Latest / Clean Power Hour / Why Grid Power Can't Save AI Data Centers (Microgrids Can)

Transcript

Why Grid Power Can't Save AI Data Centers (Microgrids Can) · Clean Power Hour

• 0:52
  Michael Stadler: I think at the end of the day, and I think already said this, a lot of people are interested, but they're still waiting. I think the sooner you get around the idea of a microwave, the better this for your company. I mean, you can wait a year, you can wait two years, but the problem is not going away, and I think it's time that people start realizing that they have to build some internal knowledge at the company so that they can deal with this idea of a microgrid and actually build it and maintain it, because it's a little bit more complicated than just purchasing the electricity from utility. So you need internal knowledge, and I think it's the sooner someone can start actually building that knowledge, the better it is for the business and your business case, because it's just otherwise you're just deferring the problem.
• 1:40
  intro: Are you speeding the energy transition? Here at the Clean Power Hour, our host, Tim Montague, bring you the best in solar, batteries and clean technologies every week. Want to go deeper into decarbonization.
• 1:51
  intro: We do too. We're here to help you understand and command the commercial, residential and utility, solar, wind and storage industries. So let's get to it together. We can speed the energy transition
• 2:07
  Tim Montague: today on the Clean Power Hour batteries and micro grids. My guest today is Michael Stadler. He is the CTO for Xendee, a design platform for micro grids. Specifically, welcome to the show, Michael,
• 2:22
  Michael Stadler: thanks for having me.
• 2:25
  Tim Montague: You know, it is heady days, as we were saying in the pre show. The market is very dynamic. Solar and batteries are all the rage, but now there are incentives for other resources, like natural gas and so you can't just rely on a single technology. You need to be flexible about what it is that you're planning. But tell us a little bit about your background. How did you come to the micro grid space. I know that you were at Lawrence Berkeley National Laboratory, but how did you get interested in this topic?
• 2:59
  Michael Stadler: Well, I think it was more than 20 years ago that I came actually over from Europe as a student to Berkeley, and I started working in 2002 and 2003 on the concept of micro grids at that time. I mean, not many people are really thinking about it, but we started thinking about, how can you design a microwave in an effective way, from a cost perspective, environmental perspective, right? And it can be a challenging issue, because you have a lot of technologies.
• 3:26
  Michael Stadler: I mean, you can have a microwave with a gas generator. You can have a micro with a CHP system, combined heat and power system, right? You can have renewables in the system. So all these technologies need to work together. Need to create an economic benefit, and to design such a system is really challenging. This is, this is how we started, more than 20 years ago, and we did a lot of research around this at Lawrence Berkeley National Lab. A lot of publications. We did all these projects, also with the military, and we have proven the technology, but it was always a technology which needed students actually or academics to really use it. And this is why, in 2018 we decided to actually start a company together for deep and really create a technology based on my knowledge and his knowledge, which is around powerful distribution simulation, to really come up with a product which is helping the client in an easy way to figure out how microchip is structured, what the savings are and and how we can run it right?
• 4:34
  Michael Stadler: And this is what we started in 2018 now a couple of years later, we are really driving, and the microG space is really changing. I mean, originally micro grids, it was mostly about PV systems, batteries, right.
• 4:47
  Michael Stadler: Then there was a couple of years ago with electric vehicles are coming into the big picture, but just recently now with with changes in the administration and how we see renewables now systems back to more combined heat and power systems, gas engines, and for example, in in data centers, which are Big push of the administration. Here we see now that micro, it's really playing a big role. So actually, micro, it's not going away.
• 5:16
  Michael Stadler: They're actually growing.
• 5:19
  Michael Stadler: They're just becoming a little bit more complicated and have a little bit more technologies in it than we had in the past. I think that's that's how we see this at the moment. Yeah,
• 5:28
  Tim Montague: and you mentioned a deep nosley, the co founder and CEO. I interviewed him last year. Check out episode 209 on the Clean Power Hour for an interview about designing and optimizing complex micro grids, which is the same topic that we're talking to Michael about, but Michael, let's kind of start from the ground up. When you think about designing and optimizing micro grids, how, how do you start to solve that problem? What is the foundation?
• 6:03
  Michael Stadler: I mean, of course, there are multiple steps to this, right? I mean, when you when you have an idea about a micro grid, you start screening it a side or a location, right?
• 6:12
  Michael Stadler: So you do, you have a screening process, and the data that you need for this is a little bit different than if you would decide a migrating detail. So we have here a multiple step approach. And traditionally, when you look at this, I mean, you need some information about, if it's an existing site, how the site is operated, because you want to make sure that with a micro grid, you're running the system differently and more effectively. So you need to collect some existing information about the electricity bills, the loads and all this stuff, so that you can start actually analyzing the system and figuring out what technologies are most attractive for this, right? And you have, as I said, already, a lot of different technologies that you can consider. I mean, traditionally, we had PVM batteries, but what about CHP systems? What about absorption channels, which have big play now with data centers, because they can cool with wasted all these technologies need to be analyzed, and you need some information on technologies, right? So we also provide this information if we want to make it easier for the client and for the user to actually analyze such a system.
• 7:19
  Tim Montague: And let's think about a data center, because this is in the news a lot, many gig what scale plus data centers are under development and construction in the US. Now, you know, I looked at a chart recently, and data center data centers for micro, I mean, for AI, are now as voluminous as all of the other internet cloud service data centers, right? So this explosion, the expression the explosion of AI data centers is very real. And these are very large facilities. They are measured in gigawatts, right? So a relatively modest one would be one gigawatt and they go up to like 10 gigawatts. So if you're let's just talk about a one gigawatt data center where you want some kind of resiliency, you want to ease the path to interconnection. So perhaps you're going to design the facility to be a micro grid initially, so that you don't have to wait in a queue for five years to get your interconnection to the grid operator, whether that's in miso or PJM. I mean, the queue lines are amazing, but walk us through, how does a customer leverage Xendee And what is the outcome that they're going to achieve by using your platform?
• 8:45
  Michael Stadler: Well, I mean, because we have such a large size here, I mean the technologies, the different sets of technologies that you can use a little bit limited. So it's clear that you cannot have a one gigawatt data center facility just with full tool takes. I mean, that's that's probably an illusion, and this is why we see this projects, that it's mostly combination between generators, batteries and PV systems and some market participation. So we also see, as for example, in Texas, that these data centers want to play a role in the electricity market. So the clients already have a certain idea about the technologies that are involved, right or should be involved, but at the end of the day, they need to figure out how big the generator should be, or when should they really trade electricity on the market to benefit from this, because at the end of the day, the data center itself doesn't want to be bothered by Oh, can I run this?
• 9:37
  Michael Stadler: Or no. So there needs to be a third party providing the energy, energy as a service to the data center. And that third party needs to figure out how they make money. And for this, they need to have batteries, generators, or the market participation in the mix to actually figure out how much they can actually save money or generate electricity for certain price so that they can offer this to the data center. And figuring out how big the generator should be, how big the battery should be, or what the potential for market participation is, is not an easy task, and this is what we are doing with them. And at the end of the day, when they're done, they know exactly how much they can save, or how much the electricity costs, or how much they should charge the data center with certain sets of technologies. So it's, it's basically the first step is building a concept on about the technologies that they need to use to achieve a certain goal, right? So that's the first step, and that's where we'd be helping them.
• 10:33
  Tim Montague: Yeah, and so when you think about in big blocks though, a one gigawatt data center that has to have electricity, 24/7, you could, of course, just plug it into the grid and you rely on grid power, and grid power is
• 10:49
  Michael Stadler: going to cannot, I think you. Cannot, because that's the problem, because it's not an energy problem, it's a power probably, I mean, I mean these data centers and areas where the energy is already the power is constant. I mean, as you said, it takes a long time to get utility connection. We have seen that this can take eight or 10 years. So actually, there's not the option to really go with the data set, with the utility for most of these cases. So that's actually why people start thinking about migrants, because the utility cannot handle this period.
• 11:19
  Tim Montague: I love that the harsh reality of money and what is most cost effective and time effective. And you know, we see a lot of news about nuclear power, for example, and smnrs and I've covered this topic on the show. But to build, to design, permit and construct a nuclear power plant is not a five year venture. That is at least a 10 and probably a 15 year venture if you're going to build a one gigawatt solar battery, or solar wind and battery with some natural gas.
• 11:57
  Tim Montague: How you know what is the time frame and what is the relative size of those chunks of sources for the project?
• 12:08
  Michael Stadler: I mean, every site is different, right? And it also depends on how much land you have to basically build your PV system.
• 12:15
  Tim Montague: So let's assume there's no constraints on land.
• 12:19
  Tim Montague: West Texas is a big place.
• 12:22
  Michael Stadler: Then it's actually cosplay. I mean, what we see is that combined heat and power units, CHP systems running a natural gas can be very effective for data centers, and they are big chunk in this game, because traditionally, when you have a data center, you purchase electricity to run your computers right, but then you have to cool down the environment, so you purchase electricity again. But with a CHP system, you generate electricity, and the heat from the generator, waste heat, can be used in an absorption drill, so called absorption driller, and you can cool the building.
• 13:00
  Michael Stadler: So with a CHP system, you naturally reduce the electricity demand, because you're not you don't need electricity for cooling from the utility, right for an electric chiller. So what we see is that basically every data center is playing with gas generators, big gas generators to power the facility and also provide the wasted for absorption chilling. Of course, a lot of this data centers also install batteries, because you can easily do backup for the system, and you can do market participation, and the PV systems also playing a role. But reality is, I mean, in all the projects that we have seen, there's a land constraint, and this is a little bit constraining the PV system, but from a cost perspective, it's very effective to actually power a data center with natural gas.
• 13:53
  Tim Montague: And so I have a problem, okay, I need one gigawatt hour of firm, 24/7, power and and then I feed that problem into Xendee is, can we think of Xendee as an AI that is then going to say, Well, look, based on what I see in this geography, for the cost of grid power, I recommend, you know, a 500 megawatt gas turbine combined heat and power system augmented with a 500 megawatt solar farm and a 500 megawatt hour battery.
• 14:31
  Michael Stadler: Yep, that's exactly what it tells you, what the rate of return is, what the cost of electricity are. So all this is output of the system, but on top of it, you're also getting an idea on how you should run the system, because installing the capacitance is only one part of it. But then, when you have the technologies on site. You need to run them in the in the right way to make the money, right? Because if you want to do market participation, you need to charge the battery the right time, or you need to sell to the to the market or purchase from the market so that you make money like in Texas, right? And when you run extended, you get the sizes, the capacities, as you said, but at the same time, it also tells you on how you can expect to utilize these technologies. And the nice thing about this is this is actually a control strategy. We are mimicking the real control strategy that will be deployed later on, and this actually enables us to take part of this algorithm and really deploy it as a controller. So our business game is actually helping people figuring out the system, the sizes, the savings and everything, and then when it installed, when being have installed the system, we also can control their technologies to really capitalize on the benefits that we have been calculating on. Primarily on so that's that's also an additional benefit.
• 15:53
  Tim Montague: I'm curious, because, as you mentioned, this is a very dynamic market. Things are changing. The Federal Administration is incentivizing certain technologies. They're putting tariffs and other restrictions, like foreign entities of concern on the ITC, the investment tax credit on equipment made in China, for example. Does Xendee take into account supply chain and the cost of equipment issues?
• 16:23
  Tim Montague: Because that landscape seems to be highly variable. The Clean Power Hour is brought to you by CPS America, maker of North America's number one three phase string inverter with over 10 gigawatts shipped in the US. The CPS product lineup includes string inverters ranging from 25 kW to 350 kW. Their flagship inverter, the CPS 350 KW is designed to work with solar plants ranging from two megawatts to two gigawatts. CPS is the world's most bankable inverter brand, and is America's number one choice for solar plants, now offering solutions for commercial utility. ESS and balance of system requirements go to Chintpowersystems.com or call 855-584-7168, to find out more.
• 17:15
  Michael Stadler: I mean, we are not directly modeling the supply chain, but of course, we change the costs so that they reflect those issues right in the timeline. But we are not in the business to really assess if this PV system, or that PV system will take a certain amount of time to come over or get manufactured in the United States. Yeah. So that's, that's, that's not our objective,
• 17:39
  Tim Montague: but, if, but, but is it such that I say to Xendee, okay, here's the cost of solar modules, here's the cost of batteries, here's the cost of natural gas generators and maybe a few other technologies, go ahead and optimize the design for a one gigawatt hour data or one gigawatt data center in this geography. Is it that simple?
• 18:00
  Tim Montague: It's that simple. To some extent, the front end, I know it's the rocket science is behind the screen.
• 18:10
  Michael Stadler: That's one point here, Tim, you're not designing a micro just for one point in time. And this is actually why the Migrate concept is so helpful, because you can design a microwave for now, and maybe now you have a gas generator and maybe some batteries or whatever it is, right? But you can think about the future, and that's how we approach this, because the data centers are growing, the demand is growing, so you can model actually the growth rates and all that stuff, the electricity price changes, the gas rate changes, and all that stuff, and also the advancement in technologies. And we did a couple of those studies, and we said, Well, let's start a migrate now with a CHP system, but maybe 10 years down the road, when small model reactors are more attractive, they can easily be plugged into a migrate concept, right? So we are not modeling this for one point in time. We do this for multiple years, because it's actually a timeline, or an investment timeline, on how you have to upgrade the system. And our point is it doesn't really matter how the Migrate looks like. Now, my odd point is you should start a migrate now with the technologies that you have at hand, because it's always cheaper than waiting for the utility or purchasing the electricity from utility if you can, if they can provide it, and then, over time, augment the migrate with newer technologies are just upgraded in every simulation that we did actually points to the fact it's always better to start with a microwave and think about the micro like a fluid system which gets upgraded over time, because then you can always hatch your risk against uncertainty in the market. And that's actually the real point there. Microgrid is a concept which can help you for multiple years, and not just for now.
• 19:53
  Tim Montague: Yeah, so is there more to say about finance and business models with regard to Xendee and the service that you're providing to developers?
• 20:07
  Michael Stadler: I mean, of course, multiple financial models and how you can approach this, right? You can do this as an energy as a service. You can model certain forms of investments, and we support our clients with this, right? But what I have seen is with the data centers, because we were speaking about this, is that they are mostly doing this as an energy as a service. So third part is basically installing the technologies, running the technologies the micro grid and providing the energy to the data centers and energy as a service contract, and this is what we most of the time model for our clients.
• 20:43
  Tim Montague: What are some other popular use cases for micro grids, as far as Xendee sees them, military?
• 20:50
  Michael Stadler: Right? I mean, we have done a lot of projects with the military. Of course, different reasons. It's resiliency. I mean, we had a lot of projects with the US Army overseas because they wanted to become independent from gas, for example, in Germany and Italy.
• 21:06
  Michael Stadler: So we, we did a lot of projects there. And that's that's a very good use case. And still, also in the United States, military bases really changing and how they use energy and how they generate it, because they have to be independent for at least two weeks, right? So military sites are a big player here.
• 21:26
  Michael Stadler: What we also seeing is communities university campuses.
• 21:29
  Michael Stadler: I mean, traditionally, university campuses have been using micro grids for a long time. We just didn't call them micro grids because they have gas engines, they have large generators. They need a lot of steam, so combined heat and power units were always playing a role for a large university campus, right? And actually, university campuses are the largest micro market in the United States at the moment, right? But communities are changing now. Communities want to become more independent. We see that third parties provide energy as a service to communities to have a more diverse energy portfolio, not just getting energy from the utility. So, I mean, yeah, the electric vehicle infrastructure is always an issue. It's a little bit soft at the moment, because there's not so much talking about electric vehicles at the moment, but at some point, electric vehicles will be an issue for the electric infrastructure. And now, combining this with the data center issue, the data centers need a lot of power in the gigawatt range. And then you have this hyper charges for electric vehicles, which also is constraining the infrastructure, the utility infrastructure. Then we have the perfect storm, I would say. And this is also why we were always doing projects and supporting electric vehicle infrastructure projects, because, I mean on a highway system, I mean in the middle of the desert, I mean, and you have a bad or poor utility connection, and you cannot charge a car still with two free gigawatts, right? So you need to build a micro to buffer the problem. And that's another market which is a little bit soft at the moment, but I think it will come back.
• 23:08
  Tim Montague: Yeah. You know, here we are in q3, of q3, of 2025, and the AI takeoff is happening before our eyes.
• 23:16
  Tim Montague: Right? Super intelligence is on the horizon now, and I'm just curious, how does Xendee See this landscape? What is the importance of machine learning and artificial intelligence in your platform, and in general, when it comes to not only designing optimizing but running these micro grids.
• 23:43
  Michael Stadler: I mean, of course, AI plays a role, especially in the control strategy of saying this so the planning tool is an optimization engine, which is not directly AI, but then when we dispatch technologies in real time, we need forecast and we need to forecast electricity loads. We need to forecast heating loads, renewable energy output that we use for this forecasting and AI, but I have to say that we actually try to use AI as little as we can, because it's it's computationally expensive. We see that too, right? I mean, if I run AI and I want to have the best outcome for a forecast, I mean, it costs money, right? And this is why we put a lot of time and energy into actually making these approaches simpler, so that they can become a little bit cheaper, because otherwise it can be an issue if we have to run a specific computer for every microcontroller, which uses a lot of energy. I mean, it's not a lot compared to the AI data centers, but it can add up. And this is why we try to be a little bit more efficient on this side and make it effective, because it can add up. But of course, we also use AI for forecasting engines.
• 25:49
  Tim Montague: What else? When you look into your crystal ball, do you see, whether it's on the policy front or the technology front that's going to impact batteries and micro grids?
• 26:03
  Michael Stadler: So it's interesting. I would say this way, man, for a long time the industry was thinking that we need a lot of subsidies in the micro space, and a lot of government funding went into the micro space. But I think we had the point where actually it's not so bad that actually people have to start thinking about the economic value of a microwave, right? I mean, and this is why we see more business, because people now start thinking, hey, wait a minute, I'm not getting an incentive or funding from the government. I need to do this with my own money. And this is why they're getting really serious about planning this in.
• 26:40
  Michael Stadler: In a good way, and running them in a good way. So from my perspective, I think microwaves are to some extent a mature technology. Of course, they're different at different sites, but from a policy side, I think it's a self run that I would say, and it's happening, and it will just grow. And I mean, as I said, electric vehicles and all this. So I think that the policy incentives are playing a minor role in the micro space at the moment. Of course, certain companies have a problem with this, but larger companies, or companies doing this right, we will survive and actually grow.
• 27:21
  Michael Stadler: So I'm not so sure if the policies will really impact the micro space. That's what I'm saying.
• 27:27
  Tim Montague: I often tell my listeners, if they want to see the future, go to Northern Europe or Asia, because those markets are further along in their integration of distributed energy resources. What is what is going on in other markets that you're paying particular attention to. So
• 27:46
  Michael Stadler: it's interesting that you mentioned this, because you are right that in Europe, and you can hear us on my accent, I'm originally from Europe, that we there are certain things which are ahead of the United States, but in terms of migrate space, I think it's actually really the US, which is ahead here, because in Europe, we don't see migrants like we see them in the United States. In Europe, it's really distributed energy resources, virtual power plants, large entities controlling this, and it's mostly in front of the meter. So basically you have large wind farms that have these PV farms in Germany, and they somehow get connected, which actually creates an issue for the system, because it's not behind the meter. Those systems actually burden the infrastructure a lot, and this is why a micro grid is completely different than a virtual power plant or large distributed energy resources, because a micro grid can help to reduce the burden on the public system, while a system like we have it in Germany or the northern countries in Europe is actually burdening the infrastructure, because they need to connect these large wind farms with the demand pockets in certain parts of the country.
• 28:57
  Michael Stadler: And this actually leads to the issue that in Europe, we see a big increase in cost for transmission distribution. So energy is not expensive, but the distribution transmission becomes very expensive because everyone has to upgrade to transmission distribution lines.
• 29:16
  Michael Stadler: And this is actually why now in the United States, the microbes can help, because we have two options for data centers. We can build the infrastructure which is costly, what we do it with Microids, which is actually cheaper for the society and more reliable. So in this regard, I think actually we are ahead in United States when it comes to micro it's because it's actually a different thinking which is helping the infrastructure which we don't have in Europe. Yeah.
• 29:42
  Tim Montague: I mean, as far as I can tell, transmission, it's both transmission and distribution are a bottleneck.
• 29:51
  Tim Montague: The grid is already in many places at capacity. And so if you're going to build a big asset that requires a lot of power, a micro grid is a natural go to because then you can Island and not, you know, lean on those already at capacity facilities. I don't understand, though, why that's not the case in Europe.
• 30:20
  Michael Stadler: I think it has to do with how, I mean, it's a cultural difference. And the utilities in Europe, the utilities are way strong, I would say, and control most parts of the systems in a different way. And the other thing is the utility and the utility system is, is, is newer than in the United States because it got rebuilt after World War Two. So Europe is not at the same stage like the United States when it comes to the aging infrastructure. So most utilities in Europe actually don't see that problem, because they say, well, well, our infrastructure still can handle all this right, while we know in the United States it cannot handle it anyway. It's old, it's it's failing, and I mean, it's just a complete bottleneck, as you say, but these things we're not there in Europe, and this is why people don't see the benefits of Microids like we see them in the United States.
• 31:12
  Tim Montague: Well, what else should our listeners know about Xendee and this world of micro grids, whether it's individual projects or portfolios of projects.
• 31:24
  Michael Stadler: Good question.
• 31:24
  Michael Stadler: I mean, I think at the end of the day, and I think already said this, a lot of people are interested, but they're still waiting. I think the sooner you get around the idea of a microwave, the better this for your company. I mean. You can wait a year, you can wait two years, but the problem is not going away, and I think it's time that people start realizing that they have to build some internal knowledge at the company so that they can deal with this idea of a microgrid and actually build it and maintain it, because it's a little bit more complicated than just purchasing the electricity from utility. So you need internal knowledge, and I think it's the sooner someone can start actually building that knowledge, the better it is for the business and your business case, because it's just otherwise, you're just deferring the problem. At the end of the day, I don't see how we can actually build the infrastructure in the United States at low cost that can handle all this, because California already we have $1 per kilowatt hour electricity rates. I mean, now if you talk this with building your infrastructure, we end up with$5 per kilowatt hour. I mean that that's not sustainable. So at the end of the day, I think my messages and my thinking is, the sooner we really build microbes, the sooner we out of this problem, out of this problem. Hey
• 32:46
  Tim Montague: guys, are you a residential solar installer doing light commercial but wanting to scale into large CNI solar? I'm Tim Montague. I've developed over 150 megawatts of commercial solar, and I've solved the problem that you're having. You don't know what tools and technologies you need in order to successfully close 100 KW to megawatt scale projects. I've developed a commercial solar accelerator to help installers exactly like you just go to cleanpowerhour.com click on strategy and book a call today. It's totally free with no obligation. Thanks for being a listener. I really appreciate you listening to the pod, and I'm Tim Montague. Let's grow solar and storage. Go to clean power hour and click strategy today. Thanks so much.
• 33:35
  Tim Montague: All right, well, let's wrap it up. Check out all of our content at cleanpowerhour.com Give us a rating and a review on Apple or Spotify. Follow us on YouTube and reach out to me on LinkedIn.
• 33:49
  Tim Montague: I love hearing from my listeners. You can find Xendee at x, e, n, d, e e.com, but how else can people find you?
• 33:59
  Michael Stadler: Michael, just just go to our website and Xendee.com and let's see how we can help you
• 34:08
  Tim Montague: with that. I'll say I'm Tim Montague. Let's grow solar and storage. Thank you so much. Michael Stadler, CTO of Xendee, thank you. Bye.