Next in Tech | Ep. 277: Technology in Space

View Full Transcript

[00:00:00] Eric Hanselman: Welcome to Next in Tech, an S&P Global podcast where the world of emerging tech lives. I'm your host, Eric Hanselman, chief analyst for industry research at S&P Global, and today we're looking at technology in space. It’s an area garnering a lot of attention, and with me to discuss it are analysts Ellie Brown and Johann Vermey. Welcome back to the podcast you both.

[00:00:25] Ellie Brown: Thanks. We're happy to be here.

[00:00:27] Johan Vermij: Thank you. It's always a pleasure to be here, and it seems like you always get me on the show when there's some unidentified flying object sightings in the world.

[00:00:38] Eric Hanselman: We've got, we've got some identified objects, but there is so much that is unidentified that we'll be digging into. You’ll be looking at identifying some of the activity that's planned to take place, and some that's going up there, in an upcoming webinar, the Orbit as the Next Data Frontier webinar.

And we've got some links in the show notes to that. Hopefully our listeners will get a chance to catch it. But both of you are joining a larger crew talking about a whole set of different aspects that are involved around what seems to be a significant movement to try and push technology into space.

Now, not that we don't already have a lot of tech in space in various forms. We've got satellite constellations up there in low Earth orbit that are already giving us internet connectivity. There’s a ton of money that's moving into this, and the webinar's going to talk about some of the capital flows around this.

But I want to dig into what both of you are going to be talking about on the technology side. What is it that you're seeing, and what are the things you're going to be looking into?

[00:01:46] Johan Vermij: So maybe it's good to go back a couple of years before we look at today. From a technology perspective, I've been looking at satellite constellations for about six years, especially how they have been used in the energy sector, for instance, oil and gas, the emerging of IoT satellites to monitor remote pipelines, or earth observation satellites, hyperspectral imaging satellites to monitor methane emissions, et cetera, and forest fires.

[00:02:22] Eric Hanselman: Or construction activity like Kim Jong Un's new palace.

[00:02:28] Johan Vermij: Satellite imaging has come a long way.

[00:02:31] Eric Hanselman: And it's really been democratized

[00:02:33] Johan Vermij: That's the interesting part. If you look at it over those six years, we first saw a wave of startups launching, announcing plans for proprietary, constellations in low Earth orbit, especially IoT satellites, with NB-IoT or, uh, LoRaWAN, small messages, focused on niche use cases.

The second wave, especially driven by, I think, the Russia-Ukraine war, is that it got democratized. It was no longer, a defense pretext that governments use satellites to do Earth observation, but especially a large OSINT, open-source intelligence community was leveraging, satellite images to look at the building, constructions of Kim Jong Un or Putin's Valdai residence or even finding Russian EW, electronic warfare, units, for targeting.

So that saw a huge increase in Earth observation satellites. And currently, this year, we're seeing a, a huge movement on orbital data centers

[00:03:50] Eric Hanselman: It's been interesting to see the amount of enthusiasm, and I guess maybe what we should dig into is some of the ideas that are driving it. And then there's been a lot of investment. There's been a lot of investment in space in general. I mean, you have got Rocket Labs acquiring another imaging side company.

There's a lot of investment that's going into this. But on the data centers, when we start thinking about that, there are a set of ideas about, why don't we just simply, if power is a concern, if cooling is a concern, we can just launch data centers out into orbit, and now we've got the coldness of space, the sun's energy to power us. We’ve now got all the things we need, so perfect solution, right?

[00:04:41] Johan Vermij: Well, that, that was the

[00:04:42] Eric Hanselman: Long pause.

[00:04:44] Johan Vermij: Yeah, it was the idea. It was, I think in, the market map of the private spaceflight industry that I wrote back in 2023, I first touched upon orbital data centers or data centers in space, as a concept. A few startups were pondering the idea, and I remember the European Commission awarded a research grant to Thales Alenia Space to research the options of data centers in space as being a sustainable solution. Because in space we wouldn't have the issue of cooling, or so we thought.

But at least we wouldn't have any CO2 emissions. And from that point, I think currently we're looking at almost 2,000, no, 2 million satellites being announced, orbital data centers just this year. So, we're looking at a big boom. But if we're looking at what's possible, we mentioned the cooling.

Well, it's cold in space. But Eli, if you want to fill in from a science point of view.

[00:06:02] Ellie Brown: This is one of my favorite things, is talking about the difference between space as an environment and Earth as an environment. Because we tend to think of space in movie terms a lot, right? Where it's just like Earth, except cold and dark and, we forget a lot of the practical physics of operating in a space environment.

So, things just don't work the same in space as they do on Earth, and there's a lot of considerations that you need to remember, right? So, if you're operating a computer in space, you're operating in a vacuum, right? Which means that when it comes to cooling, heat doesn't just dissipate out. You have to actually physically move it.

It takes work to move against that vacuum and to move the heat out. So, your whole idea of, well, space is cold, so it'll just, you know, be perfect. No, no, you actually have to put work into moving heat out, right? Not to mention all of the other things that you have to keep in mind. Radiation.

Here on Earth, we're shielded lovely by our atmosphere. We don't have to worry about radiation so much except for during heat waves when we go outside and we're dealing with sunburns and stuff like that, right? But for a computer, we have cosmic radiation that can mess with the calculations of these compute systems.

So that's constant, right? The space environment is constantly irradiated, so you need to deal with that aspect as well. So, you need to actually build the infrastructure differently to account for radiation. And not to mention, if we're talking about sustainability, all of this computer hardware has to get up to space in the first place, right?

So sure, maybe you're operating without the emissions that you would see here on Earth, but launching all of that compute infrastructure into space comes with plenty of emissions of its own. And the kicker is, all of that emission, it's put into higher levels of the atmosphere where it behaves differently and in many cases is actually even worse for the atmosphere than what we're seeing with emissions actually on the surface.

So, it's a really complicated question that I think often is kind of glossed over. It's just like, "Oh, space is cold. Perfect. There's the solution." And we don't really dive into the actual practicalities as much.

[00:08:19] Eric Hanselman: The cold thing I think is interesting, and I want to dig into it just a little bit just as a frame of reference. Yeah, I did some digging around and the International Space Station has cooling capabilities because it generates a certain amount of heat, especially when the sun's shining on it.

You got to shed heat, got to do heat rejection somewhere in there. And they've got their system through a whole set of big panels and is able to handle 70 kilowatts worth of heat rejection. And the average data center we're talking about, even small ones, get measured in the tens of megawatts.

So, the heat radiation, what's necessary, because as you're pointing out, w- why is it that data centers use so much water? They’re actually using water as the heat transfer medium to be able to go shed that heat, to move that heat around more often than not. But the expectation is you've got the cooling medium, whether or not it's water itself, whether or not it's air that has a certain amount of moisture in it, that's going to be the medium to move it.

Space is this wonderful vacuum, where no one can hear you heat. This is the problem. It's just this issue of being able to figure out how you actually move that. So, you've got to use infrared radiators and all sorts of stuff, which are big and clunky, and I don't remember exactly what the 70-kilowatt radiators in the space station weigh, but it's a lot.

And the thought there starts to really put a cap on how big you can build something.

[00:09:54] Ellie Brown: Well, I mean, the big issue with launch, right, is always weight, right? You're always trying to minimize how much weight you're putting into space because that's what drives the cost up, too. And so if you tell data center providers, "Sure, you can put this in space," but your launch costs, you have to launch all this extra hardware, too, like you're, kind of underwater as far as the costing goes because that's just so much equipment, right?

To your point, Eric, it's really heavy, right? And in space, heavy is a big no-no. You don't want heavy stuff.

[00:10:29] Eric Hanselman: And that adds to the cost and also adds to the size of the launch vehicle you need, the amount of propellant that you've got to use to get it up there, and then we get right back to that question of, how much propellant do you wind up burning up into the higher levels of the atmosphere? How much of the high-level soot that's up there sprinkling little carbon particles all the way as we go up.

That starts to create a set of much broader concerns on the sustainability side.

[00:10:56] Ellie Brown: On the webinar, we're actually going to discuss some of that sustainability as well. So, looking at the cradle-to-grave mindset for a space launch and figuring out what actually goes into it from a sustainability perspective. So, we're going to walk through everything from the supply chain behind actually constructing rockets and orbital data center technology.

We're talking rare earth minerals. And then we're going to go through the launch in orbit, orbital congestion, and then, end of life, what do we do with all of this technology, as it evolves out, right? That's the other thing with data centers; you like to update the compute systems. That's a lot easier to do if you're here on Earth than you know, if you've launched your compute system into space.

Where does that old computer go and what do we do with it?

[00:11:47] Eric Hanselman: Here on Earth, that's a challenge that's large enough. Look at just simply doing simple things like disk drive management in data centers. That's a large, equipment waste stream that simply moves through this pipeline. Now, we have ways of being able to manage that on Earth, but a little more challenging to manage that same kind of capability on orbit

[00:12:13] Johan Vermij: Yeah. There's several companies working on in-orbit servicing. But we'll just have to see if the infrastructure gets to that point. The ambitions that have been stated this year, and when I'm saying ambitions, it's about filings with the telecom authorities, the FCC for spectrum.

Which is different from, okay, this is a contract, and if I file an FCC request for a spectrum approval for a million satellites, we'll just have to see whether or not these million satellites will actually will be built and deployed. Looking back at the 2023 analysis of the new space economy, I looked at the FCC filings and based on all the proprietary constellation filings and spectrum filings, I would estimate that the growth of IoT satellites, Earth observation communication satellites, et cetera, from 2023 to 2033 would go to like almost 100,000 satellites in orbit. We're seeing that a lot of these IoT satellite constellation providers have had to revisit their business case. It’s expensive and I just read the 2026 report by the European Space Agency on the new space economy, which estimates that currently there's 15,000 satellites in orbit.

So, we're way short of that 100,000 that would be there in five years. Zooming in specifically on SpaceX, which is obviously the prime example of the new space economy. Back in 2023 SpaceX had the ambition to launch over 40,000 satellites during this decade. And currently what I'm seeing is that they have about 9,000 in orbit, so way short of their aspired 40,000.

But still it's 60% of all satellites in orbit. But if I'm looking at the orbital data center ambitions, SpaceX has filed for a million data centers, Orbital has filed for 100,000, Star Cloud 88,000, Blue Origin 52,000, Cowboy Space 20,000 satellite constellations. And we're not even looking at the, the Chinese plans to build out.

Looking at this year’s space activity, we've seen globally 324 launches, which is 25% up from 2024. Imagine the number of launches we need to do annually to get up to that one million constellation. And looking back at the old space economy, when NASA and US government launched the big GNNS, navigation satellites up into geostationary orbit, they were built for many years.

I think the Russian GLONASS system is now falling apart after 50 years, without renewed satellites. But in low Earth orbit, what we're talking about is five, 600 kilometers up from the Earth. The satellites are much smaller, much less rugged, so they have a lifespan of five to seven years maybe. So, the amount of launches you need to do annually just to replenish that five to seven year lifespan is immense.

[00:16:15] Eric Hanselman: And these little guys are going to be falling out of orbit on a regular basis, so we just need to be launching every hour on the hour. No problem, right? Well, because it's something we see with Amazon's aspirations, with Kuiper and now Leo. Their whole focus, I don't remember what their count is up to now, but their thought had been they were going to have thousands and thousands.

They’re still getting up, trying to break 1,000. All of the various other existing providers, Eutelsat, these are things that they're logistical complications to make that happen. But I'm curious from a, a satellite standpoint, you talked about what really has been the shift from these large geosynchronous satellites to low Earth orbit satellites that are smaller.

I’m wondering, is that a path forward? Because if we start building more small satellites, then you don't have quite the same kind of cooling concerns. Maybe you pop a couple GPUs in each individual one, they talk amongst themselves. You now have one big mesh, and that's how you build your data center in space.

That was kind of the prospect that certainly Starlink had been talking about, and the SpaceX capabilities that were up there. But it seems like even that starts to get a little challenged, just simply in terms of the number of launches and just the density of devices that are on orbit.

[00:17:31] Ellie Brown: Yeah, I mean, we've seen a shift towards these smaller and very formulaic, CubeSat satellites. So, instead of everything being a custom build, you have these nice Lego blocks of satellites that are a specific size, specific configurations. They're designed to do that sort of meshing, and they're small enough that a lot of times you can actually piggyback on an additional launch, right?

So, we're seeing some of that sort of activity where a company will develop a CubeSat and then they'll say, "Hey, can we hitch a ride on this existing launch?" and we'll just pay for the extra weight that's going up. They don't have to launch it themselves, right? So that is a kind of evolution of the economy that we're seeing as part of this new space economy. Is that sort of mindset of, okay, how do we make this easier, right?

But you still run into that issue of you need to get them up there, right? And so, yeah, they're smaller. Yeah, they can hitch a ride, but, the smaller systems, you know, you need more of them. And if that's the case, you still need those regular launches. You still need it to work, right? I mean, we're talking about the launches that do work.

Part of the problem is even though launch technology has gotten much better, obviously we've seen some catastrophic learning experiences, I think is what they're being called by the launch operators. They don't call them failures anymore. They call them learning experiences or something like that.

But we're still very much not great at consistently launching things into space, and we really need to streamline that, that pipeline and that operational system if we're going to be able to get anything really systemic up there, right? It's gotten better, but it's still not perfect, and we keep running into these failures that set things back quite a ways.

[00:19:30] Eric Hanselman: Which gets us back into that financial consideration of what does this cost? We’re now starting to see insurance for vehicles, now that there's a market for actually managing what that process looks like, but it is not without risk

[00:19:44] Johan Vermij: Now if we're looking at the cost, if you look at the old space economy, with the geostationary satellites, you had probably the cost of launching a rocket was like $170 million, which comes down to almost $30,000 per kilogram. And these big satellites, they were like 1,000 kilograms or whatever. To Ellie's point, the CubeSats that we're now seeing in low Earth orbit, they're much smaller and much more lightweight. The CubeSat is like a 10-by-10-by-10-centimeter cube, and the satellites are like a 6U or a 4U. So, four or six or maybe 10 of these cubes in various configurations.

But looking at the economics, Kepler Space, Kepler Communications, a Canadian company that actually has one of its first orbital data centers in place, which is a satellite that has 44 NVIDIA GPUs. I don't even want to make an estimate of what the amount of chips an AI hyperscale campus has these days.

Looking at those economics, we're coming from $30,000 per kilogram. To make it feasible for one million orbital data centers, the cost needs to come down to $300 per kilogram, and I think we're somewhere between four and seven, depending on who's paying for it. Then on the piggyback thing, SpaceX Starship could potentially lift 100 satellites at once.

But we're still needing a couple of launches before we get to a million

[00:21:46] Eric Hanselman: We've now got, China's coming into market. They've now successfully recovered a booster. So, this is the next stage. We're starting to get to a SpaceX style, reusable rocketry kinds of environment. That'll bring things down, right? And pretty soon we'll be booking Ubers for our satellites to space, right?

What could go wrong?

[00:22:07] Ellie Brown: I do think the shift towards reusable technology is critically important, right? Obviously, doing everything one at a time, and not reusing, it just, it was never going to be feasible, right? So, we're in a good spot drifting that way. Part of what we need to keep in mind though, is that it's not just one or two pieces of the launch system that we need to think about reusing.

We need to think about every time you launch, you've got to make sure that you've got a consistent launch system. You need that fuel, right? Fuel for rockets is not negligible in terms of cost. If you're reusing, you have to make sure that you can reuse as many times as possible. Reusing twice doesn't really quite count as reusing, right?

You need the stuff to not blow up on the pad because then you can't reuse it at all. You need to have a consistent launch pad and site that allows for that reusable consistency. And then you need the regulations that allow for those launches to happen at the cadence that you're looking at.

Right now, a lot of the launch system, regulatory system is still very ad hoc. You have to file for environmental monitoring for every single launch and it's evaluated on a case-by-case basis in most cases. You need to plan out satellite launches individually. It's not anywhere near what we've got with a general airport system that ferries millions of passengers around the world, right?

We're going to need to develop systems like that, not just individual reusable technologies, but an actual system in order to make this happen, and that's going to involve regulation as well.

[00:23:49] Johan Vermij: Yeah, I want to double-click on that, Ellie. It’s a good point that you mention here. So recently there's this company called Reflect Orbital that filed for, it wants to bring up 50,000 satellites that reflect sunlight to Earth, and it's like their pitch is, "Okay, we're providing a carbon-free alternative to city lighting.

And we can optimize solar arrays on Earth, bringing light to them when it's dark, so you get those revenues instead of needing to fire up gas powered plants." But all this light pollution that comes from it, and even now if we look up at the stars, sometimes we're seeing that string of pearls called Starlink flying over.

Looking at a million data centers each with their solar panel arrays at inclinations that reflect light to Earth, and this direct sunlight-to-Earth reflection, that may pollute, that light pollution alone may change wildlife cycles. And the worrying thing is that the FCC granted the application for spectrum, and it said that non-RF environmental factors like optical reflectivity and light pollution fall outside its statutory purview. So, the FCC doesn't really look at the environmental impact. So, we need more governance, and the other thing, more governance probably will be driven by sovereignty as well. We're increasingly becoming dependent on satellite intelligence. Europe is recognizing it's dependent on US tech for military information.

So, we're seeing a, a slight push in sovereign initiatives in Europe, with more investment, increasing investment, than the US. The US investment in space sector, government investment fell about 3% last year, while Europe's investment in the space economy grew by 12%, but it's still trailing the US by far.

But we're also seeing that we're getting dependent on those private, or commercial constellations, and that one man in charge may decide to switch off service for a country or a region in the middle of a war or whatever. So there needs to be more regulation.

[00:26:28] Eric Hanselman: Will that ever happen, Johan? Well, this is, this is exactly the challenge, right? As with any set of interesting new technologies, we go through this journey of interest, early speculation. Now, especially in orbit delivered connectivity, we've got a strong dependence that's been built out, and now we're starting to suddenly come to terms with some of the different aspects and the implications about this.

Now, again, from the governance perspective, from the scaling perspective, from the sustainability perspective, there are so many different pieces that are all coming together around this. It seems like it's getting to be one more of these technology issues that we wrestle with day in, day out.

 

[00:27:12] Eric Hanselman: Took us a while to talk about AI, but yeah, there's so many different pieces to this.

[00:27:16] Johan Vermij: Yeah, absolutely. But we've been focusing on the limits of possible and the critique that there is, and the concerns that we have from an environmental perspective. But we should probably also look forward to the opportunity. I grew up reading Asimov, Robert Heinlein, Arthur C. Clarke, et cetera, and somehow it always starts with the Earth being demolished and screwed up before we get to something new. Even Star Trek, and Star Wars. But, looking at the growth, we're not entering the Star Trek Federation era anytime soon. But looking at maybe these, these data centers in space, they might be ambitious. And we're a long way from getting to 1 million, 2 million data centers and I do feel that it's not realistic to see that they will be the alternative for AI compute on Earth anytime soon.

But as we're seeing the amount of data captured from space, the Earth observation satellites, hyperspectral imaging, synthetic aperture radar, camera, optical sensor that's looking at the Earth and gathering information that helps us to detect fires or methane leaks or any other thing that is helping us, that data needs to be processed.

And then we're depending on our backhaul, a terrestrial backhaul. Can we send that satellite image data to Earth and then start processing the high-resolution images, or do we use compute capability in space to do that pre-processing and determine, okay, this 99% of the Amazon forest is not on fire, but this 1% is on fire, and we're sending that data down to the fire department. So, there will be that need for orbital compute. and there's already other interesting use cases developing as we get further into space. Asteroid mining might be a few years out, but Ellie, how about lunar mining?

[00:29:37] Ellie Brown: Yeah, we've talked about that as part of my other coverage area with quantum, right? The possibilities of mining specific things that we need for quantum computers, like helium-3. The moon is really rich with helium-3. So, we've discussed with some companies about, yeah, we're going to the moon, we're going to mine helium-3, we're going to ship it back to Earth, and that's going to power the quantum computing boom.

So it's all connected, right? To your point, Eric, there's just so many moving parts, and they're moving in very exciting ways, and we'll see kind of how they all come together. And to Johan's point, hopefully they do so in a very positive way and not in a start of a sci-fi film where we have to watch the destruction of the Earth before we get to the good stuff.

[00:30:23] Eric Hanselman: Ho- hopefully not too soon. But, that, that's a good upbeat note to take us out on. Thank you both for raising the level to look at that future. But we are at time for this episode. Thank you both for being back on.

[00:30:35] Johan Vermij: You're welcome. It was a pleasure as usual.

[00:30:38] Ellie Brown: Yeah, it was a fantastic discussion. Thank you.

[00:30:41] Eric Hanselman: And I will point our listeners to the webinar that's coming up, so you'll be able to get this in even more detail. But that's it for this episode. Thanks to our audience for staying with us, and thanks to our production team, including Sophie Carr, Feranmi Adeoshun, and Dylan Scheible on the marketing events teams.

If you like this episode, please subscribe or like us. I hope you'll join us for our next episode, because there is always something next in tech.