Joining me. I'm John Thornton, Citi Industrial Specialist, and I get the joy of leading an interesting power panel to talk about what's going on around the globe in data center and power generation. I'm going to be joined with Rob Yeager from Emerson. He is president of Emerson's Power and Water Solutions business. Next to him will be Jeff Wiltrout from Cummins, vice president of corporate strategy. And lastly, from Vestas, Henrik Andersen, president and CEO. I got a series of questions to go, some targeted at each one, and we'll make our way through here. So let's start off with Rob and Jeff. This will be kind of directed towards you. U.S. power demand growth has largely been relatively stagnant for an extended period, but now seems to be increasingly meaningfully driven by significant data center-related power demand, but also increasing adoption of electrification more broadly. So can you comment on how the shift to more significant power and growth demand impacts and how you think about potential opportunities and challenges for your businesses? Yeah, I got to tell you, John, if I was sitting on this stage 24 months ago, I would have said the global power industry is a headwind. But now it's a strong tailwind for us. And I think it's a combination of, you know, certainly the data center growth, renewable addition around the world. But it's been a really pleasant surprise for us. A pleasant surprise has been the life extension of fossil units. Life extension of fossil units. And the last, it's hard to believe, in the last 12 months, we've received orders for 100,000 megawatts of plant upgrades in the United States. To put that in perspective, it's like two whole Southern Company of generation or two whole NextEras, whole generation upgrades in the United States. So why is that? Why? Because it's a combination of not being able to get the megawatts on the ground fast enough. The utilities just can't build fast enough. A combined cycle plant takes two to three years. A nuclear plant, twice that much. PV solar, a lot faster. And the second thing is Inertia. As renewables come onto the grid in the U.S., it has kind of a destabilizing effect on the grid because you need these large spinning pieces of metal, gas and Steam Turbines that instantaneously convert mechanical energy to electrical energy to provide this dampening effect on the grid. Otherwise, you get frequency variations. And the grid has to stay at 60 hertz. If it doesn't, the lights will start dimming. Major equipment will fail. So I think that's what's really driving it. The challenge for us is in the next outage season, fall, spring, fall, we'll be managing 500 outages around the world. 500 outages. That's a lot of megawatts of work. You know, so it's challenges for the supply chain, for our labor force, and these outages. You know, I know a lot of plant managers around the world, and most of them don't have a sense of humor to begin with. And if you tell a plant manager you're going to take their 1,000 megawatt unit offline for 30 days and return it to service, you better make it, so good problem to have. Record backlog that we're seeing just in retrofits and a really aggressive outage season coming up around the world. Yeah, I'll just expand briefly on a couple of different points. You alluded to the outages, Rob, and of course, for Cummins, our primary role in the power space is as a provider of backup power generation, largely through diesel gensets, and so one, the combination of the new demand growth, data center-driven, as well as other dynamics paired with the increasing likelihood and frequency of outages, weather events, and other dynamics means that's been a good recipe for our business to provide stationary backup power generation into those markets and trends. Most notably, our data center customers have a huge requirement, more so than even your plant managers you refer to, to ensure appropriate reliability and redundancy of those power requirements, and so they drive a higher portion and relationship of backup power to the prime power requirements they're expecting. And that's, of course, been quite good for our business in the U.S. and globally, which I'm sure we'll talk a little bit more about. And then we ask ourselves pretty consistently, how do we participate in a more decentralizing power grid infrastructure, right? We are not the grid-level players, but we do have some offerings and opportunities to participate as folks continue to not only find backup solutions, but to then do demand response and other alternatives for how they solve this inherent structural shortage we have in the power landscape. So given the growth in power demand expected from Data Centers and AI and the need for the power to be available 24/7, what's the role in the different forms of powers, whether it's renewables or gas or nuclear, in powering the growth in energy needs and AI? And how are you positioning your investments to best take advantage of the growth? I'll open it to the floor. I can try to start off here. I think one of the things in any energy transition, it's one of the biggest transitions the world has ever embarked in. The energy transition here, we need more of all of the energy sources we can get to. It's also interesting that when you look at the energy consumption, around 20% of it now stems from electricity. The rest, of course, comes in various parts. If you look at that, somewhere around 1.6% comes from the renewable part. We sort of say we just started that journey, and we need to be able and be better in combining the buildout. Because I think one of the points you picked up on is time to energy is critically important for the world. Because if we get that wrong, we get the demand and the supply wrong, and light goes out in parts of the world. And you're rightly saying, I think we have been blessed in many parts of the developed world where we haven't seen outages, but where a lot of the not-so-controlled demand and supply, we see outages, which, of course, is part of it. And it's interesting, then we all look around when there is an outage, let's find somebody to blame. Because if we can just find somebody to blame, then at least we've fixed that point. And then we can talk about that energy source or that company doesn't work. And that's not the right way of thinking about it. Because actually, if there is an outage, in many ways, it's because we didn't have the right access to energy generation in the first instance. So we need to build much more capacity. And of course, I'm part of this. I know I've been called a destabilizing factor. But renewables is part of it, but they won't be the 24/7 100% solution. It is not going to be, but it's going to be part of it. Because if it's not part of it, guys, it will be extremely hot, and the CO2 will go soaringly as well. So there is both a cost element, there's time to energy, and there is, by the way, also a cost of energy, which at least the renewables and the Wind Industry has now solved over the last 15 years. And we can go anywhere in the U.S., and I can show you that. And we can go anywhere in the world, and we can show you that. So we need to find that. The good thing, do you know what that is? We are every day as a consumer, we are taking more and more decisions that create the backup. We either have the car, we have the transport, we have the buses, we have the electric heaters. We have all of that that are now in our homes. And at some point in time, they will be the ones that are also storage it. So I always go, and when the wind doesn't blow, it doesn't matter because your battery in the car, unless you drive the full battery length every day, which, by the way, you probably shouldn't, you find a new job then. But then what does that battery give us as a storage right in our drive? So. Yeah. I think you're right on the money, Henrik. I think the key, John, what you said is 24/7 and people to blame. But as new renewable generation comes in, what the utilities are facing is they now all of a sudden have this portfolio of fossil, traditional fossil fuels and renewable energy that they have to manage. So our challenge is we've invested a lot in software, a lot in software. So the first thing you have to tackle is how do you coordinate and manage thousands of megawatts of wind, solar, battery, coal, gas, nuclear, and have a common look and feel, very reliable, very available, right? And that rolls up to the remote operation centers. We're seeing a lot of investment now in remote operation centers, especially in the United States, because a lot of the solar sites are unmanned or nearly unmanned. So if you don't have staffing, you have to have strong software, or you'd have to have reliable software, and it has to. You want to see the same thing that you see in the field in the remote operation center. If you go around most utilities in the United States, now they're going to have a ROC, we call it, that they manage all their assets. Second thing is software you need. Anytime you have this mix of generation, right, you still have to bid into the real-time markets. I mentioned the grid reliability and Inertia in the system. Henrik touched on it. You have to. Most have to. Most of our plants in the United States are now doing primary sustained Frequency Response, right? You have to be able to look at the frequency of the grid, and you have to move the units around to maintain frequency. We work very closely with the balancing authorities in the United States. Our software, you know, they work with our software to create this frequency. The second thing is you have to run, if you want to bid into real-time markets, you have to do what's called automatic generation control. When the RTOs and the ISOs send a signal to the plant, you're bid into the real-time markets, you better be available, and you better be able to move that unit to respond to the ISO. Our software does that. A lot of the investment that we've made to tackle this problem is in software and in those two very critical areas to keep the grid stable and to keep the markets robust. Yeah. What is the biggest impediment to power related to the growth at the moment? Policy uncertainty, costs, grid connections, finance, other factors? And then maybe how much of the power demand might just end up being met by replacement upgrades versus new construction? Yes, probably a better position than I am, Alex. Well, I'll just start. We see the backup power generation side. So we aren't as deep into the regulatory dynamics, the permitting approvals and things like that. I'm sure these folks can speak much more eloquently than I can on that dynamic. There is inherently, we're going from a world where we just haven't really built structural capacity for a very long time, and that's changed quite quickly. And so there is not only a regulatory permitting dynamic, there is literally in our market a supply chain capacity buildout dynamic to be able to keep up with the pace of some of the demand. And that's a real and very practical constraint to being able to satisfy the demand we're seeing, especially for backup power generation in the data center space. I can comment again from a software perspective. You know, when you have this combination of renewable and fossil generation, and you're bidding into the markets and you have peak loads and you have to respond to the market, you have generations that are spinning reserve, right? And they have to be ready to go all the time, right? Because if you're bid in and if you have a PPA, power purchase agreement, or you're bid into the market, you better be ready to perform. So what we're seeing a lot is backup power is a lot of the old gas and Steam Turbines. We've developed software solutions for those. We're the leading supplier of software solutions for turbines. Hard to believe that. We've retrofitted turbines. We've probably close to 200 gas, steam, Wind Turbines that we can go and replace all of the old antiquated Control Systems and put in the state-of-the-art controls. Why you need to do that is because you have to be able to coordinate those backup units with the rest of the generating capacity that you have. If you have an old control system that can't respond or it's buggy and it's not reliable, there's no way you can do it. We run something called full auto. When you walk into a power plant in the U.S., you go into the control room, most of the operators, they're busy, but they're not managing the control. You don't want them to touch the control. If we lose a pump or a fan or a motor, if it's an 800-megawatt combined cycle, our software gracefully does a run back to 500 megawatts and we save the unit. So it's a combination of reliability, the software, the backup power, bidding into markets and to create this stability of generation for the utilities. I think here also, I think still we have to accept that whatever part of the world we are in, unfortunately around the permitting and the acceptance into grid and the modeling with operators, I think it's not only getting more complicated, it unfortunately also gets with a bigger and bigger red tape. The permitting backlog in some sense still creates challenges. It also creates some concerns because even it prevents a number of areas to actually apply latest technology because the permitting process runs so long that you permitted with something that was, I call it the iPhone 5 instead of the iPhone 16. It's interesting with that, we all know what we have. We don't go in and ask for an iPhone 8 if we can now, because there's part of it. How do we also get to that? Because there is quite a lot of the technology. For us, we have 90,000 turbines across 80 countries, and to return here also to say that sits with an enormous load of data on when does it work beyond capacity, and therefore, instead of curtailing it, how do we store it? How do we sort of take all the energy out and we can then have it for when we need it, so there's a lot of that which AI will help with, but also the storage nature of how far we have come. Take the price of storage cost compared to where it was 10 years ago; that will give an easier life at a point in time, but it doesn't solve the underlying if the energy generation is not there on time. There isn't any magic wand, so that's the investment. That unfortunately comes back to the permitting, which I think everyone looks for somebody to blame again for not. But we got to have bold voices in getting faster to the permitting process. Maybe sticking with the joyous regulatory topics. Given the incentives for domestic content with the Inflation Reduction Act and now the new potential for increased tariffs with importing into the U.S., how is the production footprint and supply chain for onshore wind in the U.S. evolving? And then backing up, and I'll ask Rob, does the U.S. administration change how Emerson approaches energy transition type projects versus investing in more traditional energy versus power projects? Yeah, we are fortunate here to keep our localization in the U.S. We expanded in it, and we have had that for more than two decades, so for us, it's a natural. I'll also say it's Monday. I was in Windsor and Brighton, Colorado, and standing there to see that we have now more than quadrupled the number of employees there compared to 14 months ago. It is a lot more fun factory to visit in that sense. So the localization has happened, but let's not forget that localization happens and capacity gets utilized when there is something to produce, and I think that's back to also the point is the legislation we have had in the U.S. and also other places. Let's avoid that it becomes a 24-month stop and go and rather a continuous planning because the assets that are getting produced are intended to have a life cycle of 30 years and should not be subject to 12 or 18 or 24 months cliff or changes all the time because that actually prevents some of it. But, for us, clearly a strength that we have localized and we will continue to localize as long as it creates also the clear, transparent way of what legislation you're dealing with. Yeah, I don't think it's any surprise to anyone in this room that the U.S. energy policy is dynamic. It's not like China where you have, I think they're on their 14th five-year energy policy where they know exactly how much coal they're going to build, how much renewables, and that's okay. That's okay because what we've done over the years is we've invested, we know what's going to change. So the things I've talked about in just the renewable energy and the software for that, we know the reason we do that is because our customers are faced with the same issue with the U.S. energy policy. They don't know what to build, right? So what we need to do is make sure that our software, that we stay relevant, that we stay relevant as air fuel mix changes, we can change. We have software to do all kinds of fuel types. We automate half the generating capacity in the U.S. Coal, gas, we have 3,500 hydro sites, 65,000 Wind Turbines around the world, 100 operating nuclear plants. So all fuel types with this common look and feel, this easy to use, debug to troubleshoot no matter what their strategy is. Second thing is as an investment in software together with Aspen Tech, we have a great opportunity. If you think about where renewable generation is being added, it's not traditionally on, you know, traditionally electrons flowed one way in the United States. They're in the transmission and they flowed into the distribution side. A lot of the renewable source is being added on the distribution side. So it was traditionally a load, now can change to a generation source in hours or even minutes. So this bidirectional flow of electricity from distribution to transmission is challenging. And the way it works now is there's SCADAs. They've been around for a long time. The cap banks, the tap changers, the reclosers, there's SCADAs software that does all of that, but not necessarily in real time. They can monitor it, they can change it. Together with Aspen, we're creating, we have 40 million lines of code on the generation side. When you walk into the control room and you, I explain running in full auto, that's a lot of code. It's very mature code. It's been around a long time. Very visible. You can watch the logic execute to the millisecond. You can see in a generation plant a 1,000-megawatt supercritical unit. You know it's going to be reliable. We're taking that 40 million lines of code and we're moving it to the grid edge. Think about that for a minute. Now, for the first time ever, you'll be able to see the distribution side of the network with the voltages, all of the variability that you have in that in real time, like you see in a generation plant. The major breakthrough, moving that to the grid and being able to generate, you know, to automate it and view it in real time is very significant. So that's one of the biggest things that we're doing from that perspective in the U.S. market and investing software there. You all have significant global footprint here in the U.S., and I think there's significant excitement regarding the expected uptick in data center-led power demand. But how is that demand translating globally? And are there any drivers, either positive or negative, that you see globally? And then back to North America, as the U.S. moves towards more decentralized energy grid, how can power generation companies ensure the integration of distributed energy resources while maintaining system stability and security? I can start to speak a little bit to the first one. You know, again, from a diesel genset man, you're right, that is very much a global business and has been for some time. You've had the more mature markets where the grid has been more stable for a period of time where we have sold true standby gensets. There have been regions we participated for a long time, you know, India, Africa, and things like that, where we play the role to certainly supplement what are much more limited grid infrastructures that are now starting to catch up. You specifically alluded to this data center dynamic, which is of course a global phenomenon. We've seen the most growth very actively here in the United States and participated in that. But we have a large power generation business in China and are selling the same type of equipment, large engines and large gensets to support the data center build-out domestically in China. The same story is true in India, in Europe. The underlying structural trends are driving that element of power generation demand and the need for redundancy and reliability is certainly very consistent with some different regional dynamics, but very consistent as you go to the different regions around the world. Exacerbated as well by the domestic cybersecurity concerns you see. Every country wants to ensure that that data that is being generated, stored, held is being done so within their own borders, right? So that drives an inherent need for the data center build-out, the associated power demands, and for us, like I said, our ability to drive that power capacity in regions around the world. So very similar structural dynamics playing out slightly differently as you go in different places around the world, but one we're excited to be able to participate in everywhere. There are other things besides Data Centers generating a strong power market that I haven't seen in a long, long time, if ever. It's just the economics as well. If you think of megawatts per thousand people, megawatts per thousand people around the world, we're spoiled. In the U.S., it's three and a half or four. China, one to one and a half. India, a half. So we're seeing a lot of generation being added just because they want to reach our standards, which is a nice thing to have. Jeff, you mentioned China. We have over a thousand systems in China, probably closer to 1,500, mostly on the 600-megawatt and 1,000-megawatt ultra-supercritical boilers. So these are massive fossil units that generate a lot of the China-generated capacity. In the last 18 months, to put it in perspective, we've got orders for 22 new coal units in China, about 20,000 MW. That's an entire U.S. utility just in the last 18 months that we're working on in China. 70,000 MW are either planned or in construction right now, China coal. So you can see in the Chinese EPC contractors, right? We're getting a lot of work from them because they're exporting into Southeast Asia, into Vietnam, into Indonesia, into Thailand. You know, they are experts in building. They can build a 1,000-MW supercritical unit in 18-24 months. Who does that, right? So they're taking that technology and exporting it around the world. And that's another growth opportunity for us for automation. Maybe just here, and this is the wonderful thing with ACE is that you return to these conferences every year and you see sort of the topics change from time to time. I think it's interesting to sit here and elaborate a bit around when we were here three to five years ago. It was all about how we get the hydrogen and how we get all of that to sort of play and other stuff. I remember one of my famous sayings is, "Hydrogen is not a natural source. It's actually a produced something from electricity and other parts." Wherever you go today, that sort of disappeared because we all right now know it's the criticality of having the base access to energy whether in a format, it's in, it's beyond. And when we then hear that we're now reopening old coal or coal mines or whatever, the world never spent more coal than it did last year, period. That also means that whenever we are going to have a conversation around the CO2 and temperature, and for somebody that sat here five years ago, we sort of all did this theoretical calculations of we could still manage 1.6 degree and it will be way beyond that point. We thought we could control it to 2050, but we cannot forget it because, guys, if we sit here in five and 10 years, next generation will never forgive us. So we also have to say grids, the baseload, it's 24/7 play. We got to find ways of storage. We got to find ways of doing it because just adding things that we know is just adding to the CO2 or adding to the temperature increase is going to give us a bigger thing. I can see here and I'm not upsetting, maybe I'm setting you a little bit, but I'm not upsetting the two old people up on the channel here that they won't forgive us. I don't know what they will do when they stand in our graveyards, but they will say they promised us something better and they didn't keep that promise, so this energy transition is the long run, but it is interesting when we sit here. The topic of emergency access to energy is just so much bigger because most countries have gotten a surprise of the demand cycle, and that's not nice. And unfortunately, that is because they didn't get the permitting, they didn't get the supply organized correctly, and now we are there. So let's not forget the big picture. And then we need to build time to energy now. And that for me, then we got all, come on, I got sensors and all, I can tell you the weather, not only hours from now, but 24 hours from now, anywhere in the world you want. So let's go there. And I know when it's working and not working. So the AI giving us all these amazing opportunities, whether we have it in real life or we have sensors, but it doesn't solve, we need more to run those models. How has the competitive landscape evolved in power? What new players or emerging technologies do you see in the power space that are catching your attention? And are you concerned at all with the new players taking share and affecting your ability to price? And then separately for Henrik specifically, given the apparent return of profitability for onshore wind, is the industry now structurally better placed to deal with inflation spikes or volatility in raw materials, wages, and freight? Every time I get the question, it almost sounds like it's the onshore wind that's creating inflation and all the structural challenges. It's not. But I will just say here from a levelized cost of energy and comparison over a lifetime, happy to do the competition every day of the week. It's 15 years ago. Anything that has technology life of more than 15 years can be repowered today with a huge uptake in the efficiency and also in actually benefiting the consumer of electricity. So the technology has delivered. The onshore industry is also ready to do that part. And instead of blaming onshore for inflation, it's actually taking and contributing to the opposite. So therefore, getting some of that on is critical. I cannot avoid thinking five years back. Part of a panel five years ago was part of that. People have drawn these extensions of lines and levelized cost of energy. People were actually saying there's a risk for that energy will become free. They said, "Really? Energy can never and must never become free because then we use it unlimited." And today we know it will never happen. And the graphs and the lines indicated that you could build renewable and other stuff almost at a levelized cost of energy that will trend towards zero. It's not going to happen. But unfortunately, we took decisions at a point in time five years ago, which had that there were endless opportunities of capacity and it wouldn't cost us anything. That has changed. But we are still not there where you go into an election and promise people more capacity and not cheaper pricing. And that's maybe the mistake we are making. Yeah, good point, Henrik. I think when you look at the levelized costs of energy, clearly renewables are at the low end, you know, $20-$30 per megawatt hour. You get into combined cycles and nuclear units, they can be well, you know, a new nuclear unit could be $180-$200 per megawatt hour. So the challenge is, you know, it's affordable. The LCOE for renewables are very affordable, but Henrik alludes to, and I started this whole talk about is you create a weakening of the grid with respect to stability and reliability. So you got to have these prime movers that you add. They always have to be a part of the equation. So that, you know, that's why we're seeing such a strength, you know, in the modernization of automation around the world and new build, right? We're seeing a lot of new construction for simple cycle gas turbines, combined cycle gas turbines, two to three years. With respect to, John, with respect to competition, one thing is I fear and respect our competition, but we don't have a lot of competition. We have, you know, the traditional ABB, Siemens, GE. Scott was here this morning from GE Vernova. But if you have a thousand, you know, a thousand megawatt ultra supercritical boiler online, I mean, it's 15 stories high and the thermal stresses of that, if it trips at full load and the safety's lift, you can hear it from miles away. So it is very difficult for a startup company, you know, to say, "Hey, I'm here. I just wrote this code and I'm here to put it on your thousand megawatt supercritical unit." I'll go back to the plant manager. They'll say, "Well, who is this person and why are they talking to me?" You know, you got to have, you know, our reference is a book, you know, that thick, not only a thousand megawatt, but what boiler type, what country, what do you want to see? So from that perspective, the barrier to entry, the barrier to entry to this business is very high. Yeah, and I would just say from our perspective, and we have the internal combustion engine version of a very similar story, which is there's only so many people that make big engines that can provide, you know, two, three, four megawatts of power at a time. And that's not an easy game to play. And it's an expensive one to enter. And the engine will remain a pretty important and elegant solution to provide reliability, flexibility, and cost efficiency for how you just supplement all the elements that Henrik and Rob are referring to as it relates to building at the baseload capacity. So we same thing, certainly watch and expect people to continue to work to compete effectively. We pay close attention to that. At some level, you will see storage as a requirement play a more impactful role in how the whole thing plays itself out, both in terms of the prime load and then supplementing and providing redundancy and reliability. It absolutely is going to be a big piece of the economic and operational equation for how this is going to happen. No, and I think, come on, I mean, in many industries, it's always when you really need it, then you think there are too few suppliers of it. And when you don't need it, there are way too many, and then they are financially in each other's way. The world will equalize that over time. I think here, when you talk about energy, energy has now come to everyone's minds as high as almost the defense industry has been for the last couple of decades. And if you're in doubt, you just need to go around in Europe these days and see how the priority has switched from a lot of other things to prime defense and prime energy investments because you know that is the one you want to sit and have a conversation around. This is actually what gives us independence. So I think priorities have changed over the last five years quite dramatically. And I think industries and barriers to entry, yeah, that's right. But on the other hand, some of our things here, you don't expect more to come in. You expect some of the ones adding continuous technology investment in here. And I can just see how we have expanded on the supply chain in terms of medium, smaller size companies also in the components. That is the way I think we should think about it instead. I don't think necessarily out of my supply chain that will come and new investors suddenly doing turbines, but there is definitely here somebody that is on a global roadmap so they can travel, they can work, they can supply us across the globe. And therefore, that expansion is a very welcoming thing. So, I think those sort of growth and you can almost call them growth ventures under the larger companies is natural. And I think we all either have the ventures part of our businesses where we allow that or in the supply chain. And then can we talk a little bit about, sorry, can we talk about a little more about capacity constraints and lead in power? What do power focus backlogs and maybe the Ford pipelines look like today? And how should we be looking at them both in the near term and the long term? How visible is the growth runway? We assume your conversations did not change after the DeepSeek news, but maybe you can comment on that as well. Yeah, I can talk about visibility of the power market. We love market studies, don't get me wrong, but we don't really need market studies because we automate about 1.8 million megawatts around the world. It's about 20% US generating capacity, global generating capacity. So we have this, we're like on, we're finger on the pulse and we're like in the trenches. So we see what's happening around the world from a lot of utilities. So we know what they're planning for outages, for what their schedules are. So we can see what new build plans that they have. You know, I mentioned the 500 outages that we're going to have in the next fall, spring, fall cycle. So we have this tremendous visibility into the market. I can tell you, I can tell you that I've been doing this a long time and I have never seen, I have never seen the market this strong for what we do since I've been doing this job for sure. With respect to, I know there was a lot of concern about DeepSeek. I think it was clever computer science, no disrespect. I think they did some really nice things on optimizing the prompt technology. I think they did reinforcement learning. They did on the inference engine, they did some nice things. They used an eight-bit computing architecture instead of 16 or 32 bits. So they optimized the microchip sets and the GPU. But I believe that this is normal. I mean, this is going to happen naturally. I think when you see the generation that's being planned because of Data Centers, I think that's baked in. There'll be a natural progression of algorithm development, but I don't see it changing the forecast for generation expectations around the world with respect to Data Centers. We're working on AI itself. We have our, in May, we have our first ever, maybe the first ever in the power industry, GenAI-enabled software release where we're creating the smartest person ever to walk into a power plant, the smartest person ever to walk into a power plant. We're running the large language model locally. What I mean by that, we're ingesting all of the manuals, thousands of pages of logic that operates these plants, all the historical data, not only the historical data from that plant, but from the historical data from other plants. The real-time data 24/7 is also being adjusted, so it just gets smarter and smarter and smarter. And so you walk into the control room, right, and you'll see that nothing's happening. You'll see no alarms on the screen, nothing. But the agent will be able to detect. It'll be able to see that in a large plant, you're about to experience loss of lube oil on the steam turbine, for example, which you wouldn't normally see because there's usually no alarms for that. Anyone knows about anything about a steam turbine, you'll know loss of lube oil. You blow up the bearings for sure, and probably the rotor and the plant will be down for six months or a year. That's what we call a $10 million save. Control systems don't do that now, right? They don't. They don't have the ability to do that because the AI engine is able to interpret. There's been an AI software around for years, but it could find things. No, something's different, but it doesn't say what is different and you can't, it interprets what it is. And not only that, but you could say, "Look, what do you mean I'm about to lose lube oil on this steam turbine? Let me interact with you." Either voice recognition or I'll type it in and say, "Okay, I checked this, I checked that, now what?" Right? So it will be able to not only diagnose the problem, but allow you to fix the problem, right? Right in the control room before you have a $10 million issue. But maybe coming back to the DeepSeek and other things, I think here, I think it probably illustrates a bit how we all feel, and I spent a very early part of my career in the traders' environment. If there is a trader that has a wrong position, they will start looking for hopes and probably wait until Sunday to go to church to have the ultimate hope of that that position is changing, and I'm just sort of saying here, it doesn't change with that when you sit at it right now, that the trend of demand is going away from you and we are sitting with a position that is underlying wrong. We need more capacity, and that's why, so when somebody says, "Oh, that was DeepSeek, that could maybe take a bit of the pressure on the demand side." No, it's not. It's not because it's in there anyway for it's not only AI generated. Look, and if somebody had kids back home, look at their behavior and how they spent the energy today. So there is just that little hope every time we hear, "Oh, that took a little bit of," it maybe only take a degree of something of a sense of urgency. Go to Germany. They have gone from something where they took and relied on something that is no longer there. And that country has to make hard priorities, as had done high priorities in 24 months, which I would have said three years ago, impossible to do. But they have done it of need, not because they wanted to. Okay, I think that's where we're going to end it. We're out of time. Thank you, everybody, for joining us. Rob, Jeff, Henrik, thanks for joining us. Thank you.