First thing every space company should do is book the first available launch they can before they built the thing that they're going to launch. Before they built the thing, before they probably even know what they're going to launch. Booking a launch is such a good forcing function for a space company. So we founded the company January 1st, 2024. January 2nd, we booked the first available SpaceX rideshare launch, and we were like, okay, something is going to be on that rocket. I'm not sure 100% sure what it's going to be at this point, but something is going to be on there.
Sounds like the same advice for all the software companies. He's just going to keep launching and launching. Same thing.
Welcome back to another episode of the Light Cone. Today we're sitting down with Philip Johnston, the co-founder and CEO of StarCloud. StarCloud is building data centers in space to address the energy bottleneck that AI is creating here on Earth. Earlier this year, they raised $170 million led by Benchmark and became the fastest growing unicorn in YC history just 17 months after demo day. Philip, welcome to the Light Cone.
Thanks so much for having me.
So to start off, why data centers in space?
So we are very quickly running up on constraints on where we can build new energy projects terrestrially. And so by building them in space, we get access to this almost unlimited low-cost energy in the form of solar. Of course, there's other costs like the cost of launch, and the cost of launch is actually very rapidly trending down, you know, with new launch vehicles coming online, the Falcon 9 program and Starship on the horizon. And so that's the reason we're building data centers in space.
How did you come up with the idea? It's not sort of like the typical idea that I feel people come up with.
I'd actually been working, I was with McKinsey for a few years working with the space agencies of different governments, and I could see that the launch cost was very rapidly trending down. And actually in early 2023, I decided randomly on a weekend to take a trip down to Starbase, Texas, where they're building the Starship launch program even before the first launch, and not many people were paying attention back then. And I could see that they were just building an enormous capacity. So they're building these two Starship gigafactories. They're designed to produce something like three Starships per day. And because Starship is reusable, you know, on a sort of three or four year time frame, that could lead to us having thousands of times the capacity to get things to space. And so I started thinking, okay, let's just run the clock forward. If the launch cost was 10 times lower than it is today, and the launch capacity was maybe a thousand times more than we have today, what would make sense that currently doesn't make sense? And so I started thinking about some of the concepts from sci-fi that I remember as a kid, you know, like space-based solar where you have these huge solar panels in space, you beam that power down. And essentially, yeah, I just reached out to a few folks I knew in the space industry and got connected with a few others and started ideating on, okay, what would make sense if the launch cost was much lower than it is today?
I'm curious, are there any other ideas that occurred to you during this process when you were thinking of what people could build in space cheap?
Yeah, tons. I mean, we definitely looked at manufacturing in space, and there's VA. We looked at asteroid mining, and you know, obviously YC has AstroForge. We looked at space hotels. We looked at basically anything which could make sense if the launch cost was a lot lower than it is today.
And how come data centers was the best choice?
Data centers makes a lot of sense to be the first thing you do because you don't need to re-enter a product. So with things like asteroid mining or with manufacturing in space or space hotels, they require this very expensive re-entry process. Initially what we actually started doing was space-based solar. So huge solar panels and then you beam that power down, and it's a concept from I think like the 60s. Even Asimov in the 40s was writing about it. The problem with space-based solar is you actually lose 95% of the energy in transmission from space to Earth. And so for the first two months of the company, that was literally what we were doing. We were called Lumin Orbit because of that reason, luminosity in orbit. We were thinking, okay, once we get that power down, what are we going to be using it for? And even back in 2023, most new energy projects were being built for data centers. So we were like, okay, well, either directly or indirectly, this power is going to go into data centers. Let's rerun those numbers. So initially, we'd run the numbers to know what's the launch cost break-even that makes sense for space-based solar. And we'd come to a number of like $50 a kilo. So then we rerun the calculations. We were like, okay, let's spend a month figuring out what would the launch cost break-even need to be to make data centers in space break even. And we came to a much closer to reality number of $500 a kilo. And that's what we think currently that the break-even launch cost is. And so then we rapidly pivoted the company towards that.
So, Philip, you launched StarCloud 1 in November of 2025. And that was a real emotional moment. Like, can you walk us through what that was like?
Yeah, I mean, it was incredible. So just to tell people a bit about StarCloud 1. So we'd set this launch date, you know, 18 months in the future, and we initially started with something quite unimpressive. Like we were going to fly some Nvidia Jetson chips. They've been flown in space before, and we were going to try and do some new workloads on there. Yeah, so it was Addy that came on and said, no, we need to do something way, way cooler than that. And we actually ended up putting five GPUs on there. Three from Nvidia, two from ARM. But the most interesting one was this Nvidia H100. We did some crazy things like, for example, you have to put it through thermal cycling, and we didn't have time to book the vacuum chamber, the thermal and vacuum chamber, but we needed to know, okay, if we heat up this phase change material and cool it down, heat it up and cool it down, is it going to crack anything? And so at like 5:00 a.m. the day we had to ship it down, working through the night, and Ezra like dunking this thing in an ice bath to cool it down.
And then we pull it out the ice bath and I'm like, are you sure it's going to be fine with the electronics? They're like, don't worry about it yet. It's totally fine. They pulled it out the ice bath. Then we had these like hair dryers to heat it up and melt all the wax, or like hot air guns, and then we're dunking it back in the ice bath. It is a miracle that it works, to be honest.
You can just do things. Yeah. So this is the kind of thing you can do as a startup because we actually had a quote from one of the primes on what it would cost to do StarCloud 1, and they said $75 million to $100 million, and we did the whole of StarCloud 2 including the launch for $2 million.
So yeah, leading up to the launch, I mean, we got the most incredible separation video. I think we're going to play it now.
StarCloud 1 separation confirmed.
So half the time as it deploys, it will be behind the shadow of the Earth and you don't see anything. The other half of the time, it will deploy not into the silhouette of the Earth. And so I mean, it just is the most perfect deployment video I've ever seen. And this was played by Jensen as he walked on stage at the GTC conference this year. This five-story tall screen behind him playing the StarCloud 1 separation video.
So no, it was the most incredible feeling seeing this deploy.
Where was the team for this? Did you all...
We took the whole team down. Yeah. There was only 12 in the team at the time StarCloud 1 launched. So we took the whole team to Florida and we had like a hundred friends and family investors. Some of our team brought their family and kids down and like it was a really nice day. We're going to release a behind-the-scenes video of some of the family moments in that launch.
How long was it from like when the rocket lifted off until the thing deployed until you powered it up and were able to verify the thing actually worked?
Yeah, so it lifted off. We were all on what they call the banana bleachers. It's like the famous from the Apollo era, but it was in the middle of the night. So it lifted off at around midnight and then at 1:00 a.m. they closed the bleachers and everyone had to get on a bus and leave. And we're like, oh, I thought we were going to be able to watch the deployment. And then I'm sitting in the back of a cab like watching the live feed on SpaceX.com. Literally, I was, oh my god, it's deploying. I can't believe it's deploying. And then it took about 12 hours to make first contact, which is not bad actually. It usually takes at least 24 hours to get first contact. And then it's like a two-week commissioning period. And we had a whole bunch of software issues we had to work through. Like the whole satellite kept restarting every 2 hours. And what was happening is one of 20 different failure triggers was triggering and we didn't know which one it was. And we can only get a ground station pass at least every hour and a half, maybe longer. And so we had to manually turn off each one, turn back on everything else and wait for a ground station pass to know which one it was. So that took like 3 days just to figure out that software issue and then yeah, two weeks after that then we started commissioning and turning on all the different payloads and then we trained the first model, ran the first version of Gemini, we did the first fine-tuning of a model in orbit, did the first high-power inference on satellite imagery and all this kind of stuff.
So StarCloud 1 right now as we speak is just left Chinese airspace I think about to enter Japan.
It's going quite fast actually.
Yeah. So it travels around 17,000 mph. It's about an hour and a half to do one circumnavigation.
So I mean, by StarCloud 4 or so, this won't be like one. It'll be a giant network sort of covering the Earth, actually.
Yeah. From StarCloud 3 onwards, we'll have 88,000 and they'll all fly. This one is in what they call mid-inclination orbit, so it doesn't go over the poles. And that's another point here on how janky this one is. It doesn't even fly in the right orbit, but at least it enables us to test the hardware. The next one will fly over the poles, and that gives us 24-hour always in the sun.
In order to do that, you need to put thrusters on the satellite to move it into that orbit.
In the end, they will be deployed in that orbit, but you need thrusters anyway to maintain the orbit because you have drag from the upper atmosphere there.
What are the engineering and physics trade-offs you have to make this happen? Because what you basically are doing with StarCloud is you're getting infinite energy because the sun is basically the natural fusion reactor, it's just a giant mass that generates a nuclear reaction with the giant mass and you have infinite energy basically abundant and lots of space. But as a result of that, which we're constrained of that in Earth, you have a bunch of other really hard problems to solve like things around interconnect. How do you send all the data back and forth back to Earth? How do you handle cooling? People think that space is cold, but actually there's no, unlike Earth, there's no air to circulate it back. So you need some crazy solutions with cooling. There's things around radiation to flip the bits in processors. And there's probably other things around how to get the solar panels to become really big and expand. And many more. I mean, these are the ones that I could think of. I mean, there's like so many things that are just so hard to build.
Yeah, you are correct. There are many engineering challenges to be solved. The two biggest ones that are outstanding and most of our engineering team is split basically 50/50 down these two lines is number one, as you mentioned, how do you get rid of this heat in a vacuum, and then number two, how do we make the chips work in a higher radiation environment. Some of the other problems are being solved by other people. You know, the interconnect. We've just signed a contract with SpaceX for our next 20 satellites to have a Starlink laser terminal that gives us very high bandwidth, low latency connectivity. Other people are solving some other parts of this. The core ones we're solving are that. And so for the first one, for the thermal management side of things, we are building a very large, low-cost and low-mass deployable radiator. Now, radiators in space are not new. It's not a new physics solution. It's more of a manufacturing and engineering challenge because the International Space Station has had a radiator which dissipates lots of heat already, very similar mechanism, so a liquid-cooled loop. The challenge with it is making it cheap and light. So our radiator is at least 10 times less mass per watt of dissipation than the ISS radiator and 500 times less cost per watt of dissipation than the ISS radiator. So very low bench, very easy benchmark to beat on cost for the ISS. But yeah, and then on radiation, it's just a lot of ground testing in different particle accelerators. So we've done several rounds of testing at the Brookhaven National Lab particle accelerator for heavy ions and also at the cyclotron for high-velocity protons down in Knoxville. So we ship all of our hardware down there. My co-founder Addy runs in and out wearing this protective thing. Addy is a lunatic when it comes to this stuff. He's like, oh yeah, it's the same as flying transatlantic. Don't worry about the radiation. So yeah, basically, you expose it over a 24-hour period to the same radiation dose that you would have in a 5-year mission. And then all of that telemetry and data then informs our choice on both shielding and software to mitigate bit flips. I think we're the only people in the world now that know where both an H100, a B200, H200 will fail if you blast it with high-velocity protons and heavy ions.
Do you have to change any of the underlying electronics to solve the problem or you just build shielding around it?
Both. So we do build shielding around it. There's a lot of it is just component selection. So for example, all of the SSDs, all of the power delivery system, all of the power converters and everything else, we test 10 components and whichever the best one is, we pick. We're trying not to use space-rated components. We're trying to use off-the-shelf like automotive-style components because it's way cheaper and most people have never tested those in radiation chambers. I think that's one of the secrets of how SpaceX has actually reduced a lot of the costs of a lot of the satellites as well. They use not space-grade electronics which would be exorbitantly very expensive, but instead you take regular components that there's a supply chain and run through all the tests and make sure if it works. And I think actually this is one of the things that Starlink does as well to make sure that telecommunication satellites actually work in space because they're also in GEO which has more radiation. Are you in GEO as well?
No, we're in LEO. Yeah.
When you first told people that you were going to put data centers in space, what were people's reactions?
For a very long time, their reaction was, this is the dumbest thing I've ever heard. So yeah, we actually applied to YC once and we got... we then tried to raise and it took... we tried to raise $2 million at a $10 million post on a SAFE. It took 3 months and we got rejected from at least 100 VCs. And then even after YC, when we went out to raise after demo day, I think we got rejected from at least 20 VCs before we got to the first check.
This is one of these ideas that was like extremely unpopular with investors and if you had listened to what all the investors were telling you, you would have quit and not ever done this idea. I'm curious, why did people tell you that they were not investing and what did they get wrong?
People had a hard time visualizing a world with low-cost launch and to be fair, we're still not out of the woods on that front. You know, lots of things need to go right with the Starship program for that to make sense. So that's one thing. I think it just sounds too sci-fi and two, yeah, to make this work, there's a convergence of two factors. One is the launch cost coming down and people have to believe that. The second is it becoming way harder to build stuff terrestrially. Maybe two years ago that wasn't quite as obvious as it is today. Now people are like, okay, well they just banned building new data centers in New York. They're going to ban it in like seven other states. Now those two things I think converged to make it more investable essentially.
And I think the thing after hearing you explain all the engineering and physics, it actually is doable. It doesn't sound like sci-fi from first principles. It's like it's actually buildable.
The reason I had confidence in it is because we had such a strong engineering team. So if you have, you know, the world's best space engineers saying this is possible, and to be frank, my background isn't as a space engineer, but you know, I have full confidence in my two co-founders, one from SpaceX, the other was building satellites for NASA, saying it's possible, then you know, that's good enough for me.
Once you had the idea and then I say once you're in YC, how did you sort of break the problem down into like a plan for what to do? Obviously in software it's, you know, you think of your MVP and you launch quickly and you get feedback, but hard to do that with data centers in space.
I have a very tangible piece of advice for all space companies. The first thing every space company should do is book the first available launch they can before they built the thing that they're going to launch. Before they built the thing, before they probably even know what they're going to launch. Booking a launch is such a good forcing function for a space company. So we founded the company January 1st, 2024. January 2nd, we booked the first available SpaceX rideshare launch and it was cheap. It's like 300 grand, you know.
Which was how far out was the launch?
It was 18 months out and then it got pushed to 21 months out and we were like, okay, something is going to be on that rocket. I'm not sure 100% sure what it's going to be at this point, but something is going to be on there.
Sounds like the same advice for all the software companies. Just got to keep launching and launching. Same thing.
Yeah. And what we initially planned to launch was very unimpressive actually. It was like we were going to put a Jetson chip which has been flown on many satellites before and we were going to use it for kind of edge processing and things. We actually had a co-founder change quite early on and when we got Addy from SpaceX, Addy was like, oh, that's lame. Like, let's put an H100 on there. We're like, Addy, you can't put H100 on a satellite. He's like, sure you can. The amount of people who said it was physically impossible to run a chip that's that power dense in orbit, like people just thought it was like literally impossible to run data center grade GPUs on orbit.
What would have caused it to be impossible?
There's two things. One is the thermal side of things, they're very power dense, they produce a lot of heat, and the second is they had never been tested in the radiation environment of space. And both of those... so for the first one, like it's kind of a wacky solution, like we submerged the entire thing in this phase change material. So everything on StarCloud 1 is submerged in this phase change material. So it's like immersion cooling for all of the components, all of the power delivery, memory, everything is submerged. Like nobody's ever considered doing anything like that in space before. It's not a particularly scalable solution. It's not like you have quite a low duty cycle. You have to wait for it to melt and then it solidifies every so often. But it enables us to prove that you can run the H100 in space. The next one is this direct chip liquid cooling architecture that we can run continuously. But yeah, I mean, I give full credit to my co-founder Addy for coming up with some pretty wacky and crazy ideas.
To get to a data center that's like commercial scale where you can actually like make money on it, it's actually like profitable, it's going to require many steps. How did you like think about like sequencing out like the journey to get there?
So as I say, first step was book whatever the like first launch lowest MVP we can have available. Second step is producing something which is a product we can sell to customers. So that's StarCloud 2. That's a 10 kilowatt spacecraft that we can sell compute to government and military satellites. And then the third step is the product that we're going to be able to sell to hyperscale data centers. So that's the StarCloud 3. It's a 200 kilowatt, 3-ton spacecraft. It's about 6 meters long. With that we can fit 50 of them per Starship. So about 10 megawatts of new compute capacity per Starship. And we've just filed with the FCC for a constellation of 88,000 of those. So that means on the order of 20 gigawatts of new compute capacity.
And we can fit up to 10 terawatts in the dawn sun-synchronous orbit. So that's like 20 times the entire US power grid.
For context, what's the largest data center deployment on Earth? Because you're talking about 22 gigawatts. Oh my god. What's the largest one here on Earth today?
