Greetings everyone, and welcome to this presentation of the launch of the ABB Infinitus direct current portfolio for data centers. With me here today, I have Giampiero Frisio, our President of Electrification business area. I also have Adrian Guggisberg, head of our division Distribution Solutions, and I also have Massimiliano Cifalitti, head of the Smart Power division in Electrification. I am Ann-Sofie Nordh, Head of Investor Relations. Today, we will talk through how ABB is leading the technology transition in the data center segment. This is enabled by our portfolio that we are launching now, supporting customers across the spectrum of direct current architectures, facilitating data centers with scalability, higher power levels, and improved efficiency. After the presentations, we will open up for questions as per usual. With that said, I will hand over to you, Giampiero.
Just before we get into the details of the portfolio, let us start with some history and the development of AC and DC to get us all on the same page. Please, Giampiero.
Thank you so much, Ann-Sofie, and welcome to this key event for the Electrification business, where we are unveiling the portfolio, Infinitus DC. As Ann-Sofie was saying, everything started with the DC, and not many people know it started 150 years ago. Because most of the people do not know that the very first grid started in direct current, in DC. It started in London, in New York, in Milan, and there it was the first application. It was for the street lighting. It was done mainly by Thomas Edison. But of course, there was one key problem of the direct current. You need to generate power very close to where you use it, because there was a limitation. The limitation was that it was difficult at that time to step up the DC voltage and transmit the power for long distances.
It was generating a lot of potential losses, and that is why at a certain moment, it popped up exactly in that period of time, just a couple of years later, it happened that in Rome, in Turin, at the Niagara Falls, there were popping up generation in alternative current. Why? Because it was easier to use. It was much easier to use. It was possible with the invention of the transformer, to elevate, to step up and step down. So it was easier to have a big power generation in one place and to have a transmission of this power at long distances with low losses. This, in a summary, in many books, and there was also a very interesting movie that it was done a few years back.
It happened that there was Thomas Edison and Tesla. They were the two inventors and the two scientists that they were fighting together in a big battle that it was called also the war of the current. It happened at the end of the 19th centuries. The winner at the end, you know better than me, it was Tesla because Tesla was capable to leverage the invention of the transformer and demonstrate that for the long-distance transmission, it was the right to use the AC because it's more simple. It is easier to interact because the AC voltage is passing to the zero, so for the breakers it's easy to interrupt when it's passing through the zero, the voltage. On top of that, it was easy to make long transmission line with very low losses. This was the battle that happened 150 years ago.
Now it looks like we have a revenge. We are coming back, or there is still, for 100 years, everybody was using the alternative current. They were forgetting a little bit on the DC, but now the technology evolved. With the evolution of the technology, there are now some advantages. I think that, as we were saying, for the alternative current, we have a lot of advantage. The advantage that the current is passing through the zero, so it's easy to protect. It's easy with the transformer to utilize the step-up transformer to make a long transmission line and to transmit in long distances. It's much simple, and everybody are trained on the alternative current.
On the other hand, today, with the advent of the electronics during the last 60 years, all of the devices that we are using today, that you are using today, that you are listening to us with the mobile phone, the computer, the television set, the LED lighting, everything that you're using, your cars, everything that we are using in our life are working in DC. So what is happening? It's happening that we have a super simplified transmission line that is super effective, but when the current and the voltage is arriving at our home, we need to make a lot of conversion from AC to DC with small devices that are converting AC to DC for our computer, for our mobile phone, with the EV charger for our cars and so on. Those conversion are inefficient.
They create losses, and they create losses in a big way if you are using the power in a big way. Now, of course, as we said, we are utilizing all the devices already in direct current. On top of that, we have a surge in the DC demand that is unbelievable. If you think about that, all the solar panel that we are using in our house or in the big solar farm, they are working all in DC. Today, they are generating current in DC, and we are transforming in AC in order to transmit. When they arrive into the final utilization, we are converting again. So again, are losses. You know that renewable energy, particularly the solar, they are growing three times that what we have today in the next 10 years. On top of that, we have that battery energy storage.
Battery, of course, are working in DC. The battery storage, there we are foreseeing, and everybody's foreseeing, that the battery storage will grow eight times what we have the install base today, because they will be needed in order to stabilize the network. On top of that, of course, as you know, the data center density, it will increase in an unbelievable way. We are talking about the single rack today, and all the server infrastructure are working in DC, and so basically they are arriving in the AC to be converted, to be utilized at 54 V or 48 V by the server rack. They are moving now from 30 kW power rack few years ago, or few months ago. They need to move up to 1 MW. So huge amount of power density.
The conversion losses that you could have between the AC and DC could become very relevant. That's why NVIDIA, most of the hyperscaler, we are working with all of them during the last few years in order to try to guide and help this kind of transformation and revolution. But we are starting from a big history. I think that you know that a few years back, four years ago, we were the first in the market that we launch a solid-state circuit breaker certified IEC in the world. We are working for the DC vessels since 2013, and the DC vessel is like to have a small, let me say, microgrid that is going to the sea and need to be full independent, need to be generating power, and it need to be used power, and with the propulsion system that is working in DC.
We are the leader in this market since already 13 years. On top of that, many people doesn't know, but we were investing in DC, and in particular in DC data center, already in 2012. I was coming in Zurich to work and live here in Zurich in 2010, and just after two years, I have had the opportunity to work on the first project for having a data center in DC that is still here. Is done by green.ch. It was 1 MW data center, is still in function. I don't know if it was the first data center in the world, but for sure it was one of the first that it was installed. It was installed with the technology that we used to have already at that time. You can imagine now, because of course now the evolution of the technology was very huge.
But on top of that, today we will talk about the portfolio of the Infinitus DC, and with my colleague, we will go more in depth. We talk about that we are doing now for the data center because we need speed, and there is an immediate need for the data center. Because there will be AI factories, going to talk about gigawatt of power. But on top of that, this portfolio and this technology will be used many other industry. It's going to be used, as I said, for example, in the marine vessels, it's going to be used in the renewable energy. It's used in the rail already since decades, and we are leader in the traction for rail because some of the things that you will see today is derivated version of technology that we already in our portfolio in the motion business.
It will be used for sure in the industrial sector, in the industrial manufacturing, in the EV industries, because, again, most of the things are working already in DC. Most of the motors, most of the drives, most of the things, they need to be supplied in direct current. Anyhow, let's make a step back and go in what's the benefit of the direct current. Let's imagine the benefit for the data center. Here we have a few examples. For example, let's imagine a data center that is O.5 GW, 500 MW. It's becoming a standard, this kind of power.
For a 0.5 GW Data center, if you apply a DC technology instead of the AC technology, we demonstrated together with BCG also in our white paper that we released a few days ago, we demonstrated that you can have a gain in less conversion losses of more than 5%. What does it mean, this kind of gain? It means that you could save almost 220 GWh . To make a translation on 220 GWh , what does it mean? It means the electrical consumption of 60,000 houses for one year in Europe. That's huge. It's big. We are talking about a mid-size city in Europe. If you think about the saving that the data center and the hyperscaler could have, it could be an immediate benefit because they could save in electricity consumption, and they could save some tens of million dollars.
On top of that, if you still think about the 500 MW data center, if you save 5%, it means that you could have also 25 MW available to generate more revenues, to generate more token. So to generate more revenues, that we can estimate that it could be up to or even more of the $100 million of revenues on a yearly basis. So these are the important advantage. That's why most of the hyperscaler, they are super interested to go in that direction. Now let's think about what's the saving from the CapEx standpoint. From the CapEx standpoint, as we said, for sure, the direct current has some advantage. One of the key advantages is that it's using less cable than the alternative current that is using in three phases.
With the direct current, you can use less cable, then you can have less losses because you don't have the induction losses typical by the alternative current. If you made estimation and you made a calculation just in copper, again, for the same 500 MW data center, you could say up to 10,000 tons of copper. Make a rough calculation. Today, the copper cost is around $14,000 per ton, so we talk about almost $150 million just in copper. On top of that, you need to put the installation, the manpower to install, the hours that you need to install, the additional footprint that you need to have, the anchors on the walls, and many other things. So at the end of the day, the direct current, it could enable the hyperscaler to have a higher power level and so to have the optimization of the token per megawatt.
At the end of the day, it's the profit margin that the hyperscaler could realize. So it's super important if you go in very high in AI factory infrastructure. How many of them there will be built during the next few years? At the moment, the capacity installed at the end of 2025 was roughly 80 GW. If you look at the estimation that everybody's doing, and you saw also the big investment that is going on from all the hyperscaler and all the colocation, we have talked about that in five years' time. So from 2025- 2030, there will be an addition of almost 170 GW. That is a huge increase. I was, a few days ago, together with a big hyperscaler where we are partnered with them, we were working. We are already looking on what it will happen in 2028 and 2029.
So we were planning the capacity needed for 2028 and 2029 by this hyperscaler. And this hyperscaler was telling us, "You know, I can give you some number. During the last roughly 15 years, we installed almost 11 GW capacity installed of data center." And they were telling us, "You know how much is going to be the capacity that we will put in execution just in 2027? 10 GW." So they will install in 2027 the capacity that they were building up in the last 15 years. That's a race. When we talk about many times about the race of the data center, this is the race that is going on. It's the massive investment, but the massive investment is translating in maximum capacity.
If you make all the calculation that we were saying before, with all the saving and all the saving in electrical consumption, but also on the generation from the revenue standpoint, you can imagine how much money, how much revenues, and much profit the hyperscaler could really generate if they move from AC and DC. And that's why they are so interested. And that's why also we are here to try to help the hyperscaler, the colocation, and most of our partners to move from a simple server room that was the data center from few years ago, and there are still some of the data center built up, to the AI factory. Before the data center were rooms where there were server that were fitted in and they were supplied with 48- 54 V. For a certain standard, it was working.
When there was 30 kW for rack, it was fine. Everything was working. It was the right setup, and this setup is right until 300 kW, 400 kW After that, it will be 300 kW, 400 kW per single rack. Because otherwise, if we go up and we increase this capacity, you need to go up with the power and with the voltage, because otherwise, you will have a lot of losses, and you need to utilize tons of coppers. And so it's becoming almost impossible to grow the power of the data center. This is an important part. That's why we are developing a new portfolio, and today we are talking about that, and because we will move to 400 or 800 VDC, whatever is going to be chosen by the hyperscaler, and we will be super flexible.
But in the meantime, remember, with the DC, we will reduce substantially the power conversion losses that will increase the potential revenue generated by the hyperscaler and will reduce the electrical consumption for the data center. How we worked in order to arrive to the portfolio that we are going to show you today? As we said, and I was saying before, we are in the DC technology since at least 15 years. We were already installing our technology in one of the first data center in the world that it was installed, that it was 1 MW here. It was done and is still here in Zurich. We are in this business since long time. In fact, we have multiple divisions that are working already for the DC. We have division in Electrification, we have division in Motion.
We are, for example, as I was saying, many things that are happening in the power electronic conversion, and most of the technology is in our Motion business. We have a traction business where we have a lot of knowledge, and we are doing since decades. We are talking about that we have inverter that we are using for the marine vessels since decades. This is the beauty of the ABB way, the fact that we work in a decentralized manner in order to have speed on the technology development and to be focused on some segment. For this particular thing, we said, okay, we have many divisions that are working in parallel on different technology, on different topic, on different segments, on different type of power electronic.
Now we need to build up a system architecture, and that's why we built up inside Electrification an incubator that is pulling together all the technology that is built up by the various divisions, and they built up a DC power solution offer that will enable our customers and will enable the colocation and in particular the hyperscaler to deploy faster and in a reliable way, in an efficient way, the future DC architecture. This is how we work, and this is how we are working, and this is what Adrian and Massimiliano, they will talk more in depth. We talk about the DC now and the movement and the transformation from the AC to DC. Many times in the data center, we talk about white space and gray space. What's the change that we are foreseeing for the white space and gray space?
Before we were saying, and it is happening, the DC data center will reduce the full capital investments for the hyperscaler because there will be less copper, there will be less installation material, there will be for sure the possibility to have more footprint. In general, it will be a reduction for the CapEx for the entire data center. But the big things that will happen is going to be the shift from the white space that today is around 45% of the total power infrastructure. This part, it will be reduced from 45% to 25%, and there will be the gray space area that will increase. It will increase step by step because there will be an evolution of the DC architecture. My colleague there will talk about that.
It will not be a digital trigger from one infrastructure to another one because there will be an evolution of that in at least three, four steps. But at the end of this fourth step, for sure, the big part of the gray space is going to be three quarter of the total power infrastructure. What does it mean for ABB? It means that today our strength, it was and it is in the gray space. This gray space, it will grow during the next few years, and so we could benefit much, much more of this transformation from AC to DC. For sure, we will not forget DC, because not all the market will move in DC. It will move a portion of this market, mainly on the high infrastructure and the big data center.
We are estimating, making the analysis also with the ecosystem that is around the world and many studies, we could estimate that in 2030, between 25% and 40% of the data center installed in 2030 will be in DC. It will be mainly the AI part of the data center, where we talk about the big power and not, let me say, the data center of 25- 30 MW. That it will remain for some type of segment, and those ones for sure, they will work in AC. We will continue to focus on AC, but we will help our customer to move and make a transition on the DC power. Summarizing in a nutshell, what we are launching today is going to be a portfolio that will enable our customer to move in the journey from the AC to DC.
But now I think that with Adrian and Massimiliano, it's time to go more in depth.
Technology has evolved rapidly, but the way we power it hasn't. It's time to take the next leap, rewiring power architecture to eliminate conversions and directly serve demand. The direct current revolution is here. ABB is transforming decades of DC expertise into the Infinitus portfolio for the innovations that come next.
We are super excited to speak about Infinitus, and now we will go a little bit more in deep. Infinitus is our direct current portfolio, and today we will be focused more on data center for DC 800 VDC application. As Giampiero said, this will be an enabler also for other heavy industries, for transportation, for renewable, and so on. What we want to give today is three main messages. The first one, we are the first to give to the market a full portfolio from source to rack. The second message is that we will have all the building block that our customer can use, not only for the landing solution, but also for all the hybrid and transitory solution. Third message, thanks to this building block, we can capture all the opportunities around the world. This is a race.
There will be customer that will be very fast, early adopters, and some other that will be more traditional. We can partner with all of them so that we can optimize and maximize our opportunities around the world. When we speak about the transitory solution is exactly what we want to show in this slide. Where we are now, most of our customers are using traditional AC distribution network. Now what will come in phase I, there will be already the need to have sidecar close to in the white space to support the high power in the rack. This will be very fast. Immediately, our customer will need to have the distribution at 800 VDC in the gray area because it is there where they will make money saving in OpEx, CapEx, and footprint.
There will be a stage where they will experiment with different solution using, for example, DC UPS or other alternative like the rectifier from motion colleagues. Then the landing solution will be the solid-state transformer. This is where our star will take place and our customer can benefit for the full advantages that Giampiero just spoke.
Now we have heard from Max a little bit on the evolution on how the power distribution will change over time. We heard from Giampiero earlier about the change from AC to DC. Now, with DC, there comes a major challenge. DC is not standardized to the level of AC as it is today. AC is super plug and play. Everybody can build AC systems. This is why we choose a system approach. Our building blocks, they enable to build a full system. In order to get some modularity in that, we structured our offering into five major building blocks. We start with the powertrain. The powertrain is basically the element which connects the data center into the grid or even into behind the meter power generation or any hybrid solution.
It secures that there is reliable power, and it also secures the power connection is stable. The second part, and that is probably a bit of the heart of the new solutions, is the conversion from AC directly into the DC required in the DC distribution. From there, actually, we have the distribution part of the DC, which replaces the traditional AC distribution or low voltage. A core element in that is the protection. That is the big challenge when you change from AC to DC. Then not to forget, there is the whole cooling. Why are we mentioning here cooling? If you look at the efficiency or the losses or the energy use of a data center, there are two major consumers. One consumer is the data hole. That is what everybody is talking about. There is another major consumer, that is the cooling systems.
Also in the cooling systems today, there are multiple conversion stages which create losses. What we do with the new solutions, we also reduce conversion stages in the whole cooling system, which again, will increase the efficiency. Now, in the next step, let us look a little bit deeper into the products within these building blocks. You got a bit of an impression about the scale of this portfolio, and we just look at this here. Why is it so important to talk about the portfolio is a single component is not bringing the solution. We are working heavily in standardization. We are very active because we believe standardization will help to speed up, but before standardization will happen, we need to bring the portfolio together to help the customers to build the systems and the solutions.
Let us deep dive in two, three components, which are super essential to make this successful. Max, maybe you start with the first one.
Yes, thanks, Adrian. One of the main product that we want to deep dive is the HiPerGuard. Let us say that with this HiPerGuard, we were the first in the world to have the medium-voltage UPS with solid-state technology. Now we rely on almost 10 years of experience. Imagine how we learn from the field, how much installation we did. We already installed more than 2 GW of HiPerGuard over the last years. This help us to improve the product, to learn from the product, optimize the product, and so on. One of the most important characteristic of the HiPerGuard is the scalability. It is designed in a modular way, so that you can put 20 million voltage UPS in hard parallel to arrive to 50 MW UPS system block.
This is very useful when you want to be very fast in the implementation and installation. Using our medium-voltage UPS, you can also improve the OpEx, because the high level of efficiency is super important when they are using these 100 MW of installation. This is another saving that you can have every year on the bill of electricity. Then on the footprint. The footprint of our medium-voltage UPS is 30% less compared to traditional architecture. So it is also a class one performance product, so it means that the power quality is super high. Now let us move to the solid-state transformer, another important building block.
Absolutely, Max. The SST probably is something you read a lot, you hear a lot. I think a lot of people talk about this. Let me talk a little bit about what we are doing here and why we believe we are going to have a solution which will help the industry to adapt fast. There are a few points. First of all, I think the technology has to be done in a way that it matches to the data center needs. That means you have to have the right power and the right reliability of the product in order really to have a building block which becomes relevant. Many things you would see in the market right now is probably way more from other industries like EV charging, which comes in much lower power than what a data center needs. So that is the first step.
The second step is we have invested in this technology for many years. I think Giampiero mentioned about this also in our research centers, for example. We are running this since almost 15 years on this technology. What we choose to do when we saw now comes the moment when this is going into industry and will scale to choose a solution which is based on proven technology building blocks, because reliability is key. The other part is it is key to be able to ramp up very fast the capacity to produce and to deliver. That makes it unique. So it is not all invention in terms of every single piece or component is a new invention. It is the way we apply and bring things together. Also, Giampiero mentioned our traction business. Some of the components and building blocks, they are coming from a well-proven traction system.
As you can imagine, trains are not stopping. They need to run. They are very critical in terms of reliability, so we are using very proven building blocks. With that, we believe we have here a component which is really there, ready to scale up very fast. There is more than converting from AC to DC, and mentioned before already, one of the key challenges in DC is protection. Max, maybe you add a few words on this.
Yes, Adrian, only a few years ago, during the Capital Markets Day in Frosinone, we launched and we spoke about SACE Infinitus, the first solid-state breakers certified. Now, we understood that at that time, it was super important for marine application, for other segment. Now we see a lot of interest in the data center, and now we see also opportunity with smaller size. This is the reason why we will have, at the end of next year, also two new frames. This will be at 150 A and at 500 A. Why the customer needs super fast breakers? Because a solid-state circuit breaker can be 100 times faster than a traditional breakers. Why? Because of selectivity. Because there is a lot of power electronic, as Adrian described, up front, and you need to have a product that can protect very fast.
When this is needed, and is needed in microsecond, then is where solid-state breakers is used. Giampiero, over to you for the final comments.
Thank you so much, Adrian and Max for going in depth in the portfolio. I hope that you enjoy, and you are excited as we are, for this new portfolio that is the Infinitus DC. I think that summarizing the flexible portfolio that we are showing you today is going to support the deployment of to our customer, is going to facilitate their deployment to make a transition during the various evolution of the architecture that will happen in the next few years. The second important point is that the data center, the CapEx spend of the data center in the power infrastructure, they will move from the white space to the gray space, and this is our strength. This is where we are realizing most of our technology. This is where we will deploy our strength.
For sure, the portfolio that we are launching today is going to be supportive for the performances of the entire ABB Electrification, and it will help to reach the target of the operational EBITDA for over the cycle. Ann, now to you.
Yes
For the last words.
Indeed. We will now open for questions. As per usual, I want to remind you that for those of you who have dialed in on the phone, just press star 14 and you will be able to register a question. Also, as you do so, please remember to mute the webcast when your line is opened. Let's see how many questions we get. Just in case, please limit it to one question per person, and before we ask you to get back in the line. We wanted to try and get through as many, and let through as many as possible. As an alternative, you can also put your question through the online tool on the webcast, and then I will voice it over from here.
With that said, we go ahead with the first question. We have both through the online tool, but we start with the conference call, and we open up the line for Will Mackie at Kepler Cheuvreux. Are you with us, Will?
Yes. Thank you. Yes, can you hear me?
We can.
Can you hear me? Perfect. Good afternoon. Thank you for the presentation. For the question, I guess my one question would go to how you see the pace of the adoption of your technologies unfolding, and how do you see the limiting factors to that adoption? Does it relate to employee readiness and knowledge around DC, manufacturing capacity in the industry, or certification processes in the regions in which your technology will be utilized? Thanks.
Do you want to-
I could take it, I could answer. Thanks for the question because it's very relevant. As we were saying also during the presentation, we do believe that the adoption of the DC, it will happen, and there will be the first execution of data center in the next two, three years. It's going to be the adoption also because with the new chips and the new GPUs and TPUs that will be released, it's going to become a need. So it's going to become needed simply because the power of the computing capacity and the power of the chips that needs is so high that it becoming also physically impossible to use just the alternative current.
I think that, as we were saying, within 2030, we do believe that for the installation that will happen in 2030, we are estimating that the DC data center is going to be around 25%-40% in 2030 in terms of installation in that period of time. It means that starting from 2028, we can see the ramp-up of the DC data center. After that, in terms of capacity, I think that Massimiliano was mentioning also a little bit, we do believe that the capacity, for example, for what we are producing, it could be done already in the investments that we already defined, that is already in our roadmap, that we already announced in terms of footprint.
Today we announce, or better, during the last couple of days, we announced this Infinitus portfolio that will be ready during 2027, and from the technology standpoint, we are ready to have it.
I can actually, we have another question here on the timing. More exactly, when will the portfolio be commercially available?
Yeah. There are some things already available, some things will happen during 2027. In the next 12 months, it will be fully available, in particular, if the question is related to the solid-state transformer, that it most probably is going to be also, this is the question related, is going to be in the second half of 2027, the availability for the commercial standpoint.
I'll tie on a little bit more here, a question from Alex Virgo. Is it all your own IP, your own silicon and design, et cetera?
It's our IP, for sure. We develop internally. It's all the things that you saw here is developed internally under our IP, right. Of course, we are using a huge ecosystem of supplier partner that are working with us, also from the semiconductor standpoint, but the intellectual property of what we are showing today of all the solutions and the portfolio is owned by us.
Okay. Then we open up the line for a question from Alasdair at Bernstein, please.
Oh, yeah. Hi, good afternoon. Thanks for the question. My question is really on the power quality side of the portfolio. So the DC UPS, the SST, the TRU. You seem to be aligned here with some of your U.S. and European peers, basically in terms of pointing towards more than one technology path to centralized DC, at least in the medium term. Really, I guess the question is, to what extent do you think that the TRU, the transformer rectifier unit, is really going to be seen as a genuine alternative to the SST during that period? How might that influence the pace or timing of the solid-state transformer adoption? Maybe the answer to that is in the type of deployment or power requirements each can maybe best address, but maybe you could touch upon that as well.
The final question, I suppose, on the SST is, it is obviously maybe a more dynamic technology, but you also see a different system level efficiency benefit of the SST.
Yeah
Over the TRU as well. Thank you.
So multiple question. Adrian, I think that you can answer.
Oh, sure. Thanks for asking the question. I think some of them go a bit technical, some may be a bit more on the more generic side. Let us talk quickly first about, you asked about the adoption in terms of DC and in terms of the TRU solution, why this versus an SST solution. I think this has to do with two factors. It is a little bit about scalability and the overall design, but also a little bit of the pace as to the previous question on the adoption, how fast somebody goes to the full native solution of 800 VDC.
Obviously, it is connected to the second question that is a bit rational of moving from a TRU solution then to a full SST solution is the efficiency gain. So basically, what happens in technology-wise is basically with the SST, we are eliminating the transformer, the traditional transformer.
To be also very clear, within an SST is also a transformer, but a high-frequency transformer, which has this size instead of filling this room here. That's the huge saving also on the efficiency, which we gain with the new technology. So it's a little bit of both. It's the pace of adoption, and the other thing is really then to go to the higher efficiency.
Very good.
Great. Thank you.
Did you have anything else, Alasdair? Nope. Then we take a question here from the online tool.
Yeah, there is.
What is a realistic market share of ABB to receive in the 800 V DC market when they look at the competitor developments?
For sure, for the market share, as you know, we are not disclosing market share for particular product line or solution. It is going to be the same for the Infinitus DC. We do believe that with the launch of today, it is going to be, as we said, the first industry source-to-rack portfolio in DC. We can put on the market all of our strengths, the 25+ year of experience on the DC application, on the application that we did already in the past on tractions, on the marine vessels, and on all the power electronics. So we do believe that we could put all of these strengths on the market, and we can support all of our customer, whatever architecture they will choose, whatever path they will choose during the transition from the AC to DC.
Here we move to a question from Aaron. Could you elaborate on why the sidecar architecture will be a fast transition and not a sticky one? Are customers not interested in moving to full DC?
No, I can take this one. First of all, we think that sidecar will be mainly for retrofit and it will be for transitory solution, as we said before. Because if you have only the last stage of conversion, very close to the rack, so in the white space, you have the full gray space that will still have the traditional AC solution. That means we will avoid all the saving that Giampiero was explaining before on CapEx, on OpEx, on footprint. This is the reason why if the customer want to take the full benefit of the 800 VDC, must go in the gray zone with the 800 VDC.
I continue here. I have quite a few questions to-
There is one there.
Yes, we'll get to that as well.
Okay, sorry.
Here is from Mattias Holmberg. It disappeared. Now here, compared with today's AC architecture, how would 800 VDC change ABB's revenue opportunity per megawatt?
Yeah. I think that we were always saying, if you recap, you know that typically in the AC we are around a little bit more than $2 million per megawatt. That's an average, of course, because it's depending countries, it's depending many things, geography and whatever, but roughly around +$ 2 million per megawatt. Is a fact, as we said, that there will be moving from AC to DC, there will be a significant transition from the white space to the gray space. The gray space will increase from the 55% of the total CapEx on the electrical power investment, is going to move step by step to 75%. So for sure, it will be higher than the $2 million. We are confident that it's going to be significantly higher than the $2 million per megawatt.
Now we go to the conference call. We open up the line for Gaël at Deutsche Bank.
Oh, good afternoon. Thanks very much. Could you talk a bit more about the safety regulatory hurdles, the more complex permitting that we are certainly going to see with DC systems, also the industrial timelines and the much higher price points for the DC technology versus AC. All these things that could potentially slow down the deployment of DC systems. Just a clarification on your statement that you expect 25%-40% of the new capacity to be DC related in 2030. Within that, how much exactly would be full complete grid to rack 800 VDC systems, and how much would be more related to hybrid architectures with some sort of legacy AC system with a bit of a DC portion? Thank you.
Let me start, and eventually after that, you jump in, Adrian. From the electrical power system, we do believe that, or let's start in this way. The total capital investment for a DC data center, it will be significantly lower because as we were making the example before, just for the copper standpoint, for a 500 MW data center, you could say 10,000 tons of copper with the actual cost of copper, just the cost of the copper per se, we are talking about $140 million. Plus, you need to put all the people, the manpower that you need to anchor all the busway or all the cables. There are many things. But on top of that, there will be the conversion, as we were saying.
From the conversion standpoint and the saving on the conversion, if you, again, you make the 500 MW example, you save 5%. It means saving, of course, in electrical consumption, and we talk about some tens of million dollars, but the significant is going to be the additional revenue stream that you could generate with the saving. We talk about for a 500 megawatt data center, that you could have additional 25 MW that you could use to sell in terms of revenues. It's depending on is the data center is going to be for a type of application or for selling, for token, we are talking about from $100 million to $300 million of additional revenues that you could generate, and you know better than me how much is the profit margin that you could have from that part.
I think that is significantly the advantages from the DC standpoint, not just from the technology standpoint, from the saving, but also from the financial standpoint. Is from our power train, so for our electrical power, the capital investments, we do believe that it will be more a shift from white space to gray space, as we are saying. It means that the total capital investments for a data center in our market, let's say in the electrical space, is going to remain pretty similar, ± 10%, depending on the architecture or stuff like that, but it's very similar to what we have today. There will be additional cost for some component, but if you look for the total capital investment, it will be significantly lower for the entire data set.
For the regulation, for sure, there are new things that need to happen. I think that, Adrian, eventually you could provide some feedback here.
Oh, sure. Before I go to the regulation, maybe also a few words about the kind of the adoption and the pace and the speed. Several questions are related to this. Maybe it is really important to say, this is our approach with the full portfolio, saying, "We are comfortable with whatever speed this adoption will take." Here is the why. I think the first step of adoption is purely driven by the need of the higher power density of the racks. There is no other way than going to DC. If that goes with sidecar in the first step, or that goes very quick to native, we are okay either way. I think that's what we try to do to go with the speeds of the customers and how they feel comfortable with it.
The second phase of the adoption, what we expect to see is once the native DC is becoming established, the energy efficiency gains will also be taken advantage in data centers which don't need necessarily the full high-power rack density. That's phase II of this. If you ask me about what will be the share in 2030 or 2032, we are not able to answer. We are also not speculating on it. For us, actually, it doesn't matter so much how this adoption will come. What we try to do is to be the partner in any situation or any architecture. When it comes to the standardization, I would say there are two things to be considered. One is performance and quality standards. These standards don't exist today. We are heavily involved in creating these standards because we believe in standardization. This will drive even faster adoption.
The other part, which is existing, is when it comes to safety. That's the critical factor in order to bring things into operation and get the permission to deploy. There, I think the path is open to deploy these solutions. We do not see there the problem on that one. But certainly, it needs the right knowledge and the right understanding of the safety requirements in order to really to have an industrialized solution.
I am going to follow- up here with a question that came in, it relates to the white to gray space comment that you made. Can you elaborate, please, how large part of your revenues as of today go to white space versus gray space from Ole Fløttum?
Yeah. We always said that our strength is the gray space, a significant portion of our revenues already today is coming from the gray space. We do not disclose how much is it, but it is a significant portion of our revenues today in the data center are in the gray space.
I continue here. It is quite a flow of questions coming in, so that is really nice. Here is one from Daniela Costa. "Do you think 800 VDC will also be used for inference? Is that in the 25%-40% DC figure you see by 2030?
Yeah.
I can take this. Yeah.
Go ahead.
I just mentioned before, I think the first start will be on the training side because this is where the super high power density is needed. As I said, we expect this to move also into more traditional, which interference is included, in order to get the energy savings, so the efficiency gains. I think this is two phases. When we talk about the figure, obviously, that's a mix. That's not only AI training. I think it's probably more related with, if you think about the long term, it's very much an adoption curve. This is also why we give a quite big range on where this adoption could lie, because we cannot tell where exactly this will lie nor where exactly that mix will be. Again, what we know is the high power density is coming, and the 800 VDC is needed for the high power density.
That's going to happen.
I continue here. There's one follow-up question from Alex Virgo. "Does H2 2027 commercial availability of SST mean that's when you anticipate fully certified and commercialized, or that's prototype launch, and then you need testing, certification, standardization, et cetera?
When we talk about commercialization, is H2 2027 available to be sold. It means that in this period of time, we will do already pilot installation. We already done some pilot installation in some field. We are making all the certification piece, and we are making all the capacity built up, and so on. When we talk about commercialization, it means that it could be ordered or could be ordered as start in the second half of 2027.
Maybe Adrian and Max, here's for you. You can split this between you. Question from Jonathan Mounsey: "Where are you differentiated for certain product categories? Do you have any unique IP that would allow you to make more share than the customer usually allocates to one supplier?
I will start then, Adrian, please build on what I say. First of all, we were very often the first to have certain technology on the market. These allow us, over time, to build huge experiences. Let's give an example on the medium voltage UPS. We were the first, and after we are learning, it's now 10 years that this platform is on the market. We have learned a lot from our application, and just a few months ago, we launched also the 34.5 kV HiPerGuard. That this will help us to direct connect the medium voltage UPS to the grid, and so to eliminate another point of conversion in the system. Of course, being the first very often means understand the market, develop a new solution, learn also from small mistake, and to evolve the platform. This is very key.
We have done the same with the solid-state circuit breakers, where we were the first. We started even with other application, with the marine application, and now we discover that is super important also for data center. This is why we are learning, and now we will develop also to other frames.
Yeah, maybe I am adding a little bit of specifics to the solid-state transformer on this one. You asked a bit the question about why ABB believes we have the strong position. I would say there are kind of three factors or three things which we do maybe a bit different than others, even though I cannot comment on the others, what they exactly do. The first thing is we are really focusing with our design on data centers. When we look into the market, if I just look into the ratings of what we see being published, a lot of these products, they are way more focusing on EV charging from a power rating point of view. For a data center, it is not necessarily the way we understand data center designs. The second one is very much related to that.
Actually, we do a lot of core design, so we are including some of our major customers into the design we are doing in order to make the trade-offs, which are the most practical ones to have a solution which can be deployed fast. On the other hand, also meets the key requirements. Then the third step, I would say that is probably one of the most important decisions we have done. We are reusing proven building blocks in order to have scalability of the product. What we have learned in the data center segment or business is it can change super fast the way a design would be deployed once it is proven. In order to do that, I think we have the high focus on really being able to ramp up our supply chains and our capabilities to produce, to test, to deliver this new technology.
That is kind of the three things which we see a little bit different.
Very good. There are two questions here, sort of similar, on the topic of cooling.
From Will and Daniela. A broader question, what is your stance on cooling? Would liquid cooling be something you could develop organically or partner with someone, given how important it is likely to be for DC?
Yeah. Let me say, I think that from the cooling standpoint, we have a very good position in the sense that we owned, and we are supporting most of the cooling company with our technology. With our technology that is the variable speed drive, with the short-term battery for the CDUs, for example, or with our motor, IEC and UL motor. I think, in the CDU, there is a coolant distribution unit, what is the biggest portion on top, of course, on the older mechanical parts, that are the pipes and the heat exchanger and so on. From the electrical standpoint, the strength are the drives and the motors. We are, as you know, market leader on drives, and we are in the North America market leader for the UL, for the NEMA motor. So we have a very good application.
Now with the Infinitus, also DC portfolio, we are also launching a portfolio of variable speed drive that will enable the possibility to utilize directly the 800 VDC. So you can go directly with 800 VDC also in the coolant distribution unit, where our DC drives will enable the possibility to optimize also the energy consumption of the CDU. So our choice is to support all the various liquid cooling company that are growing, and we are enjoying that they are growing because we can provide our technology to them.
Let's move on to more of the competitive landscape. A couple of question here on that topic from Aaron and then from Will. How do you see the competitive landscape for SSTs? Eaton, [inaudible], Daiichi all claim SST technology. How is ABB positioned or differentiated? The same question sort of on the competitive landscape for medium voltage UPS and the GE Vernova announcement, et cetera.
Yeah, probably a bit of the same question as we had before.
You asked before.
I would say mostly it is answered. As I said, where we see us focusing, I mentioned these three topics just before, about design for data centers, about the core design with some of the key customers and really have the product prepared for a fast scale-up. So that is kind of how we differentiate on that space. Again, commenting a little bit on the products. You guys can also take a look at what is publicly available from many of the folks which are claiming to have the product. Just look a little bit on the power ratings. The power ratings, if you imagine that the rack power will be 1 MW, I do not think people want to deploy one SST per rack.
Yeah.
That is probably not a solution which would commercially make sense. So there is a little bit of a gap.
That's why we will have an SST that is 3+ MW with a very high efficiency.
Yes
Of 98.5% +.
Which is typically a common architecture in the data centers.
Yes.
Absolutely.
For the medium voltage UPS, also, I think I answer in some way, because I think the most important thing is that we are on the field since several years. We have learned, we have improved, our innovation pipeline is very dense. We are learning from our customer, and I think that the fact to be so in intimacy with our customer allow us to have a very reliable and safety product. I think this is super important because all the application are really critical application, so must be quality at 100%.
Yeah. I think that, also for the medium voltage UPS, we are now having more-
Yes
Player on the market, as Massimiliano was saying. It means that we were most probably right when we launched as a first in the market. Now I think that the fact that we will not be alone, it will enable many others, colocation and hyperscaler, to adopt this kind of technology, and it is going to be a benefit for sure for us.
Yes, because Giampiero, as you said, very often to be the only one in the market could be a disadvantage, because very often these big hyperscaler want to have double sourcing, multiple sourcing. So it is again a confirmation that we had the right product.
Here is a short one from George. Have you taken any orders yet for any of the products or the portfolio?
Yeah, we can say for the medium voltage UPS, we already taken orders, significant, I could say.
Yeah.
We are talking about some gigawatt that we already taking in our portfolio and also installed a lot of them. We have for the solid-state circuit breaker, for example, that we launched four years ago, we took already for other type of industries, and we took already for the rest, for some of the SST, for example, we are talking with many hyperscalers and few colocators also because they are super interested and as we said, Adrian was saying, we are co-designing the activity. We are in touch with them since long time, and we are co-developing also with their input, their suggestion. We are in touch with them in order to soon deploy also for them on the field.
And maybe if I'm adding a little bit something in also for everyone to understand the process of most of our customers, what they're going to do is they're going to deploy some of the solution in test or in proof of concept type of installations. That's also important for us because we need to operate the systems together with the real loads, with the real racks, with the real situation. Today, these racks don't exist.
Yeah.
It's not possible yet really to test it on the real conditions. That's, again, what I want to say is we are not the bottleneck in order to drive the deployment. That's the position we want to be.
Yeah.
I have just a couple more here before I let you guys go. Here from Andreas, who has followed us for a long while, he met with us.
That he did also a fantastic, Andreas, report that I read.
Yes
A couple days ago.
He says, at Data Centre World in London in March 2026, ABB said that 50% will be DC by 2030. Now we say 25%-40%. What has changed?
I think that it didn't change or it didn't change any. I think that there it was, or a mistalking or a misinterpretation, but the idea there, it was always to say it was 50% of the AI data center in 2030. Looking at the AI data center, that is almost half of the business is, I think that is coming back to the 25%-40% that we are saying today.
Mm-hmm. I know I said a couple, but there may be a few more. Here's from Will. Behind the meter power is fast growing for data centers requiring more control. How is ABB positioned to support microgrids control and balance between grid and DC?
Yeah. You want to. Yeah.
Up to you.
No, you go, Adrian. You go.
Yeah, I think it is absolutely true, and just given a bit the business I am in charge of, we are working a lot with the utilities, with power quality, with reliable power. So be it best offerings, be it offerings around synchronous condensers. There are a lot of things we are doing already to deploy for grid stability as ABB. So we have multiple solutions, and obviously we are working super close now with some of the customers in order to also being in a conceptual level, how to make that the best way on integrating behind the meter power, net grid connection, and the loads. So that is absolute focus area.
Yeah.
Yes.
Then the powertrain, exactly as you said, on the powertrain on the grid, on the medium voltage powertrain, there is a powertrain and module that Adrian was saying that is exactly that part. All the stability and the solution that could smooth also the loads versus the network.
Here is an M&A topic or M&A related from Max. When you look at your data center business and the broader technology shifts, do you see any white spaces that you would need to fill with potential M&A to strengthen your position?
Yeah. I think that when there is something that is part of our strategy, I think that when they did a Capital Markets Day almost a year ago, because it was already November last year, we were saying that on the four topic that are important for us, there is also what we call smart data center. When we were saying smart data center, we were meaning also now the DC. For example, recently at the end of July, Massimiliano in the Smart Power division, it was announced an acquisition of Advantics, that is a French company active with a very interesting technology on the DC converter that is going to be an important building block of the future data center. That's why we acquired them. They have a super interesting technology, and it's going to be at this part of the Infinitus DC portfolio.
When we are seeing that there is a good deal to do it, in particular with very interesting and futuristic technology, we're going to take it, and this is what Massimiliano did in July.
Here's one for you, Max. Medium voltage UPS.
What is the role of the medium voltage UPS in full native DC architecture with SST in long term? Is it still really needed? Says Andreas.
Yeah. We fully believe that this technology will be needed also with the full native solution. Of course, there will be also alternative to have a best, let's say, and orchestrating with the best after in the 800 VDC, but we still think that is the best solution because the overall efficiency of the system will be increased and also because you can save money also again on copper. We were already doing study. We have white paper where we show that the copper, you can save kilograms of copper, is part also what Giampiero already said. So we think that more we are in the medium voltage, more we increase the voltage, more the current is lower. So you can save still a lot of copper. So we still believe that is the right choice.
Yeah. Also is the right choice today if you want to have a future proof data center. Even if you're doing today an AC, with an AC architecture data center, if you use the medium voltage UPS, you are moving, let me say, closer to the grid, the part of stabilization and control. So it means that if you will need to do a retrofit on the DC application, you will have already future proof architectural data center. It's going to become easier in the future to move from AC to DC.
Maybe a bit to add, on the overall architecture, what we would see is, and maybe that's also something to consider as we are talking here partly about AI or data centers for training, they have a complete different need in terms of reliability than what we would have seen in the past, just traditional data centers. So with moving the UPS actually to the medium voltage level, we actually going to optimize the system way more, and it's actually a really big advantage specifically for the AI data center. So for these data centers designs. So actually, I would even say we going to see a higher demand towards such solutions. There will be also some buffering and some energy storage on an 800 V system or either very close to the rack or even on the system level.
That part has then maybe multipurpose functionality with, to some extent, also smoothing out the low profiles. The training profiles we know already they're going to be super dynamic. These dynamics you do not want to bring into the grid. So that needs to be smoothing out, and we try to avoid to have that throughout the system also for stability reasons. But really the UPS functionality on the medium voltage we expect to see this even more with the AI data centers.
I promise two more. Yeah.
There were already many two.
Many two. So here's from Will. Is there any extent of cannibalization of your own business with the data center customers in this technology shift? Which part of your portfolio could be impacted adversely by the transition? You've sort of touched on it, but still.
Yeah. I think that, for example, there will be something that it will, let me say, be reduced. For example, we have low voltage UPS. We have not the market leader for sure in low voltage UPS. We have some low voltage UPS that we know that it will disappear. But in the meantime, we will have that, for example, the solid-state transformer that is going to be a super important piece that it will grow, or we will have the low voltage DC UPS that is going to be used, or the TRU that is going to be used. There will be, for example, that there will be low voltage switchgear in AC that will disappear or that will be reduced, let's call it in this way, because AC will not disappear. That's one of the thing that we want to say. AC will not disappear.
Will remain, is going to be more hybrid AC and DC solution. But let's say in the DC application, there will be the low voltage switchgear AC that is going to be reduced, but we will have low voltage DC switchgear that we are part of the portfolio that will improve. We are not so exposed on PDUs. We are not so exposed on RPPs. This, it will not be a significant, let me say, disadvantage.
I would even say irrespective of this technology shift, if we look how data centers have been evolving, we are not using exactly the same components in today's designs as we have used six years ago.
There you go.
There is a constant evolution. That's nothing new for us, I would say.
Yeah.
For me, I would say business as usual.
Yes.
Now one more from Richard. Your comments suggest that the portfolio can speed up the deployment time for new data centers. Is that correct?
Indeed.
Is it at all possible to put numbers on that time saving?
Yeah. Saying the time is difficult. But, for example, if you just think about, we did a study, Massimiliano, I think that you will mention, for example, making the utilization for customer that are using our medium-voltage UPS, like Applied Digital. That is one that is using intensively, the medium-voltage UPS architecture. They are saying that they could reduce, if I am not wrong, around 30%, the times needed to install a data center. On top of reducing the footprint because they will not need a lot of devices, they could have module up to 50 MW instead of having modules much, much lower, right, Massimiliano?
Yeah. Also to build on this, I think that as we said, we are very confident, we are super excited the market will grow. Also the technology change will happen, so the DC will happen. But the fact we have a portfolio that is made by building block, we can follow our customer at any speed they will go. So there will be customer in certain geography with certain dimension that will decide to go a little bit smaller because they are more traditional, and some other that will be super fast. They will be early adopter. Thanks to our building block, we can really partner with all of them so that we can optimize our opportunity, but also help them to develop very fast.
Yeah.
Okay. That is the two questions again. We close it here. Thank you very much for joining us today. If you have any questions, reach out to Investor Relations, and we will do our best to help you with the good support from the guys here. Thank you very much.
Thank you.
Thank you.
Thank you so much.
Thanks a lot.
Thank you.
Industries are the beating heart of our world. They move us, supply us, shelter us, power us, connect us. They improve our homes, our cities, our environments. Behind them all is ABB. Electrifying. Automating. Partnering with our customers to keep them not just running consistently, but running more productively and more efficiently, to constantly outperform. At ABB, we call this Outrun, and it is how we help industries become leaner and cleaner. ABB, engineered to Outrun. Industries are the beating heart of our world. They move us, supply us, shelter us, power us, connect us. They improve our homes, our cities, our environments. Behind them all is ABB. Electrifying. Automating. Partnering with our customers to keep them not just running consistently, but running more productively and more efficiently, to constantly outperform. At ABB, we call this Outrun. It is how we help industries become leaner and cleaner.
ABB, engineered to Outrun. Industries are the beating heart of our world. They move us, supply us, shelter us, power us, connect us. They improve our homes, our cities, our environments. And behind them all is ABB. Electrifying. Automating. Partnering with our customers to keep them not just running consistently, but running more productively and more efficiently, to constantly outperform. At ABB, we call this Outrun. And it's how we help industries become leaner and cleaner. ABB, engineered to Outrun. Industries are the beating heart of our world. They move us, supply us, shelter us, power us, connect us. They improve our homes, our cities, our environments. And behind them all is ABB. Electrifying. Automating. Partnering with our customers to keep them not just running consistently, but running more productively and more efficiently, to constantly outperform. At ABB, we call this Outrun.
And it's how we help industries become leaner and cleaner. ABB, engineered to Outrun. Industries are the beating heart of our world. They move us, supply us, shelter us, power us, connect us. They improve our homes, our cities, our environments. And behind them all is ABB. Electrifying. Automating. Partnering with our customers to keep them not just running consistently, but running more productively and more efficiently, to constantly outperform. At ABB, we call this Outrun.