectionism. I mean, are the barriers higher for you in Europe if they arrive and are trying to take share on various products in Europe, or would we like to see this play out a bit like it did in China, at least at the lower end?
I know they, for example, built a factory in Hungary, so definitely they want to move into Europe. We see that. We see that also in parts of Asia. We haven't seen yet a big traction that they are successful of doing that. I think our strategy is to stay competitive in China. If you're competitive in China, I'm also not fearing that we have a problem somewhere else in the world. Yes, they want to expand.
But as I said, up to now, I haven't seen that customers we have lost and so on. I think there was a lot of discussion around it. They also tried to approach distributors, but up to now, I think the success was limited.
Just to be sure, any questions from journalists at this point in time? No? Okay. Then we continue with the analysts. Maybe James Moore, first row.
Yeah, thanks. One for Rainer. If we just step back on your business over the last few years, we've had a roller coaster. Orders doubled in the supply chain crisis, then they collapsed. We take profitability, strip out software pre-COVID, and then we take out gas chromatography, and we take out the parts that went to Innomotics, which were both quite dilutionary. And we think about where your true volumes are times the price.
You're a bit below where you used to be on volumes, but profitability is materially below. I'm very surprised at the absence of upside for your DI margin targets. I would expect it both on the automation side and on the software side. And on the automation side, are you saying that we can't get profitability up over the next few years when volume normalizes upwards as it is in orders in China at the moment and as you get some of the benefits from productivity and the price equation and the productivity and the severance savings? Where do you think we are on the profitability curve versus where we used to be? And are we going to remain structurally lower as we are now, or can we actually return to where we should be excluding those effects and currency?
I think Ralf said we don't expose margin of different business units, so therefore I would stick to what Ralf has done. For sure, we're always striving for higher productivity and higher utilization, and that will drive profit, but we will not expose any numbers.
Any additional? Phil?
Yeah. All right. Thank you. It's possibly a question for both of you. There's a lot of discussion about German stimulus. What's your assessment of the opportunity set from that for both rail and perhaps general CapEx in Europe and Germany in particular? Thanks.
I can start. Of course, the German stimulus is being discussed intensively, and particularly for rail. I mean, there's a lot of talk about three-digit billion numbers to be needed. One has to remember that maybe 15% of those numbers end up in my market. Much of it is civil.
We have substantially increased our order intake in Germany in the past three years already, so there is a lot of renovation going on. We have created even more capacity for the future, so we really need the government to pull through. I think the need is obvious. I think the willingness is there, and I think what you can do with technology, we've seen today. So I'm cautiously optimistic, even though I would never commit to a timeline when it comes to these stimulus projects. I think what is important for us, if you look at the global market, one never below, there's a lot of volatility in every single country. I mean, Germany is not unique. You would see the same discussions in the U.K. They stop the budget and they do a budget review for one year.
But if you all add it up together and we have a good footprint in the main railroad countries in the world, then there is a substantial growth.
Yeah, and maybe adding on, we see a mixed picture in Europe. There are some countries we've seen also from Ralf's presentation, like Italy is growing significantly. Germany, there are pockets of growth where we said on certain verticals, for sure, aerospace defense is booming, and we are participating on that one. On the other side, in Germany, the machine building market is a very big market. And if you look at the numbers of VDMA, the Machine Building Association, you saw that machine building is not doing so well. So we are suffering somehow with these machine builders in Germany.
That will not compensate. That stimulus project for a government will not compensate a weakness, especially on the machine building side.
There is time for one last question. I think Martin, you raised your hand already.
Thank you. Going back to software and also with AI inside the automation business, how often is that offering and the go-to-market combined? I mean, obviously, we see a lot about adopting AI in design and simulation and how the customer builds a product, and then you're obviously using that in production. But are these quite distinct, or how do we think about that as a go-to-market being combined?
Now, have a look at that at the booth around the corner. If you see you want to create PLC code, you always have it 100% attached with the TIA Portal.
And that also gives us then a certain scale-up because somehow with the TIA Portal, then you can buy this option of this copilot, and then it's always combined. In that copilot aspect, there might be other topics which more go into analytics and helping the operator to assist, which is not 100% combined, but it builds on our Data Fabric. And with that combination, I also believe it will be a good attachment rate to our products.
Thank you.
Thank you.
Thank you, Rainer. Thank you, Michael.
Thank you.
.
Thank you. Thank you. It is my great pleasure to spend the next 15 minutes with you to talk about data centers and specifically AI factories. And there are good reasons because, as you heard this morning, I think I heard the word data centers at least five or six times.
It's been a great business for us, and it is now, with AI factories. I'm adding to the marketplace a super attractive business and a significant growth opportunity, so as I said, it's been a growth engine for us, an area where we outperform the market, specifically in the United States, and I'm leading our Siemens industry and corporate business in the United States, and for the last five years, that also meant that I've been at the helm of Smart Infrastructure, and now, with the increasing adoption of AI and the accelerated investments that we see in AI infrastructure, we are getting our arms around this marketplace. The interesting thing that we see is that the shift towards AI and the infrastructure that is needed to run these massive data centers. Foundational shifts in both technology and scale are taking place.
So what I will spend now the remaining time on is to walk you through the reasons why we are really, as a company, best positioned to capitalize on the changes that we see on the horizon. Data centers have been an important piece of our overall business. You see here, just last year, you already heard the number this morning. We've been able to increase our revenues just last fiscal year by 40% versus fiscal 2024. And that is even more remarkable because prior years, I mean, we showed significant growth already. As a result of this, we have been able to increase significant market share, seven points over the last couple of years. So, of course, the question is, how do we do this?
And you hear us talk a lot about customer centricity, but that is really at the core because we have great relationships, very regular interaction with all the hyperscalers that are playing in this business. In addition to that, we are working actively with co-locators that are becoming an increasing force in the data center space. But we don't stop there. Partnerships with GPU manufacturers are part of the ecosystem that we successfully manage. And to ensure customer proximity and really be where the business is, we've established eight centers of competence around the world, and most recently in India and Chennai, where we have now over 200 people really working with our customers to craft the technology of the future. From a portfolio perspective, of course, we draw on the strength of market-leading portfolio, whether that is in electrification.
You heard Rainer talk about our industrial automation portfolio, but also in the building space, and when needed, we also have the ability to help our customers through financing, and where customers need solutions beyond their expertise, beyond that, we have this ecosystem that we can draw on, and I'll talk about that more later, but of course, the revenue growth that you've seen doesn't come without investments. You heard earlier, just in the U.S. alone, in my area of business, we've invested over $285 million to increase capacity, and of course, the latest addition to our factory fleet in the U.S. was Fort Worth, with a significant investment in our switchboard assembly capabilities, but we're also investing in other places of the world, and most specifically here in Germany, with EUR 100 million going into Frankfurt.
And these investments that we've made have allowed us to be really the reliable supply chain partner that our customers need. And that means serving them on time as promised with market competitive lead times. So I hope you see that we're playing from an absolute position of strength in this place. And so with that said, let me shift to what we see ahead. AI factories. Very attractive market, a large market for us, and one that is expanding a lot faster than what we're expecting in the cloud space. You see, we see about a four times faster pace for AI investments than we see in the traditional cloud area. And what that means is that by about 2030, we expect that about 70% of the data center fleet around the world will be able to run AI workloads.
But with that come pretty significant technology shifts, and I'll walk you through what that means in a little bit. For us, this translates into a significant need for innovation, a significant opportunity to bring new technology into the marketplace. But we're also talking about significantly different higher performance requirements because complexity is going up. And that allows us to really differentiate in all core areas of our business and in the interaction of these subsystems in electrification, in industrial automation, as well as in digital solutions. You heard Tony talk, but also the interplay of all these different portfolio elements. So this is what we see in terms of market potential. And I need to start here with a disclaimer because we're talking about a very active marketplace.
And the numbers that you see up here on the page, they assume that the technology that is needed to enable this growth, but more importantly, power to feed these data centers is available in the timeframe that is needed to support this growth. So overall, when I look here at the portfolio, you see four different buckets where we're positioning today: digital and software, which is strategically the really most important one because this is where we can really support our customers, starting with design and simulation that we've talked about, for this real high-speed journey that does not leave any room for error. Secondly, automation and the building space will grow a lot more in importance given the complexities that I'll touch on in a minute.
Electrification and power has been a stronghold of ours, and that will continue to be a very important piece of our business in the future, and last but not least, also services. Another opportunity for us to increase our share of recurring revenues, so what is this AI factory that we're talking about, and how is it different from what we know in the cloud space? A couple of key factors. First of all, AI factories are a lot bigger than the data center of the past, so you see here, we're talking capacity and installations going from megawatts to gigawatts. We're already working actively on 100-megawatt designs. The market is talking about going to one gigawatt. There's even talk in the industry about scaling up to four. To put this into perspective, the city of Munich, on average, draws about 800 megawatts.
So just to give you an idea, so this is a lot of power. It's driven by the shift from CPU computing to GPU computing. And in order to do this and scale like this in reasonable footprint, the power density in racks needs to increase. We're going from, if you say, I'm like, we were 1x a little time ago, we're going to 10x, 40x. In some cases, there's talk about 80x. What that means to us, I'll touch on. At the same point of time, we don't only see technology change, but we're also seeing innovation cycles compressing. Where things used to change in years, we're now talking about six-month increments because this is what the GPU industry operates at, and this needs to be absorbed in the electrical infrastructure.
At the same point of time, the heat in a gigawatt data center needs to be brought out of the data room. And this means we're shifting from air cooling of the past to liquid cooling. And so I hope you're getting a sense of the complexities that are coming into play here. What this means for us is that system performance is becoming a whole lot more important. So let me translate this into the portfolio that we, as a company, can bring to the table. I'd like to lead with digital because this is really where I see the opportunity to create the most significant differentiation and add the most value to our customer base. Again, starting with design and simulation, which, by the way, also includes thermal optimization and thermal management, we can enable customers to shorten time to market.
This is important because what we're talking about here is a race between the hyperscalers to dominant position in the AI space. Every month they can launch sooner counts. How do we do this? Through digital designs, we enable the hyperscalers to design for constructability. I'm like, through skidding solutions, we help them cut the time to bring infrastructure to life while also in parallel optimizing energy efficiency. When you're talking about a gigawatt facility, every 0.1% that you can shave off in energy consumption counts. That is real money. At the same point in time, we're leveraging the common data fabric between all the systems that are touched on to train our lifecycle digital twin, optimize in the digital world, and then feed it back into real life to improve outcomes on an ongoing basis.
There are some very specific challenges also in the AI space that I haven't touched on, and that is the power characteristics. Here, we're already working on a real-time digital power twin to generate data sets that the utilities that now need to supply power to these data centers desperately need and are asking for. Because of all this complexity, we believe that automation and the importance of PLC-based high-end industrial automation will increase drastically. It needs to because human beings can no longer handle the complexity of the system that I'm talking about here. Of course, we're building here on Rainer's market-leading portfolio. And the secret sauce that we plan to deploy here lies in the connection between the three main pillars of the AI factory being compute, cooling, and overall the power supply.
Whoever manages to control this system in the best way is the one who shall win the race. At the same point of time, we have to handle a lot more data points. Imagine millions of data points that need to be handled quickly in real time. You can only do this with industrial automation. From a power perspective, what we're seeing is a shift from alternating current AC, what we're all used to, I'm like, from home, to DC, direct current. This is the only way to get the energy that is needed by these AI factories into the compute room. There's no other way around it. So what that means for us, because we want to do this safely and reliably, is a significant innovation opportunity because we need to switch DC, which is very different from AC technology.
DC operations require different protection and very different distribution systems. And we already have some of the portfolio, and we've been working on developments in this space for quite some time. And we will make new portfolio available as early as 2026. I touched a little bit on the different power characteristics of an AI factory compared to the traditional cloud center. And without going into many details, a cloud data center runs millions of processes in parallel. So imagine a very steady load that we can handle extremely well. The AI factory is very different, and that lies in the way I'm like, how the processes and the calculations run. But imagine a very erratic power profile. And that power profile is extremely hard on the grid, opens up for us another innovation opportunity because that load, that load pattern needs to be compensated and stabilized.
So we're looking into that as well. So can we do this alone? Not really. We believe today that our key differentiation opportunity as Siemens really lies in our core in the digital space, in the automation space, and of course, in electrification, in the connection and the optimization of these building blocks themselves. But if we really want to unlock the full value that we can add and really provide our customers with a truly optimized AI factory, and I'm talking from chip to grid all the way, we're working with partners. That's the only way to do this with speed. So the logos you see on this page here are all market-leading companies. I'm like, nVent is one of our strategic cooling partners in the United States. Delta Electronics, of course, a global player where we're partnering on UPS and energy storage, so battery systems.
But you heard this morning from Jensen, and we're also working very closely with NVIDIA as the leading entity right now in the compute space. Doing this really allows us to work with market-leading players in this field, with that in turn fostering innovation, interoperability between all these subsystems, and most importantly, do this collaboratively, faster, and in the scale that is needed. So the three key things that I would like all of you to take away from the session today is, first of all, we have outperformed in the data center market, and we intend to continue on this journey. Secondly, AI factories will come with a lot of change, technological shifts, significant scaling, will require a lot of innovation and system thinking.
For all these reasons and everything I just walked you through, we are simply best positioned to capitalize on this challenge and the change that we see ahead. With that, that brings me to the end of my remarks. It is now my pleasure to introduce our next speaker, Executive Vice President for Data and AI, newly joined from Amazon AWS, Vasi Philomin.
Thank you.
Good afternoon, everyone. It's been just over 100 days since I've been at Siemens, and I'm saying that because as I interact with people in the company every single day, I feel like a kid in a candy shop because the opportunity for innovation is just huge, really, really huge to push the boundaries of what industry can do, so it's great to be here with all of you at the end of what's been an incredible series of deep dives. I've learned a lot and hope you have too. Over the last few hours, you've heard how Siemens is connecting data across every part of our portfolio, how the Data Fabric now runs through our factories, our grids, our trains, and our buildings. You've also seen glimpses of how AI is already emerging inside those businesses.
So what I'm going to try and do now is to try to bring it all of it together to show what happens when that data fabric meets intelligence at scale. That's what we call industrial AI, the point where all of those threads converge, where Siemens' digital and physical strengths come together to change how the real world works. This is the next stage of growth for Siemens, making industrial AI real, safely, responsibly, and at scale. So let's start first with how our customers are already experiencing it.
The whole automotive industry changes with AI. We need AI in the shop floor for better performance, for higher quality products. The Siemens Industrial AI Suite and the Industrial Edge helps us to deploy the future use cases in our factories. For us, it's very important, and it's a standard solution because we can roll it out on every shop floor, and it's scalable.
Running a data center is complex and can waste a lot of energy if not done right. We use Siemens AI to improve our HVAC systems. We make them smarter and more flexible. The results are impressive. We cut our energy costs into more than half.
We have to do maintenance, all checking, debugging of the system post-delivery. So if this could be done more efficiently inside or offsite, that's a key benefit, in my opinion, for Industrial Copilot.
The place where AI can have the biggest impact in our factory is in the engineering of processes, in the maintenance of machines, and in making the life easier for our people here.
When you ask our customers what industrial AI means for them, I think their stories say it best. These aren't pilots, what you're saying. They're real production results. And together, they show you something very important. The next breakthrough in AI is unfolding in the real world. So in the past, AI has mostly lived in the digital world, analyzing data, generating text, like the kind of experiences we're used to interacting with chatbots like ChatGPT. But now you're starting to see AI move into the physical world, into the machines and systems that make, move, and power everything around us. This is industrial AI, intelligence that acts, not just thinks. And that shift from the virtual to the real is what comes next. Just as electrification defined the 20th century, I think industrial AI will define the 21st.
It will reshape how we design, how we build, and how we operate the physical world from bits to atoms, from simulation to steel. In the process, the entire industrial value chain will be redefined, where design, engineering, manufacturing, and operations become one continuous intelligent loop, and no company is better positioned to lead that change as Siemens is. Siemens is the global leader in industrial automation and software, connecting the entire value chain from design to operation, as you've seen many times today. Roughly one in three industrial machines worldwide run on a Siemens controller, so we combine decades of domain expertise with advanced simulation, automation, and digital twin technology, as you've seen before, and that breadth across design, build, and operations is what will allow Siemens to bridge the digital and physical worlds like no one else can.
It's that connection that generates something even more powerful, data. Let's look at the scale of that data foundation. Across our portfolio, Siemens technology generates trillions of real-world data points every single day. Data that comes directly from machines, grids, and buildings. We now have millions of connected devices in the field, and we have a partner ecosystem spanning hundreds of companies on Siemens Xcelerator. The way I think about this is that each connection feeds a common data layer, a living map of the physical world. That's what gives our AI models the context and accuracy that others can't match in the future. When you combine that strong data foundation with AI, you are actually starting to see impact everywhere. Across industries, our customers are already seeing it.
And you've heard a lot of the examples earlier today, but I'll point out to a few right now. At DMG MORI, production ramp-up times have fallen by about 40%. And if you look at Tata Steel in the Netherlands, the predictive maintenance has cut breakdowns by two-thirds. And at the Bank of Montreal, AI cooling has reduced energy by more than half. Now, each of these results looks quite different, but the pattern is exactly the same. When you bring intelligence into the physical world, you create measurable value. And as many of you know, value is hard to come by when we talk about AI. But here, you've got no other option but to show measurable value in order to be successful. So the question now becomes, how do you scale that value from individual projects to an entire ecosystem? And that takes us into the future.
We scale it through four strategic layers that compound intelligence over time. The first step is that we bring intelligence into the machines, directly into the machines, the factories, the grids, and the buildings, and at scale, every physical asset that can sense, decide, and adapt becomes a continuous source of efficiency and value, and over time, these capabilities will compound, creating entire ecosystems of connected learning assets. That's how Siemens scales industrial AI, not as projects, but as a living learning network where there's continuous data flow in and out and allowing our AI models to get smarter and smarter over time. So let me now give you a glimpse as to how that future could actually look like. This list that I'm about to show you is not exhaustive by any means, and it's also in no particular order.
We're just scratching the surface with some of these ideas. So the first idea here is generative simulation. Today, as you may have heard, simulations calculate, but tomorrow they will foresee. And so what's happening here is that weeks of simulation is compressed into minutes, and that's time-to-market redefined for a lot of our customers. So let me actually take you through a simple example. Assume you're an engineer trying to design an electric car. One of the ways you get there is maybe you start with a traditional car design, and then you try and take things out of that so that you make the car lighter. And you want the car lighter because then the battery takes the car much longer distances like you would want for an electric car.
And as you're doing that change and as you're removing things and adding things to the design, every single time you have to run these simulations. You have to run these simulations for aerodynamics using computational fluid dynamics. You have to look at thermal cooling. You have to look at things like the flow of energy there. And so you've got to run a lot of these simulations every single time you make a change. And these simulations are computationally extensive. And you have to run things like computational fluid dynamics calculations and also things like finite element analysis techniques. Now, what you can do with physics-aware generative models is that you can simply learn from thousands of simulations that you've seen before, and then you can instantaneously predict what's going to happen in those spaces.
For example, you can come up with the thermal profiles instantaneously, and you can also do things like figure out what the drag coefficients are instantaneously, so the next idea I've got here is on the industrial foundation model, and you've heard a lot about this in the past. Today, models are mostly assists. They assist you in doing your work, and hence the copilots, but tomorrow, I think these models will create, so if you're familiar with generative AI models today, what they do is they just generate text for you. If some of you have tried some of the more advanced models, the models generate images for you and also videos for you, but there's no single model out there, generative AI model out there that can take your intent and convert it into an engineered design, and so that's what we're talking about here.
This is going to be a very powerful tool when our customers have access to it. Moving on, idea number three is in the area of buildings. Today, buildings are controlled, and tomorrow they will self-optimize by themselves. If you look at buildings today and if you look at operators that operate the building today, the way they operate the building is pretty rule-based, and it's very reactive. They have to deal with very many different systems that are out there that help with the operations of the building. They have to deal with complex things like occupancy dynamics, and they have to deal with the weather outside.
Assuming that you've got a generative model that is able to combine all these signals together, you could make the lives of these building operators a lot more effective when buildings are able to sense and orchestrate energy, safety, and comfort on their own. The last of the ideas that I had today, it's on production lines, essentially. Today, factories are fixed, but tomorrow they will adapt. So if you look at a production line today, there's a bunch of cells in the production line. And some of those cells have robots that have been very programmed in great detail to do exactly a specific task. And what that means is you've got precision, but you don't necessarily have the agility. So in case your requirements change, it's going to take you a lot of time to reprogram the entire production line.
And so with models today, you have world knowledge in the models that we have today. And that combined with things like vision language action models that run on robots, you can completely change that equation. Production lines can reconfigure themselves overnight, combining precision with agility. So hopefully, you're seeing what I'm seeing, which is each of these ideas shows AI moving further from the digital worlds to the real worlds. And all of them point to one simple truth. Industrial AI is no longer theory. It's execution now. That brings me to the end of my presentation, and I want to leave you with a visual image of what we've discussed. If you walk through a factory today, imagine you're walking through a factory today, you see something really rare. You see design and steel living in the same room.
Together, I think they symbolize what Siemens uniquely enables: the fusion of bits and atoms, code and craft, thinking and making. I feel like that's where AI belongs today, embedded in the rhythm of how we build the world around us. That's the future that we're working towards. I think Siemens is leading the way to make industrial AI real and at scale. Thank you so much.
Last opportunity to raise questions for the next 10 minutes to Ruth and to Vasi. Daniela, first one.
Hi. Thank you for the presentation. Just wanted to ask on data centers. If you could elaborate a little bit more on the market share and your positions. You mentioned seven percentage points increase. Is it more on, I guess, which areas of those addressable markets is it more on? Is it white space, gray space, cooling? Sort of where do you sit and where do you still think you have furthest room upwards?
So I would say, let me differentiate this a little bit. So if you look at it from a geographic perspective, the largest market in the world is the United States. And there we have, I would say, excellent market share position without going into details. And we really have our fair share and sometimes more than our fair share with the hyperscalers, with the colocators, and now diversifying further down. Of course, we're working with the same companies around the world. So that's the geographic differentiation. Our highest share position, when you look at the different layers of the marketplace, we're extremely well positioned with the companies that drive technology and that drive innovation. And that starts with the hyperscalers, and from there you go down. Yeah, I think so this is geography.
From a portfolio perspective now, of course, our stronghold has been and is the electrification portfolio, where now with this change that I talked about, we expect to expand further. Again, AC to DC is a great innovation opportunity for us and a chance to really apply a lot of new portfolio into the marketplace, which is incremental and represents an expansion of our addressable market. Automation, I talked about in this shift towards the AI factory, there is no way around automation. So I trust that we will leverage our customer access to get more than our fair share in this space. And for cooling, so I'm getting closer to the rack.
This is where really at this point of time, we decided partnering is best for us because these areas move so incredibly fast that working with companies like NVIDIA and nVent, who are fighting this war every day, or pulling in Delta Electronics on portfolio pieces, I'm like that are hardware-centric is for us right now the best way to do.
There's also a question in the second row. Andre?
I'll ask one to Vasi. I think you tasked with investing about EUR 1 billion over the next three years, including building an industrial foundation model. Just wanted to ask on kind of what kind of KPIs will you be tracking the closest to assess the success of that investment, and how can you help us track it as well?
Yeah. I think in terms of KPIs, the first thing is, are we investing in areas where there's clearly value? And so you're going to have to find the right customer problems to work on. That's number one. And then number two is the speed at which we're able to iterate and innovate together with some of our lead customers. That's going to matter as well. And I think one of the important things to track is also how quickly we pivot away from ideas where the value is not and move into areas where value is. So very typical things that you're used to doing if you've done a lot of zero-to-one businesses. So we're going to have to track some of those metrics.
There's one question in the back.
Hello. Filippo Santelli from La Repubblica, an Italian newspaper. I would like to know if this project of industrial foundational model entails some sort of new legal and economic agreement on data ownership and data sharing with your clients because, of course, you have a lot of connected devices and PLCs, but the data are mostly your customers?
Absolutely. Yeah. So obviously, I've talked about all of the data that runs through our data fabric. That's going to be a huge asset. I think the reason why you haven't seen a lot of advances in the industrial space is because the data has not been very readily available for anyone out there. But I think that we have an advantage there in terms of the data that's flowing through our data fabric. But then in addition to it, we've been very busy, if you've noticed, we've been busy announcing partnerships with customers of ours. And so, for example, at the EMO fair just recently, we announced partnerships with the machine builders for data sharing. And so we're working through those kinds of agreements.
We're going to need all of the data we can get our hands on because of the need of how AI has to perform in the industrial space. It's not like the digital thing where you can just retry things again. So you're going to need high-quality data in order to come out with successful solutions. You're going to probably see more of those partnerships going forward to answer your question.
Any more questions? Gail.
A question on data centers. So when we hear OpenAI announce a 10-gigawatt deal with NVIDIA or another 10-gigawatt with somebody else, and when we hear the doubling or the trebling of Oracle's backlog from one quarter to the other, so in terms of the timeline, I mean, when is this when you actually see the impact of these orders flowing into European orders, but also into revenues?
Yeah. So we're already working today on some of the first large AI data centers for our hyperscale customers. But really, when we look at the timelines that come with one-gigawatt data centers and really when we start to see the scaling, I would call that 2027, 2028. Of course, power availability being one of the influencing factors we don't control, but I would say that's when we really expect the lift-off.
Can you hear me? Yes. The second question was on the so you talked about the market size and the breakdown of it between electrification, automation, and software, as if automation and software was nearly as big as the electrification part, at least for the next few years. Earlier, I think you talked about a revenue with data centers being EUR 2.9 billion-ish. Is it pure electrification, or does it include also the business you do within automation and software?
At this point, it includes both. So the EUR 2.9 billion includes the whole portfolio. But when you look back at that chart, and I should have pointed out the footnote that was in there, the way this external information that this projection is based on is structured, it also includes services in the automation and building number. So that number is not just PLCs or building automation or building infrastructure. There's also a large services piece that's grouped into that. So that maybe explains why, compared to electrification, it looks so large.
So within the EUR 2.9 billion, how much exactly is electrification?
Oh, now you're putting me on the spot. I would say it's the lion's share at this point of time.
Okay.
Any more questions? Martin.
Thank you. Just coming back to data centers again. In terms of understanding the scope of what you sell, do you have a metric like dollars per megawatt or anything like that that we can get some understanding as to what your opportunity is at the moment? I know you've mentioned things like moving to DC, and some sources tell us that's going to be an important but small part of the market. Others say it's going to be much larger. Just to understand how you think of that dollar per megawatt potentially growing over time.
So I'd like to take a pass on the dollar per megawatt because I don't want to give you a wrong number. But overall, there's a couple of different influencing factors, which is why I want to be a little cautious. So overall, AI factories will represent really pure training facilities, 30%-40% of the market a couple of years out. There is a high level of uncertainty just in that share. And in that space, you cannot operate without DC infrastructure. So now you have the higher potential development cost of DC infrastructure, different price points as we continue to scale. At the same point of time, the overall power infrastructure also gets a whole lot simpler in terms of the different devices you need.
So overall, there's a couple of trading factors that I think we need to understand and then factor in economies of scale before I can really give you a solid answer on that. Just the trade-offs are just too large for me to give you anything reliable.
Thank you.
So, we can take one last question. Ben.
Yeah, another question on data centers. I wondered if you could talk a little bit about lead times that you see across your data center business and how you've seen pricing develop in that market as well and where you see it going. Thank you.
Yeah. So overall, it's no secret that out of COVID, when everything was electrifying and everybody was working from home, lead times went beyond a whole year. So since then, with the capacity investments that we have made, and also, in all honesty, our competition has made, those lead times have largely pulled back. But for us, one key aspect that we focused on was really to be a reliable partner to the large customers that we support, where we have a lot of strategic supply agreements, where we have long-term visibility into the demand that's coming from these customers, where we're working very actively on standardization to really pull the lead times into what our customers need. So I hope that answers your question.
And pricing?
From a pricing perspective, overall, let me answer this way. We're still investing in capacity. That tells you there is still a capacity shortage in the industry. So we have not had any issues demanding price.
Thank you. Thank you, Ruth. Thank you, Vasi.
Thank you.