Infineon key visual in RGB

Powering AI at scale: grids, HVDC, and what comes next

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

May 12, 2026

Podcast

AI is pushing our energy grid to its limits, but keeping up requires a sustainable approach, backed by the latest advancements in semiconductor technology. In this episode of our We Power AI series, Kelsey talks with Peter Wawer, Division President of the Green Industrial Power Division at Infineon, about what it really means to power AI sustainably, from the grid down to the server rack. They cover why battery energy storage, HVDC, and DC-powered data centers matter, and look ahead to what “physical AI” could mean for industry, robotics, and society.

Listen on Spotify Watch on Youtube

Speakers

In this episode of Podcast4Engineers, host Kelsey Markl speaks with Peter Wawer, Division President of the Green Industrial Power division at Infineon.

Kelsey Markl

Host:

Kelsey Markl is Head of Strategic Marketing Communications and Content at Infineon with more than a decade of experience in communications, content strategy, and storytelling. Not being an engineer is part of what makes the journey interesting. Kelsey approaches technical topics from the perspective of a curious outsider, asking the questions that help make complex ideas more accessible and relevant to engineers and broader audiences alike. Passionate about innovation and the people behind it, Kelsey enjoys exploring the trends, challenges, and breakthroughs shaping the future of engineering.

Peter Wawer

Guest:

Peter Wawer is Division President of Green Industrial Power at Infineon Technologies, a role he has held for nearly a decade as part of his broader 13-year career at the company. Holding a PhD in Electrical and Electronics Engineering from the Technische Universität Berlin, Peter brings deep technical expertise and decades of industry experience spanning Siemens, Q CELLS, and Infineon. In his current role, his vision is to enable a world of unlimited green energy — overseeing a division that spans the entire energy chain from renewable energy generation and power infrastructure to transportation, automation, and home appliances. A recognized thought leader in power semiconductors and the AI energy revolution, Peter is a frequent speaker and commentator on topics ranging from solid-state transformers and wide-bandgap technologies to the future of physical AI and humanoid robotics.

Additional information

More episodes

Guest: And while now, with help of renewables, the energy generation becomes decentralized, the grid has to cope with it. You are able to manage the huge fluctuations which come from time-dependent demand from data centers. Also, in the future design of an electricity supply for a data center, electricity storage system will play a very critical role to handle and stabilize the whole system. What about physical AI? What happens— automation, robotization? We as Infineon are perfectly set up to support these kinds of applications, like for example humanoid robots.

 

Host: This is the Podcast4Engineers, and this is our third series on We Power AI. The next few episodes, we're going to be taking a look at the technology, architectures, and innovations that are powering the AI revolution. revolution. I won't give too much away, but today we're starting with a very special guest, our division president from Green Industrial Power, Peter Wawer. Peter Wawer, thank you so much for being here.

 

Guest: Yeah, thank you, Kelsey, for inviting me.

 

Host: Let's just set the stage for the rest of our episodes. We've looked at powering AI and the technologies, the semiconductor technologies that are behind this technology and the energy requirements, the different architectures inside the data center. But now let's look at maybe a little bit of the big picture. Can you tell us a little bit about where we're coming from right now?

 

Guest: Yeah, happy to do so, Kelsey. I think the growth projection of AI and also the business opportunity that AI generates for us is unprecedented. So, I'm since a couple of decades in this industry, and what we now see as opportunities, I think it's really unprecedented. The key thing is that AI requires huge amount of electricity. And luckily, we are in the power electronics area. We are powering AI from grid to core, meaning we can address all applications that are required to enable AI to grow.

 

Host: Okay. Can you— I mean, you mentioned rise in electricity and rise in the demand for energy. We see a lot of technologies now requiring more and more energy. How do we make sure that we do that sustainably?

 

Guest: I think that's a pretty kind of challenge, right? Because the growth projections are very steep for the electricity demand of AI. And interestingly, the only source of electricity generation has proven to cope with this kind of high growth are renewables. The thing is that renewables are able to deliver the additional electricity that is required by AI. But you know that we also have a challenge because predictability and predictable energy generation, electricity generation by renewables is also a challenge. So, we have to find technical solutions to address this kind of growth. And of course, also the reliability, make sure that electricity is being available in a reliability, in a reliable manner.

 

Host: Okay. How do we compensate for that? How do we compensate for the fluctuations that renewables bring with them?

 

Guest: Yeah, maybe first of all, renewables alone will not do the job. As always, we have to find the right mix, while part of the mix will be renewable, of course. And now looking into what is required, the investments are so huge that we talk about gigawatts and up to tens of gigawatts who will be needed looking forward.

 

Host: For a single data center?

 

Guest: For a single data center, we will think about a kind of energy and electricity power supply of up to 1 gigawatt is doable or is imaginable for a data center. And then if there are many, since there are many data centers built, right, it will add up. So, 10 gigawatts and more are definitely in the scope.

 

Host: Okay. That's a lot of energy. And we said renewables in this healthy mix. Do we also need to look at the grid to kind of compensate?

 

Guest: Exactly. On top of this additional demand, we have today also in certain areas a stress grid because the grid has been built for the old traditional style of energy generation, having central big power plants. And while now, with the help of renewables, the energy generation becomes decentralized, the grid has to cope with it. Giving you 2 examples, right, we know that in Germany, where the electricity generation has exceeded 50% of total electricity being generated for Germany, the grid is already very much stressed. Same similar situation in China, which is the biggest electricity producer of the world. The grid reaches certain limitations. And therefore, you need to invest in the grid to expand the capabilities to handling bidirectional energy flows in the grid. And that is where a huge amount of investment will go in the nearer future.

 

Host: Okay. That's a good example of this bidirectionality. How does this grid stabilization look like then? What does the new landscape look like?

 

Guest: I think a very, very big portion of the stability will be provided by battery, battery electricity storage systems, BESS. This is something where, of course, you on one side have to stabilize the grid, but on the other side, you are able to manage the huge fluctuations which come from time-dependent demand from data centers. So, also in the future design of an electricity supply for a data center, electricity storage system will play a very critical role to handle and stabilize the whole system.

 

Host: Okay. Yeah, you mentioned a lot about the healthy mix of renewables. I think in a previous episode, one of our guests mentioned that we produce, taking the Germany example, a lot of energy up in the north in the offshore wind parks. But actually, where we're using a lot of energy is in the south, in Bavaria, where we're sitting today. Do we have to rethink how we're transmitting our energy to make sure that we meet the demand for AI data centers?

 

Guest: Absolutely. This kind of challenge exists not only in Germany, but also likewise in China, right? There it goes from west to east. And the solution is to invest into high-efficient HVDC transmission lines. A lot of semiconductors again are being used for that one. You have to transform the electricity into a higher voltage and then you're capable of minimizing the losses while transmitting a huge amount of energy and power.

 

Host: Okay. I can actually just visualize all the energy going across the world back and forth. But now that we've kind of visualized the big picture, what does that look like In the little— in the data center ecosystem?

 

Guest: Yeah. Actually, what is true for the big picture is likewise very similar for the data center. A typical data center uses a huge amount of electricity and that in a very volatile manner. Depending on the algorithms that work in the data center, there is a very high time dependence. And not a single source of energy can deliver these kinds of very time-dependent energy peaks that are required. But battery storage. Battery storage is the solution. I do expect a huge amount of DC-DC converters will be needed to then cope with this kind of energy variation and energy fluctuations. And that is true for the big data center overall. And it goes all to the single rack.

 

Host: Okay. So, we're actually moving into the data center building now.

 

Guest: Exactly.

 

Host: Yeah. Okay. Is that— it sounds to me like we're thinking about building little microgrids, I guess, around a data center. So instead of the big connected grid to everything, each data center is building their own kind of independent grid. Is that true?

 

Guest: I think the design of these data centers exactly will evolve in that direction. While today typical energy is being supplied AC and then converted AC to DC, to then power the single rack. Tomorrow, we will see an evolution where first of all, the power demand increases. The so-called switch modes power supplies, which are required to supply these racks, will first of all increase with respect to voltage. But at a certain point in time, that will not be enough. So, you need to then convert the whole data center architecture to a DC grid. So, in the future, until end of this decade, we do believe that AC current flowing from the middle— medium voltage grid with a voltage around 34 kV will be converted directly into DC. And to have this done in a very efficient way, right, we can think about transformers, also solid-state transformers, which are capable to convert this AC to 800 V DC. That is a standard which is currently being defined. And then you have an 800 V DC grid where you can directly connect all these GPUs, basically you need still to convert DC to DC voltage from 800 V down to the voltage level which is ultimately required by the single processor. And that we do expect to evolve in the next 5 years or so.

 

Host: Okay. Yeah, you mentioned solid-state transformers, which I just had a conversation about. We did an episode about solid-state transformers. Are there any other technologies within this little ecosystem that are further developing, popping up?

 

Guest: I think the solid-state transformer is an important piece of the puzzle, but you need then protection features because 800 V DC is a different beast to today's distribution structure. And there you need protection features. Solid-state circuit breakers together with solid-state transformers, we do see evolving. And luckily, we provide the right semiconductor power solutions to also provide these kind of security and safety features.

 

Host: Okay. Yeah. It looks like there's a lot of things happening in this area. From the evolution of our entire grid to actually the evolution of the energy flow within the data center, this development has happened over a very short period of time. But what do we expect in the next few years then? Let's talk about years because decades I feel like is too big of a timeframe for this discussion, right? What do you see coming next?

 

Guest: Yeah, that is very exciting. And remains to be seen, honestly, right? But if I would like to predict a bit what will happen then, I think still one of the big questions, I think still open questions is why everybody does understand the productivity increase with help of AI. What justifies these huge investments? And I think there's also a topic where we start now to discuss with customers, what about physical AI? What happens?

 

Host: What is physical AI? Can you just explain that a little bit?

 

Guest: Actually, how will we then benefit from AI also back to the physical level? How do we interfere with our daily doing, manufacturing operations, and also maybe other areas where we can think of, right? Elderly care, for example, in an aging society. How will we benefit from or how can we benefit from the capabilities of AI? And if you think about autonomous systems, robots, for example, being able to take over human tasks, humanoid robots, which are now evolving, these are tremendous use cases for AI looking forward. And there are already quite some companies out there which are exactly targeting for these kinds of use cases. And I think we as Infineon are perfectly set up to support these kinds of applications, like, for example, humanoid robots.

 

Host: Yeah, because power semiconductors are very close to our heart, but humanoid robots, I imagine, need a whole breadth of semiconductor solutions, sensors and everything. Okay. And what's the thing you're most excited about going forward?

 

Guest: I think it's a mix of, of course, seeing these business opportunities, which are huge. And you mentioned, right, it's not only about power, it's about everything which you can imagine as a semiconductor manufacturer. And of course, the other thing is that it's now really a time of big, big changes, right? And understanding those kinds of trends and of course shaping them in a sustainable and also useful way for mankind, I think it's personally very rewarding.

 

Host: Thank you so much. Thank you, Peter, for being here. And thank you so much to our listeners for tuning in to our third series of We Power AI. Stay tuned for many more episodes where we dive deeper into the servers, into the ecosystem that is really powering the future of our AI revolution. Thank you, Peter.

 

Guest: Thank you, Kelsey.