Hello listeners. This is the risky planner podcast. Thanks for tuning in. I know you've been playing around with AI agents and you know, other tools and stuff, and I have too, but I think we could start the conversation off with some pretty cool conversation about, a discussion about some of the tools that we're building, because it's going to lead in naturally to the topic. It is because, you know, like, spoiler alert, it takes a lot of data center horsepower to run even your basic AI. I mean, like, it's some insane factor of 10. You know how much more power and and, like, computing horsepower it takes to do an AI query versus a search. But we'll get into all that. Like, right now, I want to talk about how my new favorite thing is pitting AI against each other to do various tasks. Oh, wow. So, yeah, so you have thing one makes a code, and then you have thing to check the code, yes, and it's you'll occasionally get, you know, snark and well, I'm not sure why this was done, or this seems to be missing, or it's not particularly well organized. It's like, wow. You know, chat, GPT has some opinions about copilots coding capabilities. Yes, yes, it does. I find Claude is really good too for challenging that. And that goes great. We talked about security and Claude with the professional, I actually, I don't think it uses any inputs, even the free model. I know the professional doesn't use inputs to train its LLM, but the free model might actually not use I don't think it does. I'm not sure, though. I mean, there's, there's definitely claude's whole thing is that their conversation will be ethical, yeah, also that, but they're trying to be ethical about how they use your data and like they want to surface everything to which is a good thing. I mean, because the open, AI, chat, CPT, little bit different. Models are very different, and they are, you know, mercilessly scraping everything you do with the with their agents to to get new and I don't blame them, frankly. I mean, this is the massive free training data set at their fingertips. So why wouldn't they use it? What have you been using it for? Well, so I have gone down the AI agent rabbit hole, so to speak. So I am building a content marketing AI agency, and it's been really interesting. So I've been watching load of YouTube videos on on, just coding this, I started looking at like automation. So there's people call them AI agents, but I'm looking at like, make.com like Zapier has come a long way. Like, these are automation the no code agent builders, yeah, and they're, they're, you know, they're kind of like the AI agents, but they're just automations. They've been doing them forever, right? So it's just now they're linking, you know, different AI tools together. So there's a bit of the agent work. What I have been doing is, in the last couple of weeks I've been I downloaded Claude, sorry, cursor AIS, text to code terminal, and I started watching videos on it, and I've started building the AI agent and my agency in cursor AI and I have never coded a line of Python in my life, but now I have, like, folders upon folders for different tools and the different capabilities, and they're all starting to produce like, they're, they're, I'm using it for different purposes. So there's one is like competitive analysis, there's like content monitoring, there's sentiment, social media monitoring, and then there's content ideation, and then then there's sort of cutting first drafts based on that content ideation. And then there's human interventions at different points. But it's pretty, pretty exciting, like it basically set out to automate certain parts of the marketing agency scope. And so far it's, it's going pretty well. So that's cool. It's not done, but by any means, but it's it's been quite a journey. Well, isn't it amazing? I mean, the fact that you can even start down this, this path and and do so for free is a real sign of the times. By the way, I get real Uber vibes from where we're at with AI right now, like it's going to get more expensive and worse, I think, in the next three years. So we should all squeeze as much out of it as we can right now. And you were talking about writing Python code, like I have written about 1300 lines of Python code, which is nothing. And actually the truth is that I've written zero lines of Python code because. Wasn't me doing it. It was my, you know, combination of adversarial chat bots that were doing it at each other. But you know, so, yeah, this is, again, a sign of the times. Like, would I ever market myself as a Python coder? No, but if someone throws a whole bunch of Python code at me, I know exactly where to go to, like, parse it, learn what it does, and, you know, make some adjustments or improvements or critiques or whatever. It's kind of amazing that, you know, this is back to AI as force multiplier, like it's, it's not going to replace, let me replace some jobs. And I know a lot of CEOs in tech and in like, the coding world are already moving down that road. Personally, I feel like it's premature. It's a topic for another episode, probably. But you know, even so that the fact that you can even think about it is pretty amazing. This kind of segues nicely into our topic. Yeah, that's a good I mean, we're gonna talk about data centers. There's a lot of construction happening with that. Construction comes planning, comes scheduling, comes cost, estimation, it comes, you know, balancing stakeholder expectations. There's health and safety. I mean, these are all the things that we are involved with. Well, this with Nate totally, right, yes, including and especially environmental and sustainability concerns with, like, a big heavy underscore under water usage, okay? I mean, obviously data centers are extremely power hungry for a wide variety of reasons. So there's just the power generation element of it, too. And we talked on a prior episode about how some of the the tech giants in AI space are currently looking to either build new or reactivate old nuclear power stations. They are to power their future AI. So that gives you an idea of how much power we're talking about here, and the the cost of building that power is one thing, but even the data centers themselves are massive. What kind of cost are we talking about for a data center? Albert, so I've got, I've got some research here that tells me that we're talking about something in the neighborhood of five to $6 per square foot, which, which equates to about $244,000 per acre. That's US dollars, okay, so we're talking about kind of a lot of money, kind of one, one, yeah, kind of a lot. And that's just the land cost. Yeah, it's just a facility. It's not even, we're not even talking about the servers, none of the cool stuff. Yeah, exactly. You don't get any of the cool stuff for that. You just get, you just get land and maybe a building. The the the cost of a data center interior can be as high as $1,000 a square foot, okay? And in fact, I'd be surprised if it were any lower than that, because that was the last time I did any benchmarking on this, and that was probably five years ago. So, you know, a lot of things happened in the last five years. You may have heard indeed so, but yeah, that's a that's an old number, but it's a high number. Well, how would you like if you're managing that amount of money, like in your your your your project manager, how are you even like beginning to think about like managing the costs, you know, the controls for just that, that that build, is it different than what we do with others, with other clients, not, not necessarily, with the exception that you know, there's, there's this. In project management, generally, there's, there's this. And in engineering and architectural design too, there's this pervasive thought that you can design one and build many, right? You've even seen it summed up as d1 BM, you know, you've probably seen or heard that phrase before. Yes, you Google that first time I saw Did you well? Because I'm a marketing guy, right? I'm a Marketing Yeah, very good. So design one build many, is the idea that, well, it's just what it says on the tin, right? Like you design one of the thing, and then you go build a bunch of them, and that is a great idea that is almost never practicable. There's a few areas of like industrial construction and commercial construction where you can really do that? Well, data centers happens to be one of them. Okay? So the the first thing that you do is figure out how you can drive economies of scale is because you're not going to be building one of these things. You're gonna be building many of these things. Okay? Data Center growth is projected to grow somewhere between 20 and 30% annually for the next five years, okay? And my guess is that it's probably gonna take off from there. So considering the sheer amount of land given over to Data Centers right now, that's gonna grow, and it's not gonna stop growing for decades, probably. So talking about cost control, i. Doing d1 BM, well, is is going to be job one, right? Yeah. So build one, it costs, what it costs? Figure out what your lessons learned are, how you can economize on future builds, and then just keep doing you almost wind up treating it less as like a traditional critical path method, you know, commercial construction project, and more like an agile project, where you're iterating on a concept over and over and over again, which, by the way, those of you thinking, Oh, well, that's not a great description of Agile. I know that still, it has more in common with with with Agile when viewed globally than it does with like, you know, if you zoom in on the micro of any one build, I think the risks involved, you know, stakeholder management, it, with regard to building a data center, is is massive, because it's not like a mine, where you're going to go do it out in the middle of nowhere, where, You know, where the where the natural resources are data centers. Are they need to be close to, like people. They need to be close to power sources. And typically, that's cities. So like, for example, right? Atlanta, I think, sort of put a freeze on data center. I'd have to look up that article that I was reading, but yeah, a bunch of municipalities have sort of put a pause on data center construction because of the demands around energy use. Well, you may have heard about the Wonder Valley, kind of it's a conceptual development here in Alberta, where I am, Kevin O'Leary and his investment group want to, kind of, I will say, exploit. I don't mean it in the negative sense, but exploit the permissive regulatory environment and cheap energy costs in Alberta to stand up a huge hyper scaling facility, you know, that will contain an enormous amount of processing power to to basically lease out to AI. And by the way, I could be getting details wrong. Details don't really exist for this project yet. It's just an interesting idea, and like, an AI generated video at the moment, but even so, like it points to kind of what you're talking about, like Wonder Valley. Should it exist? Will be in the middle of nowhere. Okay, so the development is exactly that it'll be a huge development right in the middle of nowhere to avoid being in the middle of a city or having all of these municipal zoning and permitting issues to deal with. But putting these things in the middle of nowhere does, in fact, bring its own challenges, so you have to build those at scale. But I think we're going to see more of those, like data parks, where you have huge, huge, huge developments that get service with water and power and of course, fiber optic running. New fiber is going to be we're going to have to, as a species, run an enormous amount of new fiber optic cable in the next 10 years to keep up with all this demand for bandwidth. So that's going to be interesting to watch, too. To watch, too. Yeah, the what do you think? So if this is, yeah, I've always wondered why the data centers have to be I mean, once I read about the challenges and access to power, it makes sense that they're going to be close to a an urban center where there's but the the idea of putting them out in the middle of nowhere makes a lot of sense to me. I think the like, what so dive a little bit deeper and describe, like, how would we solve the challenges around, you know, access to energy, like, you know, water, I understand. We're going to build it next to a lake or a river or whatever, but how, what are the challenges in accessing the power if you're going to build it out in the middle of nowhere, and how would you solve we're talking a few episodes about the concept of CO generation, which is like putting generation facilities right next to whatever thing you're trying to power. And the CO comes in because it's usually done in concert with grid power, right? So you have a handful of different ways that you can power your asset if it's out in the middle of nowhere. But bear in mind that, like every step of this process, the longer you make the runs like the longer the fiber is. I mean, the fiber, the length, doesn't make a huge difference. It's more like the units that are at each end and how many fiber strands you've got, but certainly for power and for traditional copper data and that sort of thing, the length matters a lot to how much loss you're getting and how high quality it is, and like the risks associated with running it. So if you can shorten those runs, you reduce your risk, and you reduce your cost, and there's all kinds of other benefits to just having a short path between your support services, like your water and your power to the thing that they are supporting. So as short as it can be is as short as you can make it basically, and that's that's part of the reason why putting them near cities has been. Popular. The other thing is, like, you can build a data center into anything. So if you've got a disused and, well, anything Nate, I mean, if you've got a tiny enough computer and you've got a spare, I don't know cigar box, well, Nvidia did that. They, they have that handheld, right? Yeah, about the size of a cigar. I know. Yeah, I know. And like, we could, we could at location. Company is not known for its incredible server capacity, okay, like, we have a box in a closet, basically, that's our server. We could put that NVIDIA box that you're talking about, I don't remember the name of, in in the closet, with our computer, and basically give ourselves the ability to run a large language model that's approximately equivalent to GPT four. Yeah, chat, GPT four, yeah, or maybe it was the 40 mini model. But anyway, it's, it's a pretty capable piece of super computing equipment that's literally the size of a cigar box. So I was being flippant, but also kind of not, yeah, no, it's, it's definitely Small. Small enough it is. You put enough of those things right next to each other, they're going to generate an awful generate an awful lot of heat. And you know, heat needs to get cleared away. So that's the real challenge, right when you're when you're building data centers at scale, is you need to get rid of the heat. So it's the computers themselves use a lot of power. But then there's also the requirement for the data center cooling, which uses a lot of power as well. If you if your data center continues running, you know, if you have 99 plus percent uptime, you're probably putting on a backup power. If it continues running, but your HVAC doesn't, then every single thing in there melts and dies and you no longer have a data center. So like backup power and redundancy requirements and robustness of your grid and all that stuff. All very good reasons to have things on premises designed to spec and yours, right? Because if, if grid power dies, because, you know, something got struck by lightning, or there's a terrorist attack, or whatever horrible things should happen, then, like, you know, you're you're screwed. And like, maybe, if there's a terrorist attack, you don't care if your data center is running, you know, there, it feels to me like i and i use Claude, I use Gemini daily. It I almost, I can't really think of an instance where they were down, but it sounds like there is so much that can go wrong to stand one of these centers up. Oh, absolutely. And, and, like, as as risky as the prospect of building one is, and as as much risk as there is just in the basic infrastructure. Like, I like to think about projects at time zero, and then think about them, you know, before and after. So time zero is like, when the thing comes online, like, that's the finished state of the project. You know, the product that you're building with your project is in the best physical condition it will ever be okay. Theoretically, it's matching all of the specifications of the design it's done right. From then on, it starts to degrade as it is used and as entropy takes over. You know, nature tries to, you know, get its insidious fingers into in between all the cracks, and things start to turn off, and political situations change and demand shifts, and you're adding stuff and subtracting stuff. So like at time zero, it is at its ideal state. So every point before that is your project, and every point after that is after that is your operation. And the operating these things, in my opinion, like the amount of time it takes to build a data center is somewhere between a few weeks and a few years, depending on the size. And that's like, even if it's a few years, there's still a pretty small amount of time, considering that most of these buildings are built for a 3040, 50 year life. Yeah. Okay. So even if you take five years to build your data center, you have to operate it for 10 times that amount of time. Okay, so whatever risks you had while you were building your power infrastructure, your your water infrastructure to support your cooling, even things like the building envelope itself, which will degrade over time and allow moisture and the elements in all those things, have to be maintained for 10 times as long as it took to build them. Okay? So that's where the real risks are, is in operating and maintaining your fancy new asset. So if a data center is delayed if construction takes long, you know, we've done, you know, research on capital projects, and a majority of them go over budget due to delays and other kind of risks, sort of coming to the surface and not being managed, like what kind of contingency levels? You know, just based on our experience with capital projects across various industries. And we've done infrastructure, we've done construction, we've done energy, you know, we've managed 50 billion in assets. What kind of contingency levels are we building into the budgets for, you know, these data centers that cost, you know, around my. Research is saying 15 million a megawatt, yeah? And that's, that's a lot, yep, you know. So you don't want to screw that one up. So you want healthy contingency, but not so much that you cripple your project's financial picture, right? And the sweet spot for, well, okay, first off, this is, this is what I like to call a How long is a piece of string. Question, how long is a piece of string? How long the answer is, as long as it needs to be, right, right? So this is one of the reasons why the design, one build many approach doesn't work super well for capital projects, because if you're building an identical data center on top of a skyscraper in Shanghai versus building it in the middle of Wonder Valley in Alberta, those are two very different projects, right? The risk profile is different. The availability of local infrastructure is different. But even if you try to control for all of the external factors and just focus on the individual project, you're still going to wind up with natural variations in price based on the supply chain in the area where you're building and that sort of thing. So it's how long is a piece of string? It's really hard to answer. But the kind of sweet spot is somewhere between 10 and 20% for a capital project that's going for its like final construction phase. And depending on the size of the project, 20% is a lot of money, okay, hundreds of millions of dollars you might be spending on any one like a larger data center. So 20% of that is a whole project in and of itself, right? So you really want to keep that contingency number down without pushing it so far down that you've actually not given yourself enough headroom to control for risk and to manage risks when they arise. So this is, you know, bringing me back to that design, one build many conversation we were having earlier, the more of these things you build, and the more successful you get. As long as you're iterating and making improvements every time you do one, you can bring that contingency number down to account for the fact that you've mitigated, Incorporated and accounted for a lot of risks in project. You know, Version Three versus version two, which itself mitigated more than version one. You know what I mean? So, and data centers are a perfect example of when and how you can do that, right? You know, I said earlier, you can build a data center into anything like, if everybody's got, like, the the corner store or grocery store, or whatever, like the building on the corner, that just whatever business goes in there just seems to fail. Doesn't matter what it is, you know, it's just like, it's just gonna die, like, data center, data center, just jam one into there, you know, you know, every corner store that has, like, the Bitcoin, you know, ATM sign on it, you know, has the past both of them, has the passport photo on it. And then also is nowadays center, with the help of Nvidia, exactly what is the like. So let's talk about Kevin O'Leary. And so he's an investor. He's, you know, I think this is a smart decision, the Wonder Valley development. And by the way, if you're, if you're sitting there wondering, are we talking about the Shark Tank guy? Yes, yeah. 100% Yeah. So I think it's a smart move, right? Let's build something. Data centers. Don't need a ton of people for an operation to operate it. They need a team. But what is, how do you think the you know, the project team is presenting risk to somebody like an investor like Kevin, that's thinking like, I'm just looking for a return on my investment, but he's probably looking, you know, to build something sustainable and long term. Yeah. So my first question is, how do you think they're kind of proposing to, like, you know, the source of capital like this is the risk that we might encounter, and here's how we're going to get around it. That's part of the reason why they're looking at Alberta, okay, because the the the top line costs will be lower here for permitting reasons, and because of the like I said, the cost of energy is lower here. And that kind of thing matters a lot, because if you're doing return on investment calculations, ROI, okay, like, for example, if you're if you're running an NPV cost model and that present value model, okay, it doesn't matter. Actually, what kind of return on investment modeling you're doing, the single largest factor in your payback period is your initial capital outlay, okay, so to put that in less wonky terms, the less you spend up front, the faster you get payback. What are they? What are they spending up front? Either? Again. How long is a piece of string? Yeah. But you asked earlier, like, what is what risks does the project team represent? And the answer is, cost, right? Yeah, your your project team represents a cost that is capitalized against the initial value of the asset, depreciated over the life cycle of the asset, and becomes part of the you know, it's part of your tax burden as an organization. That tax burden and the cost of operating the asset all figure into the ROI but again, the single greatest. Uh, you know, determiner of when determiner is that word probably is, anyway, of when you can make up words. It's, it's your podcast. The single greatest determinator of your payback period is your initial at capital cost and the length of the string, like 100% then look right exactly you got, how long the piece of string actually is, reaches all the way to the moon and back. Yeah, there's a bunch of companies that are standing up data centers. Demand for data centers is going to be is increasing, right? And so the, oh, it's huge already. Yeah, it's huge already. It's it's going to be constantly increasing. The demand for energy will be constantly increasing. So we're looking at the CO, as you said, the CO generation. You know, there's lots of solutions. How does, how does, I mean, I think about like in LA right now there's wildfires, like, just like natural disasters can get in the way, right, if you're building something out in Alberta, like, you know, cold inclement weather, you know, flooding, like are like the cold is kind of a gift, honestly, yeah, that's because for half the year you can basically do passive cooling. And, I mean, I say basically, because you still have to exhaust, just open up the barn doors. Let the, yeah, exactly. Just roll, have Farmer John get out there. Roll the barber John and his giant fan. Farmer John's Nvidia AI stack, yeah, will suddenly be cooled by nature. You know, that's not, I think that's not too far fetched, right? Like they're, they're, you know, like growing corn. You know, one decade they might be, you know, growing corn next to a giant data center. The next, who knows? Yeah, I want to see the the artfully framed photograph of that. You know, in 10 years time, we're gonna, we're gonna get that for sure. But I've now lost track of what your initial question, yeah, I don't remember about Farmer John. Can we? Can we play just a thought experiment? Because I'm really this build one, or design one, build many. How, like, let's just pretend, or a thought experiment. Well, what? What kind of frameworks or what are we? What do we have to standardize, to do and to build? You know many duplicative is that right? Did any you know many data centers? We're all making up words, yeah, exactly. Let's just have a thought experiment like play like, just, let's, let's, let's spitball. What are some of the frameworks that we need to have in order to realize a design one, build many model for data centers. Well, okay, so imagine that instead of building data centers, instead we're trying to we're trying to find someone's keys, okay? And they're lost in the grass, okay, which is an experience we've all had at two in the morning after an interesting evening. So you and your five friends are all looking for, as long as they're not the car keys. Yes, you're looking for your door keys. Oh, they're definitely the car keys. Nate, in my case, they would definitely have been the car keys. But anyway, why am I bringing this up? It seems insane and totally unrelated. Let me tell you why. Because if you are treating each of those person search efforts as an individual project, where the starting point is, I have lost keys, I'm looking for them, and the end point is I have found keys and I am using them. Okay, every single person involved in this search effort, running their own project, is going to think about, okay, well, what is optimal for me to search this entire area, to find the keys and then put them to use as quickly as possible. Okay? And if everybody does that independently, they're all going to come up with, you get five friends with five different solutions, and they're not necessarily going to play nice together, like you're going to run into your friends, you know. But if one person doesn't participate and instead controls the efforts of the others and says, like, Okay, we're going to break it up into a grid, like, this is a square lawn. So you take quadrant one, you take quadrant two, or, you know, we'll do it vertically, like, you know what I mean? Like, break it up, yep, organize your efforts. Plan them. You're gonna find the keys four times faster, probably. Okay, now, but what if you get lucky? Who cares? That's not a case that you plan for, okay? So what you plan for is the worst case, right? The worst case is the last person finds it in the last possible spot. So you make the number of spots they have to search as small as they possibly can be, and you have just increased your worst case scenarios like upside by four by 4x right? So how does this pertain to building data centers? Yep, because if you think about each one data center as its own project, and you're building 1000 of them, okay, which is awfully big number, but like, if you're building a lot close, we're getting there, yeah? Well, yeah, yeah, we're getting there, yeah. I mean, you know, Kevin O'Leary's Wonder Valley situation isn't going to be one giant building. Oh, right. It's going to be a complex of many buildings. So. And thinking about them as individual projects that all have their own objectives and are in competition with each other for capital funds and project resources and things like that is wasteful and unnecessary in the same way that, you know, having everybody just look for the keys at random, you know, would be okay. You're gonna wind up with unnecessary, avoidable conflicts, duplication of effort, and you know, it's just, it's just not a good way to manage things. So, so the the right answer is, you, you treat it as a program instead of as a project. Okay, so there is a distinction to be made between program and project management here. And you know, a lot of you probably know what that distinction is already, but yeah, give it to me because I'm not sure. Yeah, of course, you were Nate, but a program is just a group of related projects that are managed together. Okay, that's all it is. So, you know, a good example is that, is that data center complex, you know, hypothetical one that I was talking about. There's probably an administrative building there. There's probably all that generation stuff. There's, there's like, district power distribution, there's, you know, the the fiber optic infrastructure that needs to be there. Mention all the water stuff, all that stuff's still there, okay? And all, not to mention, like, there's a site plan, okay, there's roads connecting all these buildings to each other. There may even be some kind of, like, you know, miniature mass transit system, like, you know, little tram pads, or whatever. All that stuff has to be thought about in a holistic way, or else you wind up with you ever been to an old west town? Oh, yeah. The the layout of the streets can be best described. How would you say? I mean, it's just a grid, right? Well, I mean, it's a T. Well, you went to a very nice, well planned old west town then, because, like, traditionally and typically, they weren't planned at all. I was a white horse. I have you ever been to wall drug? And was that North Dakota? No, no. I'm just thinking about, yeah. Okay, so I'm just thinking about where they where they have their shootouts. Just one west town is one road. What kind of west town are we talking about? I well, okay, so think about that one road, all right. And you've got, you know, Clint Eastwood, yeah, taking off his poncho and getting ready for business. So, like, that one road is probably the only one that has any regularity to it at all. Like, you get off that one road and it's just been no roads, yeah, yeah, no, exactly. And it's like houses are just wherever they fit, you know, that kind of thing. Maybe a better example would have been like a an old city in Europe, okay, okay, where, you know, all the roads are just where the cart paths used to be, and they follow, like the contours of the natural environment and that sort of thing. So, you know, right now you have $100,000 sports cars that were designed last year driving on the same roads that you know the Romans were using to move dressed stone around 2000 years ago. So again, why am I saying all this? Because your average medieval city center is a good example of a city that didn't get planned. Okay? So imposing at least some regularity on to how that stuff is is laid out, or data centers you referring to, like, with the program, yeah, yeah, exactly. So, like, you've got to think about all that stuff in a uniform sort of way to avoid the medieval city center, you know, off Main off Main Street in the old west, kind of thing, you've got to really be cognizant of where you're putting stuff else do, do program teams kind of think different than than project teams. Well, for one thing, you can centralize a lot of the the functions of a project into a single unit that provides those almost like as services to the individual, like sub projects that are being run. So for example, if you're running a major construction project, you've probably got a project controls team right that has like a scheduler and a cost controller. Maybe there's a person who's specifically doing change management or document control is probably a risk manager. So you got, like, a team of four or five plus a project controls managers, maybe like five people or so, and that would be a pretty bare bones project controls team for like, a large project, got it four or five people, so that four or five person team would be duplicated or replicated across all of the various sub projects if they were run independently of each other, because these are all going to be like, 100 million dollar projects, you know, potentially something pretty, pretty big money going on there. So you want to have good control over it. If you run it as a program, then you figure out what requirements all of those projects have in common with regards to their project controls, and then you centralize the fulfillment of those requirements into a PMO, okay, where, in this case, the PE doesn't stand for project anymore. It stands for program, right? Okay, so your program management office then supplies those as services out to the various projects. And you probably still need one or two project controls, people per team. But if you think about like, if. You're running five related projects, and each one would have had five person team on it. That's 25 people. If you take that five person team and centralize it and then put one person on each project, now you're down to 30. Sorry, no, you're not 10. We're making up math too. You've increased it by five. We're making up math. No, that's okay. So there's a bit of centralization and standardization. That we're doing with, with the it's not just people, yeah, it's not just people. It's also about the processes that the processes, the systems that they're using, systems, yeah, you know which I'm now in extremely well trodden territory, talking about people process and systems, right? That's right. But those, those are the answers to your question, like those data, using the data and standardized getting value from that 100% Yeah, well, so interesting discussion. I love the thought experiment, and I hope our listeners got value from that. What's let's, well, let's wrap up. So what's your one piece of advice for for stakeholders who are pushing for faster delivery of these data centers in a high demand market. Well, if you're you know anything like Kevin O'Leary, and you've got millions of dollars or billions of dollars to throw around and make new investments in data centers, there's a pretty good chance you should just do what he did and go find a place where the land won't be expensive. There's a permissive regulatory environment, low energy costs, and the opportunity to draw on an extremely skilled and available labor force. Okay, so you know, Alberta and Texas are frequently compared to each other. Texas is one of the best places in terms of market value to go to business with a data center right now. Okay, so just picking where it's going to go and using that to drive your requirements is going to be a huge, huge win for anybody who's looking to stand up new computing capacity. And frankly, because the world of international business is looking a little bit more volatile than it did a few months ago, thanks to, you know, one, Mr. Donald Trump and his continual threats of tariffs and things, you know, the the creation of bilateral trade disagreements means that whatever country you're trying to service probably you should build it there. Yeah, okay, yep. So there's, there's a pretty good that that's a good isolation of like, you know, a lot of cables have been cut as part of the Ukraine war, right? So, like, Finland lost internet for a half a day or so. By the way, don't fact check me on that. That's probably wrong, but, like, the kind of things, too, people we're making, yeah, we've got Alternate Facts, but no. The The point is that whenever you're doing things at a great remove, like at a great distance, then you're introducing risk. So just don't do that, right? You're trying to serve the US market. Build it in the US. What's the best place to do that? Central Texas. Probably Central Texas, Alberta's farmer. John, so, yeah, exactly. Find, find, find a farmer. John, yeah, and yeah, do it? Do it where it's super cold half the year. Thanks, Albert, that was great. Yeah, thanks. Now, it was a fun one. All right, thanks all for listening. We'll talk to you next time. Hey everybody, it's Albert here. Thanks for tuning in to the risky planner podcast. We hope today's conversation was informative and, above all else, inspires you to excellence in what you do. If you liked today's episode, don't forget to rate, subscribe and leave a review. It helps us reach more listeners just like you. I'd also like to thank Thompson Igbo egbo for letting us use his excellent music on our show. If you like what you hear, check him out@igbomusic.com that's E, G, B, O music.com talk with you later. Foreign.