Infineon key visual in RGB

Protecting AI power with solid-state circuit breakers

We’re bringing you the latest from the world of semiconductors – straight to your ears! From quick takes on trending applications to deep dives on product innovations, our experts give you their take on the tech behind the tech.

Podcast

Aug 25, 2026

Podcast

AI data centers are driving power demands to levels that are forcing engineers to rethink how electricity is distributed and protected. In this episode of Podcast4Engineers, Peter Balint talks with Infineon distinguished engineer Leo Aichriedler about the shift from traditional AC systems toward high-voltage DC distribution and the role of solid-state circuit breakers. They explore why semiconductor-based breakers can isolate faults up to 1,000 times faster, how higher voltages and increasingly complex power grids affect protection systems, and why speed, reliability, selectivity and power density matter. The conversation also looks ahead to commercialization, standardization and the rapidly evolving role of solid-state protection as AI data centers accelerate the adoption of DC power.

Listen on Spotify Watch on Youtube

Speakers

In this episode of the Podcast4Engineers, host Peter Balint speaks with Leo Aichriedler, an engineer in the Application Marketing Organization at Infineon.

Peter Balint

Host:

Peter Balint has shaped visual and audio narratives at Infineon since 2021. He’s a video producer with 20 years of experience and has produced podcasts for the past 10 years. Over his career, Peter has interviewed speakers from all over Europe, bringing high-quality media production and engaging conversations to the forefront of his work.

Leo Aichriedler

Guest:

Leo Aichriedler is a seasoned expert in power semiconductors with over 25 years of experience. He holds a Dipl.-Ing. in Electrical Engineering from TU Graz, Austria, and began his professional career at Siemens Halbleiter in 1998. Over the years, he has worked in various positions, ranging from analog circuit design to power semiconductor technology. For the past five years, he has served as a Distinguished Engineer in the Application Marketing Organization, driving innovation and delivering cutting-edge solutions for high-voltage solid-state power distribution.

Additional information

More episodes

Guest: The most important thing that you need to improve is the speed, the fault isolation speed. And there we do need substantial improvements by almost a factor of 10 to the power of 3. It's 1,000 times faster fault isolation, and that can no longer be achieved with electromechanical components. You need to introduce semiconductor solutions. These are then called the solid-state breakers.

 

Host: Hello and welcome to this episode of Podcast4Engineers. It's the podcast you just have to listen to if you're interested in what's going on in the semiconductor marketplace. I'm your host, Peter Balint, and today we are joined by Leo Aichriedler, who is a distinguished engineer here at Infineon. Mainly working in high-voltage solid-state power distribution and specifically the solid-state circuit breaker, which is what we're here to talk about today. And in our series of We Power AI podcast, we've been looking at different systems which deliver clean and reliable energy because we have this situation where AI is using so much of it. But today we look at the circuit breaker. And of course, something is changing here too. That's why we're here to talk today.

 

Guest: Absolutely.

 

Host: So what's pushing this change with the circuit breakers in the modern AI data centers?

 

Guest: Well, if you look at the past, in the past, all data centers were basically using the traditional distribution scheme, mainly based on AC systems and for sure having a critical infrastructure distribution system. It's basically power coming in from the grid and then getting distributed to the loads. And for sure, there was like an uninterruptible power supply as a backup system, but there were some local gensets or whatever, but it was all AC. Now, obviously, with AI coming into the game, the power demand is increasing dramatically. So now we need multiple sources to supply the data centers. And they have a different nature. There might be for sure the grid, but there might be like PV infits, there might be large-scale battery storages to overcome the peaks, there are backup units and the like. And this means we have to support multiple sources and connect them to these loads. And the sources can be of dual nature. They can be like AC loads, AC supplies from the grid, but they can also be of DC nature like the PV infeed or the batteries. And we need to blend them and supply the loads. And the next thing that is also interesting is the servers, they are for sure DC nature. They need DC. And since some of the sources are also of DC nature, it makes sense to distribute DC to avoid multiple conversions, DC to AC, AC to DC. It's just more efficient if you distribute DC in one domain. And another factor is for sure the distribution voltage. So far, typical voltages were like 230 V or 400 V AC. Now, if we go so high with the power levels, it makes sense to also increase the voltage because the current is just going insane with these relatively low voltages. You want to increase the voltage level as well. And there it makes sense. We do see that like voltages in the regime of 400 V DC are introduced, but some are already going into 800 volts DC. In some other segments, in some other parts of the industry, we already see like 1,500 V. to max out the low voltage regime. And some are even exceeding that low voltage barrier, I would call it, going beyond the 1,500 V into the medium voltage range. So now with this change we achieve, we make sure we can distribute a lot of power. And by reducing the currents, we have to use less copper. Copper is a very precious material, and you want to get the best out of the copper that you need to employ. But on the other hand, you want to reduce the effort you have in routing the power to your load. And there's one connecting element. If you have these multiple sources, you have your loads, for sure there's the copper in between, but the connecting element, the smart element, is then the circuit breaker because you need to connect, you need to isolate. In case of a fault, you want to isolate the individual sources. They can also fail. And you want to have a means to protect the downstream loads and/or isolate faulted downstream loads to propagate into your distribution system. And so now, as the distribution system changes so much and the power that we— the nature of the voltage that we distribute is changing so much, we also need to adopt the protection scheme. And in general, one can say the most important thing that you need to improve is speed, the fault is isolation speed. And there we do need substantial improvements by almost a factor of 10 to the power of 3. So it's 1,000 times faster fault isolation. And that can no longer be achieved with electromechanical components. You need to introduce semiconductor solutions. These are then called the solid-state breakers.

 

Host: This is the only solution to handling this kind of—

 

Guest: Seems so far. I mean, there are hybrid solutions for sure. As always, when you have this huge transition, there's hybrid solutions as well that combine semiconductors with optimized electromechanical actuators. But when it comes to high-power distribution systems like we do see in the AI data centers, we predominantly see solid-state circuit breakers introduced.

 

Host: Okay. Now, circuit breaker technology is nothing new. Obviously, it's been around for a decade. And you have said that solid-state circuit breakers revolutionize power distribution schemes, however, both technologies will coexist. So tell us what you meant by that.

 

Guest: Well, if you look into the electromechanical breakers, it is amazing. This technology has been actually patented about a century ago, a real century. The original patent of this electromechanical circuit breaker was filed in 1925. In the German Reich, and since then it has for sure seen a lot of evolution. But basic technology is still the very same. Whatever you find in your household, whatever you find in in some most industrial installations, AC installations, is still based on that very scheme, which I think is impressive. With the solid-state breakers, we now enter a new era. I would actually call it a revolution because. They enable much faster tripping, much more accurate tripping, and we enable a much lower overshoot. In the DC grid, we have a very, very fast rise of fault current that can rise like 400 to 500 A per microsecond. And a solid electromechanical breaker takes like 10 milliseconds, 20 milliseconds to interrupt the fault. Now multiply 10 milliseconds by 500 A per microsecond. That is a tremendous amount of current. With a solid-state breaker, we interrupt within microsecond regime. You barely see any overshoot. In the end, it's totally unspectacular. If you apply for a solid-state breaker, there are no sparks, there is no smoke, there is no explosions anywhere. There's not even a click. It's totally unspectacular. And that's exactly what a solid-state breaker should be.

 

Host: Is this because it's happening so quickly?

 

Guest: It's happening so quickly and there is no huge amount of power that you see there. You don't see hundreds of kiloamps getting interrupted because we interrupt exactly at where the fault current is. If the nominal current is like 100 amps, we would eventually see a maximum current of 500 amps, while with the electromechanical breaker, we would see 200 kA. You can imagine if you want to interrupt 200 kA, there will be sparks, there will be smoke, and you have to avoid that. This is not acceptable. especially in the data center environment.

 

Host: Okay.

 

Guest: And this is then wear-free. That's also amazing that you can do hundreds of fault interruptions because this is wear-free switching. And the key benefit therefore is that you can use the circuit breaker also for additional functionality like actuation. It's possible and it's also being used. This is obviously substantially different from all legacy implementations. Now the question is, will each and every socket now convert to a solid-state breaker because it's so cool and so more capable? Eventually in the future that might happen, but this will be the far future. It's clear that a solid-state breaker is definitely not cheaper than electromechanical one. And there's also a wide range of different requirements, especially in the AC domain. What we do expect is that for, especially for high power, high voltage DC distribution, we will predominantly see solid-state breakers being applied, while in the AC domain, it might be some niche markets that start with the adoption. We already see that. And then over time, as technology becomes more available, we will also see more sockets in the AC domain.

 

Host: Okay. Now, my burning question is when we say solid-state, we think of nothing mechanical, right? There are no moving parts. But this can't be the case in solid-state circuit breakers because you do need this isolation between 2 conductors.

 

Guest: You got it. You seem to be an electrical engineer. Yes. In some implementations, it might be full solid state, especially in extra low voltage as we used to use in the data centers, like for 48 volts. You don't have the need for galvanic separation, not necessarily from a safety perspective, right? But every trained electrician knows that you don't want to touch a live node even though it says off. there will be some leakage currents. And especially with the capacitive nature of such a DC grid, that would be super dangerous. Every standard basically states that you have to have a galvanic separation and semiconductors cannot provide that. You need to add an air gap device like you have with the electromechanical breakers as well. They would open and then you have a galvanic separation. The big difference here is that now if you look at the distribution of workload, we have a semiconductor switch and we have this air gap device in series, that electromechanical actuator doesn't have to do the heavy lifting anymore because the current will be actually decayed to zero by the semiconductor and then you open the air gap device. That's also why we see a lot of players now in the market coming out with optimized disconnectors, because you don't need that fancy arc chamber anymore. You don't need all the kiloamp switching, the multiple kiloamp interruption capability. You just need to provide an air gap that can be made much easier, much simpler, much more optimized than using this very large-scale burning chambers to handle the arcs.

 

Host: Which would make the device more durable down the road.

 

Guest: It makes it more durable. It also makes it less resistive. We shall not underestimate the resistance that comes with those disconnectors. These are also electrical electromechanical contactors that have some resistance. You can optimize those contacts to have less resistance, and you can make them smaller, substantially smaller. And especially in the data center market, we see that power density is the key deciding criteria. You want to keep this contact, you add those breakers, they cannot be huge. And the smaller they are, the more appreciated it is. That's why it also makes sense to work on those disconnectors and optimize them for solid-state power distribution.

 

Host: And you as an engineer in this field, you're in a key position to talk about what might be the challenges. So I throw that out to you. What are the big hurdles that have to be overcome, or what do you see on a daily basis?

 

Guest: Well, yeah, one might say this is a super boring application, right? You're working on a thing that is sitting in a cabinet, you don't see it. It is on 24/7 for 15 years and there might be no single switching action. Compared to my colleagues that work in the power conversion business, it's a sad application somehow, right? But in that 1 microsecond this fault happens, it must do its job, right? And it's fully exposed. It does not have any input protection, does have no output protection. Obviously, the one time that it has to do its job in its lifetime, it's important. And if this thing fails, it can lead to substantial damage. It can lead to a blackout in the best case. It can lead to a fire. And subsequent propagation of that failure might also lead to harm to people. It's a protection device in the end. It's obvious that You also want to keep the conduction losses of that semiconductor that we put in series now as low as possible. On the one hand, we need to optimize the technology that is as low as possible, and it needs to be super reliable on the other hand side. And it needs to cope with that boring application environment just sitting there and not having to do anything. So that's a huge challenge for semiconductors as well. And that's also why we, like 4 or 5 years ago, we decided that we do need to develop dedicated semiconductor technologies, also packages that cope with these very particular requirements. And that's what we are currently working on. And that's also what we are already rolling out already on a first scale.

 

Host: And earlier you touched on this concept of moving higher voltage or bringing higher voltage into the rack. What does this mean for solid-state circuit breakers? Do you need more, fewer, faster? How do you compensate?

 

Guest: Yeah, it's not just the use of the increase of the voltage level. That's one thing, right? We are now rolling out a full distribution scheme, right? In the first implementations, we do see it's like a point-to-point connection. It’s easy there. You have one source, you have one load, you put eventually a breaker in between if the connection fails or whatever, also for some protection. of the source, for instance. But now what we do see is the next stage is that you really go into a grid. You have multiple sources that combine to one distribution rail, and then you have multiple loads that tap from that distribution scheme. You need a hierarchy of breakers, and that is driving it. This is breathtaking as it's now being rolled out. We know that same distribution architecture from AC systems. And that's now also being rolled out for DC systems. Although you can imagine that if you have this very fast rate of rise of the fault current and very fast tripping breakers, it's an enormous challenge to also make sure that those breakers are selective one versus each other. You don't want to have one small breaker having the job to trip, but it also would trip the upper one and the main one, and you also go into a blackout. You need Despite being so fast, you need to achieve selectivity. We need multiple sizes of breakers ranging from like 30 amps on the individual rack level up to many kiloamps on the main breaker level. And yeah, that's the key requirement. They still need to be super fast. You cannot compromise on reliability, and they need to work together. So yeah, the more complex these power distribution schemes are designed, the more breakers will be employed. And that's what is currently going on. They need to be faster, they need to be enormously capable, and they also need to be affordable. I think that's also an important part that you want to take the benefits for sure from solid-state technology, but it also needs to be commercially viable to be applied.

 

Host: So bottom line, it sounds like things have just become really complex and where we used to have a linear system, it's now become a grid.

 

Guest: It's becoming a grid in the end. And we like, we as engineers, we like that complexity, we tackle it. And I think also with the products that we are now rolling out, it is demonstrated that it works. It's definitely cutting-edge technology, although it doesn't look so sexy from the way it's working. When it's working, it is very interesting how it works together and to make this thing reliable and robust and available as well. And that's what is being rolled out also together with our customers. We are working very closely there. And yeah, we'll see first systems in place by end of the year in a high-power data center. And then looking forward to how this gets then rolled out into the breadth of the market.

 

Host: And this sort of leads me to my next question, which is commercial viability. It sounds like if we're moving forward with, with these systems, then we have to have these solid-state circuit breakers. Is the market welcoming this? Are they concerned about costs or is this—

 

Guest: It would not be the market if they were not concerned about the cost. And that is clear. And we know the cost of legacy AC systems, right? We know the benefits of DC distribution. And it's clear that a breaker where you add a solid-state semiconductor, high-voltage solid-state device in series to a contactor will not be cheaper. I think that's nature. What we do see is, as with every new technology, it's a matter of novelty. It's a matter of volume, also experience in the market, and also a matter of standardization. What do we have to cope with? What do we have to meet in terms of requirements? According to which standard will we certify such a breaker? Can we take out the overhead, that is probably, that is currently associated with a lot of standards that basically result from electromechanical breaker standards? They just have been derived a little bit, but still they're based on the electromechanical requirements. That is a huge cost burden as well. And I think the more that technology gets rolled out, the more it gets used, the higher the volumes. In semiconductors, it's all about volumes. Also, the more commercially viable will that be. That's why I mean that at the beginning, we'll predominantly see the solid-state breakers being employed in DC systems where the value proposition is enormous. It's a must to apply them. And then we'll also see that we'll more and more see them being applied in AC systems where eventually the value proposition you only have in some niche markets where you need particular requirements. But as the cost goes down, then it also becomes commercially viable. It gets very competitive also versus electromechanical solutions.

 

Host: So now for our listeners who are interested in this topic, is there anything looking forward, looking into the future that we haven't talked about that might be key points when we discuss the solid-state circuit breaker?

 

Guest: Yeah, this is hard to say what happens in a revolution, right? If I knew that, it would be easy. Obviously, what we do see is, as already mentioned, we see the voltages going even higher, going even beyond the limit of what is considered the low voltage regime up to 1,500 V, especially in the context of solid-state transformers. We also see some first customer projects that even deal with voltages as high as 34.5 kV. Now imagine you do a solid-state circuit breaker that has at least a blocking capability of 60, 70 kV. That is an enormous technological challenge. Also, isolation coordination-wise, we see a lot of things coming up there as well. I think on the standardization side, we have high hopes that we'll see dedicated standards that address the particular aspects of solid-state breakers, also the particular requirements in a DC grid. So far, there's not a big delta in the standardization between AC and DC, although those grids are substantially different. I think that's one thing. And on the customer end, we do see— Like in every revolution, there is a couple of customers that cope with that change, that address it, that even drive it. Some others are reluctant. And we'll see what the outcome is of that revolution. And that also gets new players into the market. They're just evolving. They eventually do not even have legacy products. For them, it's a new reason to enter that market. And that is also very interesting to see how that market is now evolving. It was there because the market is huge. It has been there for 100 years with the same thing, everybody optimizing it to the last cent. And now we go with something that is of high value, protecting enormous assets in a data center. And yeah, so I think it's very interesting to see. And let's see in 2 years, we can talk about what happened. And this market is so super fast that I think in 2 years it will be a completely different picture.

 

Host: Yeah, it's mind-boggling. I mean, it's something that's a technology that's been around for 100 years, but it's never ever moved so fast in the history of—

 

Guest: Exactly. Yeah. It got an impulse by DC distribution. Now, DC distribution is used in a lot of trial applications, also in other fields in the industry. Now, obviously, with the data centers embracing this technology, it gets an enormous technological push. And it also accelerates the rollout in other markets. I think that's why it is super amazing to see how things are developing. And it also got enormous push for sure also to make huge investments in product and technology development, which doesn't come for free. It also needs to be justifiable to make this effort.

 

Host: Thanks for taking time to talk with us today on the podcast, and we look forward to talking with you again in the future.

 

Guest: You're welcome.

 

Host: And thank you for all who are watching and listening. We hope you enjoyed the show and we'll see you soon. Thank you.