ARK Invest: Why SpaceX Wants To Put AI Data Centers In Orbit

Hinzugefügt: 2026-05-11

[00:00:00]through uh Starship launching thousands of times a year, they could get 100 gawatts per year. So that is the equivalent of launching roughly all of the nuclear facilities in the United States today annually at Starship.

[00:00:13]>> What's up everybody? It's LG Ducet here and welcome to Milk Road AI, the twice weekly AI show that is totally down to book a one-way ticket to Mars, but only if they promise that I'll still be able to watch the latest season of Love Island when we get there. Today is May 11th, 2026. Recording on May 7th, 2026.

[00:00:29]Listen guys, jokes aside, the trade this year and for the last few years has been AI stuff, right? Semiconductors, chips, memories, companies that might have contracts with other companies. Those have been the winners. But it's always important to look a bit further ahead and even beyond our atmosphere. My guest today tells me that the next big bottleneck, alite, a bit further into the future, is actually rockets and specifically who has rockets and access to rockets to complete the long- aaited space buildout. We'll dig into that and more with our guest Daniel Maguire from Arc Invest. Today's episode is brought to you by Ferris, the layer 1 built for Real Fi and Nexo. Earn interest, borrow and trade crypto. Daniel, welcome to the show, man.

[00:01:07]>> Great. Yeah, thanks for having me, LG.

[00:01:09]>> You don't watch Love Island, do you?

[00:01:11]>> I do not know.

[00:01:14]>> Maybe I'm not talking about rockets.

[00:01:16]Yeah, >> it's just an LG joke. It's just for me.

[00:01:18]It's just for me, man. Uh, but Daniel, um, you're a huge part of the research team at at Arc. uh and you were a big part of writing the report that we've covered a few times on the show. We've covered some of the other parts of it which was the big ideas 2026 which is a huge collaboration um from you and your team and you specialize largely like in the space part and in the energy part and I really wanted to have you on to cover both of those. Um, so kind of like the leader question for you, Daniel, what I want to know is just like with space, what exactly do the next like five years look like from like an industry standpoint? Because we all, you know, when the shuttles launch and come back and bring astronauts up and stuff like that, we all watch, but then it feels like space then just it just disappears from the from the from the front page from the timeline once there isn't a rocket actually flying up and down. And once that's gone, it's like we don't nobody even talks about it, but I feel like there's a there's a huge thing going on there.

[00:02:09]>> Sure. Uh maybe it's helpful if we take a step back and I just explain like the overall scene of the space economy and then happy to dive in in any areas. Um so really uh what's opening up the space economy over the last couple of years is the concept of reusable rockets. Uh years ago, people laughed at SpaceX when they came up with the concept of reusing part of a rocket. Uh historically, for the most part, you would launch a rocket and lose most of it. So from an economic standpoint, it wasn't very feasible. Um, however, when you look at last year, SpaceX is really leading the way. They now have over 10,000 active Starlink satellites, which is its internet connectivity service. Uh, they have over twothirds of all of the active satellites in space. Um, and really this dominance is thanks to what we believe at ARC is a 10-year lead on the industry. Um, and that's really a statement based in fact. So when you look at uh rocket reusability, uh SpaceX through its Falcon 9 rocket landed an orbital class booster. So the bottom part of a rocket back in 2015, uh they've executed almost perfectly on reusability since. The closest competitor is Blue Origin. It's uh Jeff Bezos's rocket toy and uh they landed a booster uh at the end of last year. They just recently reused in April. So that's 10 years later after SpaceX. And notably, while other companies are really just starting with this partial reusability, the bottom part of the rocket, SpaceX is really leaning heavily into full reusability. So, the bottom part and the top part uh which really h is with its Starship program and then that has a lot of knock-on effects for the space economy. So, that's kind of where we are today. Um, and it's why it's really starting to heat up. Now, >> if you're not seeing this, you're trading blind right now. There are real moves happening behind the scenes that most people don't see until it's way too late. Inside Milkroad Pro, you can track exactly what our analysts are buying, selling, and what's on their watch lists before it moves. Martin recently unloaded 40% of his cash off of one signal. And Melvin is up multiple calls by like 50% and we're already up 30% this month on some alts. This is all happening every week. You can try it out for just a dollar for 7 days. The link is in the description.

[00:04:13]>> Is Was it Jeff Bezos flight? Was that the one with Katy Perry on it?

[00:04:17]>> Is that what I'm thinking about?

[00:04:18]>> Yeah.

[00:04:19]It's the same company that was a suborbital flight. So they actually stopped they stopped doing that and they're they're focusing now on the core rocket business of pulling satellites to space. Yeah. With with a rocket that's called New Glenn.

[00:04:31]>> What like So here's a question for you is why why is everybody so far behind?

[00:04:35]Is it just that that Elon just went on this like started this company 20 years ago, 15 years ago when nobody else it was like an uninvestable area that it was just like NASA does that nobody else touches it and nobody really thought about it or was it just that I don't know I don't know the reason why everybody else was lagging?

[00:04:50]>> Yeah, I think a couple of things. First and foremost, I'm at risk of staying in the office like building a rocket is very difficult. Um so that that that's one. Uh secondly, there have been a lot of companies who've tried reusability and have failed partly because it's very difficult. H and really thirdly is that you know really before the commercial space opened up the only government contracts were really governmental. Uh they're like fixed price cost plus there's not a lot of margin in it. Uh through the reusability Elon and his company have made uh satellite connectivity commercially viable. That's we know is a positive business in terms of cash flow generation for SpaceX. So it's really demonstrating to others, okay, there's actually a return on investment here outside of building a rocket for the government. So I think it's like a number of factors. Um, and then most recently is the concept of orbital data centers, which is really the buzzword around AI right now. And that that's kind of adding fuel to the fire.

[00:05:46]>> And that's something we've covered that a few times in our in our newsletter.

[00:05:49]Um, and as well I think a few times on the show, but maybe Daniel just remind us like what exactly what exactly is the orbital data center and why why build that instead of just uh picking a field in Texas somewhere and and plopping a big data center down there.

[00:06:03]>> Yeah. Yeah. Why a develop a very difficult thing to do which is a rocket and then b send a load of mass to orbit instead of building it on the ground. So I want to know maybe if we take a step back a key bottleneck for AI hyperscalers today is access to power. I mean we just saw yesterday SpaceX is giving some power to Anthropic and through GPUs. Um and when you look at developing power on Earth uh you have nuclear power where ARC is a big fan of that's a multi-year construction project. It's really a 20 30 plus solution. Uh wind and solar are are good but they're really uh intermittent meaning it's not always. It's not always sunny. Uh you can pair it with batteries although there are some supply chain constraints there. uh gas turbines a lot of AI hyperscalers are turning to they're sold out till the end of the decade. Um and let's say you do have a power source and you want to connect to the grid in the United States that's a 5year weight. So even uh let's forget about the grid. Let's say okay we have our power source let's go behind the grid. Uh and then there's a lot of nimism or not in my backyard h where people are just don't want data centers built beside their house. So the the concept with orbital data centers is that you uh launch a satellite into space. It's in an orbit known as suns synchronous orbit where it's in near constant sunlight. Uh that means that there uh is no real need for kind of heavy batteries on the satellite. I mentioned that because mass is very important to launching stuff to space.

[00:07:23]The lighter it is, the cheaper it is to launch. Um so when you're operating in in the sun 24/7, you're getting uh free uh energy from the sun. Everyone talks about nuclear fusion on Earth, but really the best fusion to use is the sun. It's a big ball in the air. H has all of the heat we we need. And really uh that is the concept of orbital data centers and and why the likes of Elon and uh Google have a project suncatcher.

[00:07:48]There's other private companies like StarCloud. They're really leaning into this uh because of the enormous opportunity it represents.

[00:07:54]>> Okay. So, so I guess what's the what's kind of the picture you're painting for the next 5 or 10 years? cuz it does I mean I those are staggering numbers what you said back at the start that Starlink has 10,000 satellites that have been launched that that SpaceX owns 2/3 of the satellites that uh exist in orbit today. Um, but it still feels like that alone is far away is still there still a pretty big gap to actually having the type of data center that you're describing. And so I guess maybe paint that picture the next 5 10 years despite it there being a huge energy bottleneck, right? A huge power bottleneck like you're describing, you know, everything sold out for 5 years, but >> also is I want to know like is that the top priority for what these rockets are being used used for as well?

[00:08:36]>> Sure. Um so there's a lot kind of packed into that but maybe to take a step back um the partial reusability uh what that has enabled is satellite connectivity.

[00:08:45]So that's Starlink that's an economically uh feasible business. Um and this all really comes down to launch cost. So how expensive is it to launch mass to Earth. Uh at ARC a lot of our research revolves around the concept of rights law. uh for your viewers or listeners that aren't familiar rights laws states that for every cumulative doubling of a metric h cost decline by a constant percentage. So in uh the case of rockets that cumulative doubling is kilograms launched to orbit. So every cumulative doubling of kilograms launched to orbit launch cost decline it's around roughly 17%. Uh what that has enabled today is Falcon 9 to be at around $1,000 per kilogram to launch to space. As I said that gets something like a satellite connectivity uh economically possible. The critical unlock to this whole orbital data center uh concept is Starship. Starship is a fully reusable rocket as I mentioned earlier bottom part and top part. We estimate that could cut launch cost by another order of magnitude to sub $100 per kilogram. And that makes orbital data centers economically possible. Um really for those of you who aren't as familiar with the the space uh SpaceX has been testing and developing Starship. They have flown it 11 times and the test flight uh the next one flight test 12 is coming up probably in the next couple of weeks. Uh and that is really the next generation of the rocket where they anticipate will commercialize. H and really to be frank without full reusability orbital data centers aren't really going to be economically possible. So, a lot of it really hinges on on the concept of full reusability and and SpaceX is really the only company who's going after it. Uh, yeah.

[00:10:21]>> Wow. So, so what so and maybe you could you can also explain to me like you mentioned the SpaceX and anthropic deal.

[00:10:28]Maybe you could explain that to me and how how it relates to this. I think I just having trouble.

[00:10:32]>> I think I think I just try to like wrap my head around I think a lot of people are just like what you know because we look at this potential SpaceX IPO which might come this year. the the and this is on the the the crypto DGEN world, but it's like the the preIPO price trades at like 1.7 trillion, right? Which would be the most insane IPO ever known to man.

[00:10:52]Um compete competing with the anthropics and the open AIS as well. Um but I think I think what I want to understand is just like what what exactly is SpaceX as a company in five or 10 years? You know what I because there's already such a diver diversified product suite and so many different ways that they make money but so many different ways that they spend and and I'll start with that and then I can ask you my next question.

[00:11:12]>> Okay, sure. If you ask Elon and he's come out publicly and said this, he'll say that he thinks orbital datas will be the cheapest way to get compute in two to three years. The question is uh from your perspective, what do you do with that compute and where is the business model? Really there's kind of two things you could do. You could have it like as an infrastructure as a service which is what we actually just saw XAI and Entropic do where they uh lease the GPUs to uh a player, they get to run their large language model and monetize in that way. Um terrestrially, I think Brett Winton, our chief futurist, has put estimates out there. I think it's around $10 billion per gigawatt of power. Um alternatively, and really which is much more fruitful opportunity is actually using those GPUs yourself and monetizing your own large language model. Uh and that uh we estimate again Brett Winton, I think it's a tweet out there, it's around uh two to three times more. So 20 to 30 billion per gigawatt you can get for monetizing your own large language model. And that's really why uh explains the rationale for SpaceX acquiring XAI and really having the vertical integration all the way from launch to the infrastructure building the satellites and really to monetizing that with its own large language model uh through Grock. Um, so the the news uh yesterday that came out is that uh XAI's first data center, Colossus One, they're leasing it to Anthropic to give them all of the GPUs to help increase the the token limit for Anthropic because I know me personally and a lot of people at ARC, we're running into that token limit. We we use it quite a lot. So, >> how do you how do you guys decide who gets more tokens?

[00:12:40]>> Um, it's h who can convince the head of IT to prioritize the tokens. Yeah, depends on value add and and productivity uplift.

[00:12:49]>> A little pitch competition for sure. A little efficiency competition. Yeah, I think I think everybody's going through that right now.

[00:12:54]>> Um, okay, great. So, and then and then my last question or my second last question on the space side is, you know, we we discussed a little bit about Jeff Bezos and and and I know that I think that Virgin also had a rocket at one point or had like years ago was testing some kind of suborbital flights or something like that. Um, is there like will there are there any real competitors? Like who is SpaceX now competing with uh if anybody with kind of everything you've described and sure like Jeff Bezos will event I'm sure he'll eventually have rockets that could people can use. Um but is there is there is there anything outside of just being rocket companies that would be competing with SpaceX?

[00:13:31]>> Yeah. Well, I think specifically on launch uh to maybe frame it um last year out of all the kilograms sent to orbit I think SpaceX was around 85% of that. uh the remaining 15% is actually dominated by China. Um so if you remove China from the equation in terms of like the public markets, Rocket Lab is a a great company. They have a small uh rocket albeit it's not yet reusable. They're working towards a larger reusable rocket but they are the most active launch provider in the west outside of SpaceX.

[00:13:58]Uh you have Blue Origin who you mentioned. Um there are there is also other compelling companies in the private market. One I would shout out is a company called Stoke Space. uh they are currently developing a rocket which is intending to be fully reusable. So as I mentioned SpaceX with Starship is going to be a fully reusable rocket.

[00:14:15]Everyone else is only working on using the bottom part. Stoke space is intending to leaprog that and really go straight into full reusability. Um and why that's important is as I mentioned launch costs, lower launch costs, it opens up uh much more opportunities uh as it relates to what's economically possible h in space. So that's the launch and then really you could speak all day about what goes into the to the rockets. There's companies like Intuitive Machines uh who landed on the moon a couple of years ago and and uh have had a couple of moon and lunar missions. Uh there's um plenty others like uh maybe a kind of a technology convergence of energy and space. There's companies like Campco or BWXT H which people wouldn't necessarily associate with space but uh nuclear propulsion is something that's of important interest to the United States government and you need uranium to fuel a lot of the nuclear initiatives both in space if it's building a reactor on the moon or or uh propulsion for deep space. So there's like so many ways you can play play the space bet. Um and they're kind of just some off the top of my head but as I said I could I could speak all day on this so I'll pause there.

[00:15:21]>> Don't don't tempt me. Are these all private companies? Like is Stoke Stoke I'm assuming is a private company?

[00:15:25]>> Yes, Stoke's a private company. Rocket Lab is publicly listed. Campco is publicly listed. They're Uranium Miner.

[00:15:31]BWX Technologies. Intuitive Machines is publicly listed too. Um so it's kind of a mix of public and private. There are a lot of private companies that are coming to the public market uh too as well. But that's why I mean at ARC we have our venture fund. Uh, it's available to to most retail investors. Minimum $500 and you get access to some of the private companies I in uh our fund before they hit the public markets.

[00:15:53]>> Oh wow. I I didn't even know we were doing a plug for ARC by me asking the question. There was a secret plug for ARC. I was like how do I how do I know Stoke and you're like here's why Stoke actually you actually use AR.

[00:16:03]>> Well I should say yeah Stoke isn't isn't in our our venture fun we're monitoring closely. There are other space companies that are there uh too. So yeah, Portal Space Systems is another one. They do uh satellite maneuverability. Um and I'm probably Oh, Radiant Nuclear is another one. They're a micro reactor 1 megawatt electric. They're involved uh terrestrially and some uh Department of Defense initiatives, but potentially could have applications on the moon as well. Um yeah. Yeah. But tried to slide in an arc plug there. All right.

[00:16:34]>> What's what's Daniel? What's like a flaw in this plan? You know what I Like what is the what is something that is like no listen like what what are the what are the people what are the bears for space what do they say what are the strong arguments for like no like the you know the data center is just going to crash into the ocean one day and like or destroy a city when it when it burns through the atmosphere or something you know like what that's a doomsday scenario but like what's what is the the contrarian take here for for this this is a wonderful thing man I love space I've always loved space I think a lot of people are excited about this but like there's got to be a lot of good reasons why this won't happen >> yeah well I mean that's what what makes a market you've two two sides of the argument. Um course I would say in the other side uh some people argue that uh having a data center in space the radiation can flip bits in the GPUs and can throw off your LLM. Um you can rad hard a lot of the hardware to to get past that. Another argument I often hear is that uh it's very difficult to maintain and operate a satellite. If a GPU goes down in the data center, somebody can fix it. The counterargument I put to that is that it's only a single source of failure. If you have millions of satellites as SpaceX is intending to launch, you could de-orbit it and send up like 60 more the next day. Um, and then lastly, a lot of people speak about radiation because space is a vacuum. You can't cool the GPUs through convection or conduction and uh you need to use radiation. H the problem with radiators is that they add a lot of mass. Um, but I suspect someone like SpaceX who has uh 10,000 satellites in space uh and are working on on a fully reusable rocket can definitely work and fix and address the quote unquote radiation problem. So, I don't suspect that'll be that'll be >> okay. Hold on. This is like a science question. Why can't why why couldn't why isn't NA in space a natural coolant?

[00:18:11]It's a vacuum, so it's like the coldest place. But this is I feel like I'm in third grade now asking dumbest.

[00:18:16]>> So, yeah. No, I I believe it's because the the lack of atmosphere. So the radiator is uh essentially is the best way to do it. Um so it takes the heat and just radiates it out into space. Uh >> okay. Okay. Okay. So it has to Yeah.

[00:18:29]Right. Okay. So it needs somewhere to go and you you need something to push it out.

[00:18:33]>> Push it out. Yes. Exactly. Okay. It's not that space isn't cold enough to to the heat. Okay. That's why I was like listen man. Space is like the coldest place in heaven. I was like have your water flowing around. benefit of doing this is that it'll the heat can just go out into space and just go away, you know. Yeah.

[00:18:51]>> Yeah. And I should say, yeah, the means of doing that is using a radiator.

[00:18:54]>> A radiator. Okay. That's that's a connecting dot. Okay. Glad I'm smarter than I thought. That's that's better.

[00:19:01]>> Give yourself more credit, Algae.

[00:19:03]>> Thank you. Thank you, Daniel. Is is so what in the future? Let's And again, I'm trying to think further ahead because again, this is, you know, in the in the intro, I kind of positioned this as like listen, this this will be a trade that everyone's going to be piling into at some point. We don't know when but at some point this will become you know once we kind of pass this current phase what you know you told me that 85% of space launches right now of of kilograms or whatever brought to space or spaceex >> in the future what percent of compute is being generated from space like is there an estimation for something like that like what percentage of data centers or of power for data centers is being generated by this like solar orbit that you were describing or by data centers doing that. Yeah, that's a great question. So may maybe to to set the context for for people h like if you think of a one large nuclear reactor like this like you see in the Simpsons and Homer Simpson that's around 1 gawatts of power. H Elon has stated he believes through Starship launching thousands of times a year they could get 100 gawatt per year. So that is the equivalent of launching roughly all of the nuclear facilities in the United States today annually through Starship.

[00:20:10]He believes h then once you go to space h and you're on the moon which is the lunar economy the gravity isn't as intense you can use what's called a mass driver so you don't need a large rocket and with a mass driver he thinks it could be 100 terowatts a year which is a thousand 100,000 gigawatts a year so like 100,000 nuclear reactors a year albeit I will caveat you know I'm sure there's probably some Elon timeline b like embedded in this and as SpaceX say they make the impossible late they tend to deliver on what they say it's not necessarily always on time. Um, so they're kind of the rough figures to help kind of people contextualize the the magnitude of this and and maybe lastly I'll say in terms of like how many satellites that's on the power equation. Uh, SpaceX has filed with the FCC to have 1 million satellites for AI compute. Um, for context, as I said, I think the most uh contextualized figure is the 10,000 active Starlink satellites. So really they're planning kind of bigger than than than one can believe. It's kind of hard to grasp your your head on the the size of this opportunity >> and Optimus and and the Optimus robots are the ones who are going to be building that space base and launching that stuff, right? Like and that's that's the secondary play with Tesla is that like those once there's thousands of those millions of those, sure, they'll be in homes helping us with our groceries, but they'll mainly be on the moon doing this.

[00:21:28]>> Yeah, exactly. Yeah. In fact, actually last year um because we're aware kind of how complex all of this is, we actually put out an open source SpaceX valuation model. So this is an excel anyone can download. Um and included in that was an element of Mars where we had Optimus or Optimi the plural going to Mars. Albeit this this model was kind of before the whole orbital data center craze as well as the XAI acquisition and so really focuses on the core Starlink business.

[00:21:55]>> You think that this is a craze? Is this a craze or is it like reality?

[00:21:59]>> I I I say craze because I just see it on my X timeline every time I open it up every day. I I I do and we at ARC do believe it's likely going to to be a reality and just given the bottleneck on Earth. H really it's just a question of when. Um yeah, I probably I crazy is kind of a throwaway comment. I say that because sometimes I see people posting on X all the time about it and yeah, it's a it can be a bit overwhelming, but I I yeah, I think we would be of the house view that it's uh it's likely going to be the case. It's just really a question of when.

[00:22:30]>> Trillions of dollars in real world assets are stuck offchain. real estate, commodities, private credit, all locked behind outdated systems that weren't built for a global 24/7 economy. The fix: bring those assets onchain. Problem is, most blockchains weren't designed for that either. Feros is. It's a layer 1 purpose-built for real, a real world financial infrastructure that lets assets be tokenized, verified, and traded at institutional scale. We're talking parallel execution for serious throughput, compliance baked in from day one, and infrastructure that actually connects onchain and off-chain systems.

[00:23:02]They've got a $10 million incubator backing builders who want to make real a reality. Join the Atlantic Ocean test net and start building at milkroad.com/faros network.

[00:23:13]The way we build wealth is changing.

[00:23:15]Stocks, savings accounts, real estate.

[00:23:16]That playbook is getting rewritten in real time. Nexo is the platform built for what's next. It's an all-in-one digital wealth platform where you can earn interest on your crypto, borrow against it without selling, and trade a wide range of assets, all in one place with 24/7 support and institutional grade security. Oh, and by the way, Nexo is back in the US with new US clients getting 30 days of Wealth Club premier access. That means elevated interest rates, lower borrowing costs, and crypto cashback on trades, benefits usually reserved for loyalty program members.

[00:23:48]Get started at milkroad.com/nexo.

[00:23:52]So all of this, you know, a huge the whole kind of underground thesis for all of this is power, right? Is energy. And that's another huge part that you specialize in at Arc and that you wrote in the in the big ideas 2026. A lot of sections authored by you and a few others was about power. And what I found most interesting in here too, um there's a few different sections I wanted to ask you about. Um >> yeah, >> but you actually even at the very start of your energy section, you you guys wrote that during the internet boom, despite concerns about its energy intensity, economies became more energy efficient and the same dynamic could play out with AI thanks to significant energy efficiency gains. And I find that really interesting because that's always you know milk road historically was a crypto thing and people are like bitcoin sucks up too much power you know and now you have uh you know bitcoin mining companies converting to power companies um and and so many other things like that have happened um and I don't know about the bitcoin part we'll leave that out but I'm saying is that I want to ask you about this thesis that basically >> and I guess it relates to space as well this is a great example that it's like this will actually make power way more efficient >> so tell me a little bit more about that.

[00:25:02]Yeah. So I think like the the takeaway line is more energy is good. Like we shouldn't be afraid to build out more power. Um and and why we say that with such confidence is really it actually revolves around that concept of rights law again. So I mentioned cumulative doubling. In the case of power or energy you can pick power or energy depending if you're doing nuclear, solar or batteries. But historically throughout history every cumulative doubling uh the cost has come down in terms of construction. Um and that translates to lower electricity prices for all. And I mean that's what a lot of people complain about nowadays is the electricity prices are going through the roof through data centers, but really it's a basic demand and supply thing.

[00:25:38]You know, when you have excess supply of power, you're going to force the the prices down. Um and then as it relates to like more efficient, you know, if you look at the relationship between uh like a unit of GDP output versus the amount of energy back in the internet boom, everyone was like, "Oh no, we're moving the internet or we're moving the economy to the internet. this is going to like cost so much money and the electricity prices are going to go through the roof and and the reality is just when you keep building out power, like I said, excess supply, you're going to bring the cost down and and people will will uh find more efficient ways to use it.

[00:26:10]>> Is is is solar a good place to look these days? Because I just going to pull up your charts that you guys had here.

[00:26:15]Um, and we've talked on our show about TAN, which is like a a solar ETF, not to not to call out one specific asset, um, but that's one that a lot of our our analysts have discussed. they have in some of their portfolios that are uh milk road portfolios. But this I I found this chart was interesting. You have your energy intensity uh which I want you to describe in a second. But this about on the right this global power capacity additions per year dating back to 2015 and how the only thing that's really grown is actually solar power and I'm assuming that that's mainly China.

[00:26:45]like I'm assuming that that's because I I don't think that that's mainly like you know uh in the US or any other places that tell me a little bit more about that man is solar is solar the clear play is is the US going to start doing this is do do space data centers count as solar power I guess so >> yeah yeah so well answer your last question yes they do that that would be how they're powered and solar facing to the sun um so what energy intensity is it looks at okay for a given dollar of GDP output how much energy do you need so as You can see that the trend is turning down for most economies. China was kind of when it entered the World Trade Organization, it kind of spiked up with a lot of the uh heavy industries coming online. Um so that's what energy intensity is. As it relates to solar, you're 100% correct. A lot of it is in China. Um that's why you see the likes of SpaceX and uh Tesla looking to build out their whole domestic solar. I think they both have goals to build. I want to say it's around 100 gawatts a year solar. Um, however, as I mentioned earlier in the podcast, one of the problems with solar, particularly on Earth, is the lack of uh 247 power. It only works when the sun is shining for the most part. So, usually you have to pair it with a battery. Um, and that's really why we're very bullish on nuclear energy. Uh, it can be like a good firm base load power. It's clean. It can run 24/7. Um, but yeah, I mean, the solar economy is really taking off in China.

[00:28:04]If you look at any of the stats, it's kind of crazy how much they're building.

[00:28:07]Um and you know when you look at power being the bottleneck of AI hyperscalers at the moment that's kind of why there's such a a sense of urgency to bring on any power they can right now irrespective of whether it's solar whether it's nuclear or gas turbines.

[00:28:20]>> What so why is nuclear still such a small part of this chart.

[00:28:24]>> So nuclear has had a a long history in uh focusing maybe on the United States.

[00:28:30]So if you look back at that whole rights law concept I mentioned nuclear followed that uh really from the ' 50s when when it kind of started to proliferate up until the 70s this is because there were certain regulatory changes uh which really caused construction cost to balloon um and when people weren't really executing on projects they kind of turned away from it there were the incidents Fukushima 2011 three-mile island in I think 1979 in the United States which didn't really help the case but you know when you look at today's environment There are several executive orders to bring back nuclear and get it on that cost decline curve. Um there's the concept of small modular reactors which aren't aren't new but they're really emerging now where you can build a nuclear reactor in a factory and mass build them and help benefit from economies of scale and put thousands of nuclear reactors plotted around different economies. Um and then you also have micro reactors and and reactors on the moon. Again, the the really good thing about nuclear is, you know, it's base load, it can run 24/7, and it's also clean, too. Um, so we're we're big advocates of nuclear energy at Arc.

[00:29:33]>> Got it. Okay. You know, we've we've had I mentioned this to you before as well before we were talking, but we um on the subject of nuclear. One thing that's always stuck out to me is we've had uh Dr. Alex Wisner Gross on the show a few times from the moonshots and and he in his first episode appearance with us, he he shared an anecdote from the 1950s that nuclear was supposed to make um electricity too cheap to meter. That electricity would be cheap would be so cheap that it would be free. Um and that now you know compute is going to make intelligence too cheap to meter. Right?

[00:30:04]So that's always something I kind of lead with. And anytime anybody brings up nuclear I was like yeah you know and and he kind of painted that's like listen the resistance there is to AI right now you can't repeat the same mistake of being resistant like we were to nuclear back in the 60s and 70s or the US was or whoever whoever the leaders were at the time was just like you can't you can't repeat that mistake because if we they' kept going with nuclear the whole time we would be in a completely different situation by now in terms of power right so that AI AI is very similar um >> yeah in fact actually on that point and it's to chart you're currently on uh or maybe it's the next chart, but had this nuclear buildout continued and there weren't these crazy regulatory changes, we actually estimate that electricity prices in the US would be 40% cheaper than they are today. Um so again, that's why the whole oneliner is more energy is good, more power is good. We should just keep building power really results in net net benefit for everyone.

[00:30:57]>> That's great. Uh, and then the last I I kind of want to wrap this up soon, but I think you had a chart in here about how the investment in global power needs to increase to 10 trillion by 2030, which is pretty soon, Daniel, man. Like, that's pretty like, you know, that's that's that's closer to now than CO is.

[00:31:14]Uh, that's much closer. So, that's is three and a half years away. It's very very very close. Um, tell me a little bit more about that. what what are we currently at in terms of is that annual trend or is or annual spend or is that or is that total spend? How does that work?

[00:31:28]>> Yeah, so that that's cumulative. Uh so from now till 2030. Um and if you look at the chart on the right because I'm I'm well aware that $10 trillion sounds like a crazy amount of money. Really when you look at the past five years 2021 to 2025 and the previous five it increased around 1.5x the cumulative spend. 10 trillion is only doubling that. So it's, you know, it's not really an outlier as as it relates to looking at historical trends. And then really, if you look at the AI hyperscalers, I mean, they are probably going to drive a large portion of this. Um, when you look at all of the demand they want for all of this power that they're bringing on, they're trying to restart nuclear plants that haven't been started in decades.

[00:32:04]And and you know, I think it just goes to the concept again of more energy is good. And it's really an enormous opportunity both on the capital build front, but also on like the metering side. if some of the nuclear players are signing these PPAs or purchase power agreements with AI hyperscalers guaranteeing streams of income annually and yeah you know we're very excited about this space it's uh it's really starting to take off >> and this this on the left here you have cumulative global power capital investment and you've broken it down into sectors of solar other and battery and that the main one that's I mean battery is definitely increasing quite a lot here uh from the from the previous four-year period to the current four-ear period but but solar is the one that stands out as like a huge share of the pie whereas like previously like it's going to be have been 50% and now it's going to rise to like 60 or 70% almost double or triple. Is that where is that is that what needs to happen or is that what is happening?

[00:32:55]>> Uh well I mean if you just extrapolate all of China's trends which dominate solar I mean that's where they're heading and but as I said you have the US really opening up to this now particularly if uh orbital data centers are are are going.

[00:33:07]>> Okay. So that's that's data center that's orbital data centers that would make up that number. Yeah, predominantly China orbital data centers, but also the battery point you've called out is very important and stationary energy storage.

[00:33:19]I mean, if you look at Tesla's business, it's growing one of the fastest growing business lines. And that's because, as I mentioned, solar doesn't run 24/7, but you can pair it with a battery and then you really increase the capacity factor.

[00:33:29]Um, and you kind of help address that intermittent issue. Um, and you know, within within the the stationary energy storage, a lot of the batteries are sourced from China. So again, the US is trying to build up a domestic supply chain to help address that, too.

[00:33:42]>> Are they going to be successful? Can the US do that?

[00:33:45]>> Um, I mean, it's it's it's there's a lot of catching up to do when you look at China, but I mean, I would never bet against Elon and what he says. Maybe bet against the timelines, but delivering probably not.

[00:33:56]I suspect that the US will will get there eventually. for people that are investors um and maybe are you know there things are really popping right now like I said in the intro there's so many there's a lot of different categories um in the AI buildout that have that have been have done very well in recent times what what you know it feels like a lot of this is in the private market or even you know in something like China which is kind of a bit of a wall for a lot of people where where do we look where do we besides reading arcs research where do we look um what sectors do we look at to kind of be part of this over the next 5 years?

[00:34:34]>> Yeah. Um well, maybe just seeing as we're talking about nuclear, there's a couple of public companies worth calling out. Oaklo is one. H they have an agreement with Meta. Uh there's also another company called X Energy uh who has an agreement with Amazon. They actually just recently went public through an IPO. H both of those reactors are known as Gen 4, so they're the next generation of reactors. They're small modular so roughly a third of the size of a big Tor Simpson nuclear plant. H so so that that's on the nuclear side. Um on the space side as I mentioned earlier intuitive machines rocket lab we are also seeing the convergence of space and defense really where you know satellites are becoming important for national security that brings into the equation companies like Kratos defense and security solutions. It's a defense company. They have compelling ground station solutions to communicate with all of the satellites. Aero environment also have h solutions in space. H but they're more known for their kind of drone solution that worked well in Ukraine. L3 Harris is another one. Um both providing key components. They were I think they provided over 100 subcomponents to the Artemis 2 mission where just recently the the astronauts went around the moon. So, you know, there are so many ways to play it and I'd encourage everyone just to kind of pick your compelling companies in the space, defense, energy arena and really dive in and see what where they're exposed to, whether it's government funding or or h how are they positioned relative to the some of the cost declines we're projecting with rights law arc.

[00:35:58]>> Last question for you, Daniel. Uh you win the the SpaceX lottery and you get to be the only human that goes to the moon to oversee all the Optimus robots and you're you are their king. Do you take it or no?

[00:36:10]H. Uh, it depends on the timeline.

[00:36:12]>> They might kill you. They might they might turn on you and kill you or or And you'll be the loneliest person in the universe. But also have the coolest You'll be the coolest person ever.

[00:36:22]>> Maybe if it was in like 30 years, I would consider it. In the next like 5 to 10, I don't think so. I I have a lot to live and a lot to experience on Earth.

[00:36:29]I'd rather visit all the countries before going extraterrestrial and heading to Mars.

[00:36:35]>> I always laugh. Elon always said to that, you know, he wants to die on Mars, but not on impact. He wants to get there alive. So, >> yeah, that's fair.

[00:36:43]>> Certainly don't want to be the first one to go there. Let someone else who's >> he's going to send he's going to send a lot of Optimus crash test dummies up first to make sure that it's a safe landing when they get their >> Mythbusters episode.

[00:36:54]>> That's Yeah, exactly. Uh Dana Magcguire, thank you, man, for for all your insight. I mean, definitely one of the more unique episodes we've done. Uh a lot to look into. Um, and would be great to have you back uh sometime next year maybe to kind of give us updates on all this.

[00:37:08]>> Yeah, we'd love to be back. We can check in next year and see uh see where orbital data centers are at.

[00:37:13]>> Perfect. Want to stay ahead of the biggest technological shift in history?

[00:37:17]Subscribe now to get insights straight from the sharpest minds in tech and finance. Quickly, you'll note this show is for educational purposes only.

[00:37:24]Nothing here is financial advice.

[00:37:26]Investing always carries risk. Never invest more than you can afford to lose.

[00:37:30]Thanks for tuning in. See you in the next one.