
Get every episode summarized
Each time Open Circuit publishes, we email you a written briefing from the transcript — the topics, who appeared, and any specific claims, with the ad reads skipped.
Email me new episodesFree for 3 shows. No card needed.
About this episode
Open Circuit is made possible by:
“This is Partner Content from Latitude Studios. Kishin Punnaderi and Kameenburg spend their days thinking about the hidden hardware that makes clean power work. Things like recombiners, load break disconnects, and cable bus systems.”From the transcript
Get every episode summarized
Each time Open Circuit publishes, we email you a written briefing from the transcript — the topics, who appeared, and any specific claims, with the ad reads skipped.
Email me new episodesFree for 3 shows. No card needed.
Transcript ready
219 searchable segments. Every word is indexed and playable.
Full transcript
Open Circuit — The hidden power systems behind AI [partner content]. Machine-transcribed; use the interactive transcript above to jump the player to any line.
This is Partner Content from Latitude Studios. Kishin Punnaderi and Kameenburg spend their days thinking about the hidden hardware that makes clean power work. Things like recombiners, load break disconnects, and cable bus systems. Their world is highly technical, but as in what they do, they make it sound deceptively simple. I build product that allows the movement of energy very simply put. Mostly I tell people I help sell electrical equipment out in the market. They work at Sholes Technologies Group, which has spent 30 years building the electrical systems that move power from solar fields and more recently through big battery projects. Kishin is a product manager for battery storage. His job starts when a customer shows up with a power problem. What we're trying to figure out is what do they really need from a DC distribution perspective, whether it's protection, aggregation, isolation, so on and so forth.
Kameen works on the other end of that process. He leads business development for battery storage and data centers, talking with customers about those problems, and bringing the interesting ones back to Kishin. And then on the flip side of things, if an opportunity comes my way, I'll mall it over, but then I'll typically bring it to Kishin. Kishin will help me determine whether or not it's really something viable that we want to do for our business. We're trying to figure out what problems people are experiencing. So it's not just, hey, what sells or what can we build? Hey, what are you guys facing that people aren't seeing or that we're not seeing? Lately, the most interesting problems are coming from data centers. A market that is moving so fast the power design choices are constantly shifting. Every day, there's another opportunity to look at something new. Somebody thinks they have an idea of what the solution looks like. They can come to us and then we can decide whether or not that is a solution. And then in three months, they'll tell us they completely change their mind and the architectures totally different, but we'll adjust to that too.
It's a true story. That's what I'm laughing. When we hear data center, we tend to picture giant buildings, rows of linking servers or massive cooling equipment. But an entire electrical ecosystem is sitting between a data center's power source and GPUs. And right now, there is surprisingly little agreement about what it should look like. When we say it's the Wild West, what does it actually feel like from Sholes perspective? Think hectic's a good way to put it. I mean, it's growing crazy fast. Stephen, I was just going to say you're a runner now, right? So you buy a pair of shoes and then in three days, you need a new pair of shoes. That's what it feels like right now, working with all these customers. It's the Wild West because there's a lot of solutions to a problem. And we're at the infancy stage right now. Which one is the right one? And that's what we're all trying to figure out together right now.
Big questions about how power should run through data centers are still up for grabs. And the answers will determine how much equipment is needed, how quickly it's built, and how much floor space is left for servers. All the customers that we're looking at right now are doing something different. And that's not an understatement. Everything that we're working on right now is different. You know what is indifferent? D.C. distribution. And that's why we're in a really cool place. After three decades of building D.C. systems for solar and later batteries, Sholes consider storage for data centers as the next frontier. The opportunity is in some instances maybe even more significant than what we're seeing in the utility scale space for battery storage. Batteries and data centers are changing fast. They used to have one job, keeps their resolve until backup generators kicked on. But today, they're being used to smooth out volatile loads, support onsite generation, and buffer the grid.
And that changes the power system around it. In this episode, producing partnership with Sholes Technologies. Stephen Lacey talks with Kishin Pona-Durai and Kamen Burke about the expanding role of batteries inside AI campuses, the shift from AC to higher voltage D.C. and how to navigate a market that is constantly changing. A lot of times, data centers need to move quick. And that means the vendors, the manufacturers also need to move quick. That's what this industry is. I want to dig deeper into batteries now. So batteries have been in use in data centers for a long time for UPS systems. Kish, what is different about the way batteries are being deployed now? The role they're being asked to play, and ultimately how does that feed into product decisions? Yeah. You take a look at batteries historically, within data centers. And typically what you're seeing is a UPS room, right?
It's a room that exists on the data center site. It's inside the data center or inside a room on a data center. And typically, its use case is for very minimal things. It's to take on the initial downtime so that a diesel generator could get back up and running and provide the brunt of the power that the data center needs. The problem with this is that it could never live up to the scale at which data centers today are being built, specifically AI data centers. So keep in mind that a UPS room and a diesel generator still has presence in the market just for very specific data centers. When we're talking about the data centers that are making headlines today, those are not your average data centers. Those are large gigawatt scale data centers dedicated for artificial intelligence. And I think one thing that is a problem that we've experienced in those data centers is how volatile the power draw is.
So your typical UPS room provides two problems to this new way of thinking about data centers. A, it's taking up room for more compute. You could use that for a couple more racks more than a couple more racks, right? But racks or racks capacity is capacity. The second problem is they're too small to withstand the volatility of the power draw in AI data center. So that it couldn't even take on the buffer for a diesel generator to really turn on. How we're looking at batteries today is scaling those batteries and putting them outside the data center more similar to what you would see in a utility scale energy plant, right? Very similar concept. Diesel generators still can be in play. They can be used as a secondary, but the battery outside is where we're all going. Well, if we're moving batteries outside, what I would bet on is a technology that's able to increase power density immensely, meaning how can you inject as much power via batteries while keeping the footprint smaller?
You know, taking a look at energy storage now within data centers, it's no longer just an afterthought. It is we're going to need this to handle the volatility of an AI load. It's going to need to be a buffer for a grid because the grid and the diesel generator itself cannot take on the volatility of the AI load. And that's kind of where we're at now. All these different battery chemistries, all these different ways of installing batteries on a site use cases on how you're using inverters. All these different concepts coming into play. It's no longer just, hey, maybe we can add batteries. It's we need batteries. Yeah, I came in any commonalities among what you're hearing from customers like when they come to you and they're figuring out how to deploy a battery, what it's going to do, what are they actually trying to solve? They're trying to solve exactly what Kish is talking about. There's the data center side of this where we're talking about immense power draws sometimes to what Kish alluded to.
We're talking about immense power drops, which the grid obviously can't handle. That's why we see the right through requirements that we have now. It's just we can go back to just general generation and the stability of our own grid, right? I live here in Nashville and we have had some very serious winter storms the past couple of years. We had a huge ice storm this past winter, 300,000 people were without power at one point. The couple of years before that, we had an immense ice storm, just general winter storm where there were rolling blackouts. And so the grid needs batteries and the grid needed batteries prior to this data center eruption, right? And so when customers come to me, they're typically trying to figure out a few things. A, if I'm talking about just power developers, they're probably focused on how do we put just batteries online and put them on the grid.
But if I'm talking to somebody a little bit tighter to the data centers, they're trying to figure out where does my battery actually fit in the system. And so you can talk about the generation side of the data center or you can talk about the medium voltage, UPS side of the data center. And those today are typically two different applications. Yeah, what's what kind of challenges does that create if those teams are not talking to each other? To me, it seems like a missed opportunity for the power developers to not be talking to the medium voltage, UPS folks directly, right? Because let's say we have a data center and let's just say for clean numbers, we have a gigawatt. Maybe there's an announcement about it having 250 megawatts worth of battery on it. But that data center is probably going to have two full gigawatts of battery in a UPS system. What I do know is that a UPS system, maybe it utilizes a one hour battery, but I know that a two hour battery doesn't cost twice as much.
And so why not utilize some of that space maybe to put in something like a two hour battery or four hour battery that can be utilized by the power developer. And you just keep a certain amount of that in reserve for those backup power needs. There's an interesting concept here, right? Like you always think of behind the meter in front of the meter. It's in my opinion, I mean, without with how big these these sites are getting, you know, porque no los dos, right? Why not both? Because in this case, right, you're you're you're building a gigawatt worth of power, right? That's bigger than a lot of these utility scale sites that are coming up in a lot of places. Well, why aren't we looking at that at that site as a as a means to help the grid more than just, you know, more than just a buffer really. What can we do more with it? I'm kind of curious. What your observation is, you can't show about the types of teams that are working on power infrastructure decisions, you know, they're building power infrastructure fit for a small city.
Many of them, I think, have probably very sophisticated teams, some not so much. Tell me about the types of people that are actually making these decisions and the types of problems that they're being asked to grapple with, maybe sometimes for the first time. So, you know, the data center ecosystem is quite interesting, right? It's you have you have the guys that are that are financing these big projects, right? Those are the the hyperscalers that you hear of today. Then within that, you have the people who are contracted out by them to do the construction. And the people that are doing the construction are not experts in everything. So they'll go and piece out. Here's what we want to do for this data center. Here's the energy section. Here's the white space section. Here's the gray space section. And they'll go and find these different vendors that that are very specialized in these types of things. And that subcontracted out work turns into 10, 15, 16, 20 different subcontractors all working on very specific components of this data center.
You have people from different walks of life industries coming together to work on to work on these problems. Going back to kind of what relates to us is well, you know, we're obviously part of the energy infrastructure. And we work very closely with other companies that offer larger energy solutions, but they need a very specialized component. And that's how we fit into that. You look three steps adjacent to us and you have 10 different people working on the piping, the conduit, the concrete. Anything vertical is different from people that are working on the horizontal. There's so many people working on these data centers. It's actually amazing. It can be tough, not just for somebody on our side, but as a developer owner and operator, right? There's no way they could manage all of those people themselves. I think the thing that needs to be focused on though is making sure that the important decisions and the people involved in those high-level decisions aren't getting siloed off.
And that they're communicating appropriately. Let's talk a little bit about the types of batteries that you're seeing. These data center operators and developers choosing. What are the chemistries you're hearing people talk about and consider? Yeah. The three main that we've been involved with, at least have had tons of conversations around, are going to be iron air batteries. Another one is the sodium battery. The sodium battery has been a topic of discussion for a very long time. But I think we're starting to see some newer developments on salt batteries that can, I would say, better compete with the traditional lithium ion. And then we're also looking at redox batteries or in other words, a vanadium flow battery. And all these batteries have different installation methods, different footprints, different capacities. They all offer the same solution, though, overall, which is power generation.
So I think as we start to look at all these infrastructures, all these different battery technologies and chemistries, it's learning more about how they're streamlining their production. How they're making it repeatable so that you can scale it. Just keeping track of all these different technologies or chemistries, I should say, to see how they can compete when it comes to scaling as large as lithium ion. Yeah. So how does that change the sizing of the system, the voltage? These are all very different size and duration batteries. So how does that feed into the actual components and architecture itself? Yeah. So I mean, when you're talking about iron air and vanadium flow, you're typically talking about a much larger footprint for that battery to equal a smaller footprint of lithium ion. What that means is you have to really allocate square footage on your data center site to have that capacity.
There's the discussion of building up versus building horizontal. So that's one route that people are looking at, stacking batteries. But overall, how you fit your footprint on the data center, that needs to be talked about with these developers that are building the data centers. The one thing I will say is that these specifically, the companies that we've been talking to on these new chemistries, they all are operating in DC batteries. So they're DC out batteries. And I think that's by design for power quality. And as we move into 800 volt DC architecture, 1500 volt DC architecture, they're ready to go. So does it change the architecture of energy distribution, not entirely, but does it change how developers have to plan for battery generation on their site immensely? Actually, to that point, can you talk about how electricity flows into a data center and through the data center and what conversion looks like within the data center before it reaches the servers?
So taking a look at distribution of energy for a traditional data center site, it's very different from what we're seeing today. So typically you have DC power generation. That's coming from the battery, coming from wind, coming from solar. That's being fed into a grid, grid substation. What have you? That substation typically has a transformer that will convert that DC power generation into AC power generation. And from there, that transmission of energy can then get sent to the site. And the site should have a means to receive that medium voltage. And from there, the medium voltage transmission will then move to the site where then you will want to move that from a medium voltage to a low voltage to where your data center can run on that 480 volt AC. And so you're going through a couple conversions to get there. And then inside your your UPS room, if you are running UPS batteries, you have rectifiers that will change that DC voltage to AC voltage.
And then there are a couple conversions that are happening there. And a new data center, what we're looking at today and with batteries being external, you're still going to have your medium voltage balance system to receive the power from the substation. But now we're feeding it through inverters, we're feeding it through batteries, we're feeding it through a lot of components before it gets to the data center. And what we're doing here is providing a means for batteries to be a buffer for the grid, right? Like I had mentioned earlier, all this to say is in an AC data center, you go through a lot of conversions. And then you look at what an 800 volt DC data center looks like. There's a lot of possibility to where we can cut down those conversions. Conversions offer efficiency drops, right? If you look at a component, like an inverter, a converter, one of those, you're looking at a power efficiency at around 98.7, 90, sometimes 97.5 in that range.
And then you're talking about a gigawatt of power that that 3% or that 2.5% makes a huge difference. And so as we move into DC data centers, if we can cut down the conversion loss or the inversion loss, you can drastically improve your power quality over time. Are you seeing any standardization here around power blocks or is does everyone treat this like their own unique snowflake? Everyone's got their own roses, right? All the flowers look the same from a distance, but you come up on them and they're not. They're not. There are DC blocks, you know, in the solar spaces, an example, inverter sizes, they're pretty standardize around five megawatts. But when we start getting into especially the sizes of some of these, let's call them non-traditional batteries, those DC blocks can range from 500 kilowatts to sometimes three megawatts.
It all just depends how they did their original design. Then we get into the architecture of what everybody wants to do on the inside of the data center. And that's that's also completely different. Like I said earlier, everybody's kind of taking a bet on what this is going to look like, right? Are we are we going AC all the way in and then converting to DC? Are we bringing the DC all the way into the white space and through the building directly to the rack? What does that look like? It all looks like I said the same from a distance, but it is all very different when you get into the nitty-gritty of it. So if we consider this transition to 800 volt DC, what other products are you working on? When we look at switching over to 800 volt and the power distribution aspect of right, we have the busway system that typically sits above the racks today. We believe that we can use that same cable bus system to bring power to the racks to ultimately generate something that's more safe, easier to install and potentially a quicker timeline, right? Because the busway guys are full their capacity and so is everybody else.
But specifically for 800 volt open channel busway isn't going to make a lot of sense for people, especially when we talk about safety, the biggest concern being arc flash. And so by utilizing something that is already out in the industry today, we deployed almost 100 gigawatts worth of that material out into the market and is actually already being used in a 1500 volt ecosystem. We think we're bringing a benefit potentially to the market that eventually could connect both systems and parts of the product, right? If we have our power distribution outside with our medium voltage UPS, cable is going to get fed in and then potentially even a DC PDU system that would feed out into this what we're calling air length system as well. Yeah, I think the idea here is that we have a suite of products right now focused on DC distribution. The power hub can be built to work with a UPS system, but it can also be built to distribute power to many air length systems at the same time or a singular air length system depending on how much power this the rack would be.
So as we look at the ecosystem for Android volt DC and 1500 volt DC, we're ensuring that we can take more of that distribution products week. So you guys are you have a front row seat to many of the decisions that are happening inside these data centers. If we move out a few years to the end of the decade, what pieces do you think will change the most over the next few years? Generally, I think the footprint of these data centers is going to change significantly, right? I think one of the interesting things that I'm seeing is when I worked in EV, we went from 48 amp chargers overnight. A lot of people still use them today, right? You can just sit in your garage all night and it'll charge up appropriately. But we were using those in commercial applications and within it felt like a year and a half, maybe two years, we hopped up to a megawatt charger in discussions around those.
That's what's happening today inside of the data center. And so the difference is we're going to be able to do, you know, what's 10 racks and 100 kilowatts today can suddenly be done by one megawatt. One megawatt rack, right? And there's other questions around are there diminishing returns, right? Do you replace that entire space, right? Do you put 10 one megawatt racks in back into that road to fill up that space or do you suddenly only need three? And I think the answer is you will only maybe need the three. And so we'll get smaller and smaller. In order to do that, though, from an architecture standpoint, you are going to have to see this transfer over into DC. And so 800 volt is going to happen. It's happening right now, I should say. You see announcements about it all over the place. What I think will be the next step will probably be 1500 volt, right? The renewables industry has been set around 1500 volt for years now.
And ironically, we've talked about on the renewal space 2KV. And so you may see as renewable steps over to 2KV, the data industry might say, why don't we do the same? Quite frankly, the data center world should have probably started with 1500 volt DC as the preliminary discussion. So it's pretty funny that we're starting at 800, but I don't know, maybe another hot take there. Any other changes, Kish, that you see on the horizon, what do you think this market looks like in 2030? Yeah, I mean, look, the demand, like I said, is going to be there. But I think the scary thing, maybe not so scary, but the thing that has my eyebrows raised is the pace at which data center demand, energy demand is growing is far going to outpace the utility scale growth. What I mean by that to be very clear, I think the capacity utility is still going to be very, very high, but the rate at which utility is growing versus data center demand, it's night and day.
And so I'll go back to the discussion of behind the meter, front of the meter, if we're building these sites to be size for a gigawatt 2 gigawatt 2.5 gigawatts, why don't they offer services in both fronts? Well, it is certainly a wild time. And I know you guys are right up close and personal to it. And I really appreciate your time and insights here. Kishin, good to see you. Yeah, likewise. Thanks for having us on. It's really great to be talking about this with you and just really excited for this industry. Kaman, thank you. Stephen, thanks for having us. We appreciate it. This episode was produced in partnership with Shoals Technologies Group. For 30 years, Shoals has designed and manufactured the electrical backbone for critical energy infrastructure. The company builds simple, reliable solutions for complex power challenges across solar, energy storage, and data centers. Visit Shoals.com to learn how Shoals solutions are helping power data centers.
Thanks for joining us.
More episodes
More from Open Circuit

Power is caught in AI’s doom loop
Open Circuit

Empowering utility customers with demand flexibility [partner content]
Open Circuit

Electricity bills are on the ballot
Open Circuit

The breakout moment for VPPs [partner content]
Open Circuit