THE SHIFT TO SPACE:
THE RISE OF ORBITAL COMPUTING -
ROB DEMILLO

CRA: Rob DeMillo, welcome to the Tech Burst Talks podcast. Rob: Thanks Charles. Thank you for inviting me. It's great to be here. CRA: I gotta admit, I'm really looking forward to this because normally on the podcast I talk about topics that I relatively know at least a little bit. But today I'm going way out of my depth because we're gonna talk about orbital data centres. So putting compute in space. So this is gonna be a bit intimidating for me, but I'm hoping that you can educate me and the audience as well Rob: I hope I can as well 'cause that's, that is my job. CRA: Before we get into that, why don't you tell the audience a little bit about your background, because you've had quite an interesting career. Rob: I had a long and winding road that led to the same place. My degrees are in computer science, mathematics and, and weirdly astrophysics. And I had dreamed about becoming an astrophysicist until I realized that, you know, I'd be eating ramen noodles for the rest of my life. So when I went to graduate school it was computer science, heavy focus on math and analytics. And I left there to [00:01:00] work for Jet Propulsion Labs. And I was a image specialist for the Galileo spacecraft, Mars observer, God rested soul. And to give you some idea of my age the Soviet Vene spacecraft, if you remember those guys. So I did that for about seven years and left there to go work at the FAA via Lincoln Laboratories. They're the folks that invented radar and we were working on a TC problems, air traffic control problems, and basically doing the same work I was doing at JPL, but it was in the radar spectrum. And so we built Terminal Doppler Weather radar which is now employed everywhere in the world. When that project completed on time and under budget, by the way when that project completed they were gonna put me on Oceanic air travel and I couldn't think of anything more boring. So, one of the vendors that we were using. Was spinning out a startup company and they asked me to go with them, and I did, and I never looked back. And I then began a 30 year career doing entrepreneurial work and initially mostly in the media space. So media technologies, so, video [00:02:00] production like this, right? So video production video distribution distribution channels computer graphics computer graphics for animation, computer graphics for special effects. And if you track that. Going forward it, it leads pretty quickly into data centers when data centers start to show up. So the last 15 plus years of my career have been in and around data centers and, and how to do compute in those data centers in ways that made sense to get technology off-prem. The last company that I founded and sold was a company called Nimble Collective. And that was founded with, a number of luminaries from the computer graphics world. So the folks that did Toy Story and Madagascar and Golem from Lord of the Rings and, and those sorts of folks. And we took the animation platform animation, workflow that animators use. And pushed it up into the cloud. The cloud was ready for it at that point. And the rationale behind it was that animators are a nomadic bunch two years in New Zealand, [00:03:00] two years in Toronto, two years in Los Angeles, you know, that kind of thing. And it's hard. It's a hard life. So by putting the animation workflow into the cloud and allowing people to use the tools that they have always used in animation you could live in Des Moines and work on the latest Avengers movie, right. So successful, we, we sold that to Amazon, became part of Amazon Studio. I think that's called Amazon Nimble Studio Now. And after that I went to work, went to work for an architecture firm, which was way right turn left of field of treatment, right. So, the reason for that was when we were looking for, for partners for our animation platform, we we're, we're trying to figure out if we were gonna be able to pay for our next meal. And so we were looking to the verticals and it turns out that the workflow for architecture is extremely similar to the workflow for animation. So we started talking to architecture firms and when we got bought by Amazon, I let the folks at one of the companies that we were talking to, which was Skidmore, Owings, and Merrill, know that I, you know, we were being sold [00:04:00] and be working with this nice guy named Eric, and they said, well, we've been, we've been, they called me back and they said, well, we've been looking for a CTO for three years. So I thought, okay, this is interesting. So I went over to SOM for about five years. And it was largely to do what I did in animation, but for architecture, moving their entire workflow up and into the cloud. SOM, like all large architecture firms, had offices all over the world. Everything was very centralized in each office. It was hard for them to communicate back and forth. So by doing cloud, by pushing it in the cloud, it it. Circumnavigated that problem. So I left there after five years and my, my goal was to go into sort of a soft retirement. I was working with spark Labs global as you as you know them as well, Charles. And then I got a call last summer from a group called the Mandala Space Venture. Group and, and Mandala Space Ventures, if you look them up, they are a accelerator for space tech companies. And it was founded by a guy named Dr. Leon [00:05:00] Alkali. And Leon is XJPL. He is 32, 33 year JPL guy ex fellow extremely well respected. He's, he put the insight lander on Mars Grange orbiters around the moon orbiters on earth. He understood Space Tech back and forth. And so he was he, he, he left JPL specifically to start Mandela Space Ventures. And it, it's incubated a number of things, but one of the things that incubated was Sophia. So he called to talk to me about Sophia Space. They were doing a nationwide, worldwide, actually I think search for CEO. And it turns out that if you do a Venn diagram of people who worked at JPL, people who understand data centers and people who understand deep tech, it's like a tiny little.in the center. So, so we talked for about five months, and in December I agreed to come on board as CEO. January I took a position we did a few tweaks to the business plan and immediately went fundraising. That's one hell of a journey. And we're [00:06:00] gonna come back to Sophia's space in a minute, but there's one more thing I'm dying to ask you about. You've been involved in 9 startups over the years, and 7 of those have exited. So you always hear about how hard it is to have an exit. Yet you've managed a 7 out 9 hit rate. So what's the key to your success? Rob: Dumb stupid luck. I wish I had a brilliant plan I could lay out for you, but yeah, honestly, it is a combination of skill and luck. If you over index on one, you, you lose out on the other. I have always been interested as, as is clear from Sophia's space, I have always been interested in that technology that takes you right up to the edge. That takes you right up to the hairy edge where you look over and it's it is not for the risk averse, but it is the place where all, all the the coins lie, right? So when, whenever you go to those edges, people wanna do something. One of my startup companies was the first to do video advertising on a mobile device. One of my companies was the first to do, cloud-based animation, so all this stuff where it's [00:07:00] like the tech is almost there, just needs a little bit of a shove and you go over. And what's interesting about Sophia is this has that all over, like, like every, every green flag that's waving in my face about Sophia is this, this is that in truckloads. CRA: You mentioned that you were talking to Sophia's space for a number of months because you were also contemplating a soft retirement. Rob: yeah, Not, not, not to be that guy, but they called me when I was in Venice, Italy. and it was interesting because, you know, if you look at my life as an aura, Boris. I did touch on all those areas and they were all incredibly fascinating. And I would be lying if I never thought about like, you know, if you put a computer in orbit, what does that really mean? Like, like how does that, how does that all operate? How do you do Interoperation? And it's kind of been noodling in the back of my head for a while. When Mandela called and when Leon called I had the same reaction that a lot of people had until recently which was, [00:08:00] wait, what? You know, that sort of moment okay, tell me more. I'm listening. And so we talked for a while. It was less about. Me convincing them to hire me as their CEO and more about let's talk to each other. Leon's a wonderful man and, we both wax poetic for, for long time. You don't wanna be on a phone call with either of us. So we had these really sort of interesting conversations that not only spanned what Sophia was doing, but just the philosophical and morality background of doing this kind of thing. It was a good fit and it made a lot of sense and it collected so many things in my brain. It is an impressive twist because I remember seeing that you had left some and I thought you were just gonna relax for a little bit and instead you pop up, as the CEO of an orbital data center company. And I'm like, how the hell did this actually happen? So why don't we talk a little bit about Sophia Space. Rob: Sure. So lemme talk a bit about the origin story. 'cause that's pretty interesting too. So, mandala Space, Leon's [00:09:00] venture company had a collaboration with Jet Propulsion Laboratories JPL in Caltech. And by the way, that's the, that's gonna be a common theme through this conversation. They got a big, big chunk of grant money behind this. Let, let's see if we can put up a solar cell in orbit and collect enough solar power and beam it down to earth and, and see if we can do power collection that way. And so after 18 months the results came back that you know, the math says yes. The economics, say don't do this. They basically started to, to dismantle all that. And Leon to his credit, he raised his hand and said, well, if we put a server on this thing. And so that was the origin story. So flash ahead to, early 2025 and I come on board. The, the key to what we're doing here, and, and this is a differentiator for us, is we've rethought what it means to be a computer, right? We have a, a lot of competition and, competition is, is great because it validates the market. And there's certainly enough to go around for everybody. Right. But what we're doing that's different is we are [00:10:00] not taking terrestrial servers at terrestrial technology basically, and retrofitting it for orbit. We are not thinking about terrestrial data centers. And how do you shove that whole thing up into space? Right. We're we're going way back to, to first principles and saying we're in orbit, there's radiation, the servers are moving. Energy is free. It's 4 Kelvin in orbit. What can we do with all of this? And we came up with a device that we're calling a tile. And a tile is it's an object it's 1m x 1m x 1cm It looks like a tile, and it does look cool, it looks that way for a reason. So it's a completely solid state device. It wouldn't even run on the planet. At the top of the, of the tile is a solar cell that collects energy from the sun. Once you do the math, you wind up with about 400, 450 watts of power and the centre of this Oreo cookie is the compute layer. We have four, the [00:11:00] equivalent of four enterprise servers on board. So think of them as quadrants in each of that, in each tile, and the servers have been redesigned using off the shelf parts. We're not building anything special, but they've been redesigned to be flat so that the heat profile is straight down onto the board and then underneath that tile. Compute layer is a proprietary alloy that dissipates heat. And whenever you have a big delta between your heat source and the, the thing that you're radiating heat into it's a lot more efficient, right? And so, 4° kelvin, almost as cold as you're gonna get. So, now we've got the re object that is redesigned server. And if you were to run it on the earth, it would overheat within minutes, right? It just flat out would stop. So it's got passive cooling. We get free energy and you don't have the problems of real estate that you have here on earth. The next trick is the operating system. So we've designed an operating system that sits on these tiles that becomes aware of other tiles that's attached to, [00:12:00] so these tiles can be sort of infinitely stacked together. They communicate with each other back and forth, and they do the work that you would have an IT technician do in a data center, right? So they do load management heat balancing. If one tile's too hot and one tile's too cold, it'll, it'll move processing over there. It'll look for errors. It'll, it'll correct. In case of a dead tile, it'll run around the dead tile. Firmware, upgrades, security patches, et cetera, et cetera. So now we have a system that is modular. Each tile understands everybody else, and at that point you can start stacking these things together in different configurations. So that was a long explanation for one question, so apologies for that. CRA: The first thing I want to go back to is you've got this great, innovative technology that's totally cutting edge, and it's also new to market. So I know you just got some pre-seed funding, but now you have to go out there and identify the market to man, and who should we be collaborating with or who should we be talking to? So where do you go from here? What are the next steps now that you've got the pre-seed [00:13:00] funding? Rob: we've been talking to people all along before we even start, so the pre preceded. We closed actually just a, about a month and a half, two months ago. And you know, so thanks for bringing that up. And we're actually going back into a seed round raise almost right away 'cause we're taking advantage of, of certain market conditions. With the proceeds from that pre-seed, we, we've hired hardware engineers, we'd hired software engineers. We've set up a laboratory in, in Pasadena. We started the process of breadboarding the tile. So we're taking this outta the realm of math and cad, into reality to, to make those tiles. we plan on having completed demo tiles by the end of this year that we will we'll test the math by putting them, putting them into a chamber that will simulate orbital conditions, so like radiation, blah, blah, blah. Make sure that, that the tile is operating the way it's supposed to tile work. And then next year we plan on launching a couple of demo tiles to, to. You know, test it in orbit. So that's the timeline going forward. As to how we are finding clients. Once this all broke and went public, they're, they're finding us. And there, [00:14:00] there's, there's some conversation around that too, right? So, department of Defense is obviously someone that's interested in us. We have a lot of collaboration with companies like Axiom who are putting up their. Their own station nodes there are conversations like that, that are currently occurring, but what's happening in the world right now? So defense and weather and air traffic control and maritime management and all the rest of it. There's a, there's a concept called the LEO economy, and LEO stands for low earth orbit, and low earth orbit is anywhere between 200 kilometers. I think a thousand or 1200 kilometers. And, and in that layer there's a lot of attempts at commerce happening. So there's infrastructures being set up. There are new satellites going up. There are new space stations going up. There are communications pipelines happening. So it, it's a little like watching this will appeal to you. It's a little like watching a smart city become built. we're building this new thing from, from scratch. So up we go and you know, we're gonna need somebody put in the plumbing, we're gonna need [00:15:00] somebody put in electrics, so all this stuff is happening. And Sophia's going to be playing a part in that, like we are the compute for the Leo economy. So how do you get these tiles into space? Because they know it's great once they're there, but it's not like you can just use a big slingshot. Rob: Actually there is a big slingshot, but that's not us. There is there, there is, what is it? Spin orbit. It's a good question. So there is a a process we call origami, which is essentially what you're thinking of. So these tiles they fold together, right? And so a, a large array of tiles can actually be made quite small. And once they're placed in orbit, they will self configure. Right? That's, that's part of our, our, our portfolio, our, our patent portfolio. And so, when they're, when they're released, they will simply expand. There's no on switch. When they, they, they get enough power, they'll simply start up boot and, and, and go from there as to how we get it in orbit. We have no plans whatsoever to become a launch company. There are a lot of folks out there doing a lot better work than than we ever could. The [00:16:00] SpaceXs and the blue origins showing up on the planet now to get people into orbit, which is the other key. The cost of getting an object into orbit per kilogram is dropping dramatically. It wasn't that long ago where it was in the 25,000, $30,000 US dollar per kilogram. Range, which is, which is prohibitively expensive. It's now in the couple thousand range, and, and by 2030 it'll be under, you know, $1,200 a kilogram. there's so many satellites in low Earth orbit right now. You've got starlink, you've got companies and governments in the Middle East and Asia putting up their own networks and many others, but aren't we gonna get to a situation where there's just so much up there that the space junk is just gonna clutter everything? Rob: definitely a bunch of space junk up there, but there's a responsibility of the satellite owners to deorbit these things. Which simply means exactly what it sounds like. It crashes into the OS atmosphere and burns up. I don't wanna minimize this 'cause that's a real problem, but you, you'll see in the press, there's these computer graphics that show like this, this, this shell around the earth of, [00:17:00] of crap, right? And all that stuff is sub pixel, right? So, so when you go up 800 kilometers, your surface area expands tremendously, right? So it's not the surface area of the earth, it's the diameter of the earth plus 800 kilometers, right. So it, it's a larger area. Having said that, there is a lot of crap up there. There are a number of companies that are looking at removing debris. For morbid 'cause we were pretty careless for the first 40, 50 years of doing this. And so there's a lot of bolts and chips of paint and, and small objects flying around in that shell which do pose a problem. And they pose a problem now and they'll pose a problem for a while until there's, there's a cleanup methodology adopted and I've seen some fairly clever ideas for doing this. Satellites themselves. they have to be debited after a certain period of time. We've gotten better at making sure that we keep the debris levels down, but it, I don't wanna minimize it. It is, it is an issue, which is part of why the tile arrays, the operating system runs those tile arrays will route around dead tiles or dead quads in the tile because they, they probably will take a hit [00:18:00] at some point. I want to take this all the way back to the beginning, so I'm gonna ask you some very simple questions to make sure that I and the audience understand this. So, why would we put data centres in space at all? Rob: The slight change to our business plan is we talk about this in terms of putting compute in orbit. Which leads to an orbital data center. So the first few years of our existence, we are working specifically with edge computing around satellites and stations that need compute local. Right. And I can talk about why they need that in a moment. But. In a very short period of time, and we're talking in the 2030 2031 timeframe. Yes. We will be using these tile technologies to put up our own orbital data centers, which will operate like a data center, a one megawatt data center. Using our tile system is a small object. It's a 50 meter by 50 meter device. Right. Which is half the size of the ISSI think, if I remember my math. So, the reasons that you want to get compute off the ground, and the reason you wanna get data centers at, at least partially off the ground [00:19:00] is threefold. 1 is power. So power is free. When we put our orbital data centres in orbit, they'll be in something called a sun, synchronous orbit, which you can think of as being pole to pole They're mostly facing the sun almost all of the time. There's about three months a year where there's a few minutes a day where they go into Eclipse. But these things have batteries, so they'll be fine. So powers free. 2 Cooling's free. The thing that nobody talks about, about data Centres on Earth, and the reason they don't talk about it, is there's no real good solution. 'cause, 'cause the laws of thermodynamics are real. They generate a lot of heat, and heat is energy and you can't destroy energy. You can just move it. Right. And so when you talk about a terrestrial data center 40% of the power that goes into a terrestrial data centre goes towards cooling. So you're either shipping the cooling to the earth. To the atmosphere or the ocean. Those are kind of your only choices. I mean, you could, you could move it into oil and then the oil becomes a contaminant problem. So you could have all the free energy on [00:20:00] earth that you want. You know, people are talking about nuclear power centers in your data centers. She okay. But it still heats those things up, right? Which is definitely a local problem. If you think about the urban heat transfer problem you've got that, you know, 20 fold with a data center, right? You've got this immense object the size of, you know, a warehouse putting all this heat, right, right into your local environment. And there's an argument to be made that if you think about that far enough down the road. You're gonna help with global warming, right? So when you put these things in space, you don't have that, it, it, it's just gone. The heat, goes into the biggest heat sink you can think of, which is the known universe, right? And so there, there's not really a heat problem. And then 3 that goes away is nimby, right? Right. So we, we don't have anywhere near as many data centers as we want. We should clarify this just in case NIMBY stands for. Rob: . So not in my backyard. Right? CRA: Not in my backyard. Yes. Rob: These buildings are, I mean, you know, they've been made more and more attractive. But [00:21:00] they're buildings. They were windowless giant buildings. Nobody wants 'em around. And so a fun factor here is that 30% of the United States' data centers are in Virginia. And Virginia said that that's about enough of that, right? We don't have enough data centres Nobody really wants them anywhere. And the projection is that by 2030 we'll have 3X as many as we do now. Right. So you don't have that problem in orbit either. So, so in one swipe, you get rid of all of this. Okay, so now we can put compute in space, but what I wanna know next is what are the use cases? I know you mentioned defense, but what else could it be used for? ? Rob: So we have amazing satellites up there right now. And more are planned. And these things have, have sensors that collect terabytes, sometimes petabytes of data. And the dirty little secret here is that most of that data is tossed out. And it's tossed out not because it's not good, but because the compute on the satellite is not powerful enough or the bandwidth is not high speed or fast enough to get information down. Back to the planet to do the [00:22:00] computing and have it go back up to the satellite. They do some quick high pass filtering and, you know, 80 to 90% of the data is just tossed on the floor based off of this kind of rudimentary filter. So, that's what's happening right now. And it's happening every. But with edge computing right there, you don't need to do that, right? So suddenly things that were science fictiony become possible. You can do realtime missile tracking, you can do realtime air traffic control. You can do realtime environmental management environmental catastrophe management. You can do maritime work. You can do things that you could never really do before because you had to get the data from the data source down to the ground to do the compute. Then send the results back up. And that time lag was often slow enough, sometimes hours, sometimes days, depending on the size of the data set where you couldn't do anything in real-time Now you can I have an odd question about this one though. So what's going on with air traffic control now? I mean, aren't we relatively safe [00:23:00] as is? Rob: It's going to add to their toolbox. Of things that they can do, right? Air traffic control right now is fairly well mapped out over most of the land, right? The United States and Europe and the Euro Asian continent and, and Africa. But there are still data deserts, right? There are still points where the aircraft is not on any ground-based service. If you're doing oceanic air travel, you're going from Long Island to to Heathrow. There are long stretches of period there where the only information that's coming to you is the plane's own radar system delayed satellite information. This kind of goes away now, so now you've got objects in orbit that can do something called sensor fusion. Right. And sensor fusion is, and you know this from your, your city work as well. Sensor fusion is, okay, I've got a little bit of data here. I've got a little bit of data from this sensor. I've got these images here. I've got this radar data here. I've got lidar here. Cool. Now what? Right. But, but now you've got this ability to take all that [00:24:00] stuff and combine it together and come up with something more accurate, It is taking data and turning it into intelligence, and it just clicked for me right there while you were explaining it. And the first thing I thought of was that missing Malaysian Airlines flight and we didn't know where it was because it wasn't sending out any more pings. Rob: that's right. okay, now I get that. So the final basic question is, this is so cutting edge. What's changed in the last few years that makes this even possible now? ? Rob: Well, advances in computing that has pushed chip technology. Pretty significantly, you know, Nvidia leading the pack. But there, there is a lot of folks that are getting involved in this. So the technology itself has improved the ability of the operating systems to make intelligent decisions on its own, based off of its own data. Right. It has improved. That's, that's what's allowing our operating system to, manage the tiles for itself and launch cost dropping. That's one of the big ones , there there's so much competition. Eric Schmidt just took the CEO helmet relativity, [00:25:00] and there's SpaceX and, Honda just got in the act and the blue origin and there, there's a. Dozens of smaller companies that are, are small lift companies to, to push things up into orbit. And there's so much competition, like anything else on the planet. Competition has caused prices to plummet, all projections show that they're gonna continue to plummet. So it's becoming cheaper now to put things up. It is a good combination of factors, which makes it a very attractive market to go after. But how do you scale this? You mentioned that your tiles are gonna be ready to go next year. Is this something that's gonna be small at first and then grow incrementally? Rob: It is the goal for these things to self deploy even the large ODCs that we're planning on in 20 30, 20 31. Because of the size of the tile and because of the way our origami tech works they can be made quite small. And there are a lot of heavy lift rockets. Coming online, so in, in the 20 30, 20 31 timeframe, we feel confident that the carrying capacity of these, these rockets will carry up an entire one megawatt data center on its own and it'll [00:26:00] deploy in the meantime. As we're putting up these smaller racks of tiles they're small enough where they'll fit in in, in current lift vehicles. Even when we get to the point of something that we're calling a satellite companion, which is not quite a data center and not quite a rack of tiles it it's small enough where it will fit in current launch vehicles. I want to go a little bit more into the use cases. So we talked about air traffic control, and I understand that one. I also understand the defense side of it. But the one I want to go into now is the emergency responders. Is that more for disaster recovery? Rob: Think of it this way. It's a variation of the A TC problem, right? So when you're on the ground, whether you're a foot soldier or you are a, a disaster response person, you have available to you what's coming in over your headset what you see in front of you, what any sensors you have on your body collects and whatever drone information is collected right. A lot of that is disassociated, and a lot of that is the response tactics are provided by people in a [00:27:00] room somewhere that's collecting all this stuff and advising the people on the ground. Now, if you want to employ satellites to, to help with disaster recovery, people looking for heat signatures of people in a building or, you know, something along those lines that information has to get relayed, but if you are in orbit with compute and you send all this information up, which is a lot smaller task than sending the information down because the data sets are small and, and can be, can be sent to orbit fairly quickly. Let the AI that's in those tiles coordinate that effort and then send accurate information back down. Now you've got a system where you know, where somebody is under the rubble or, or you understand where the enemy is over the next horizon, or you understand incoming missiles you've got this situation where you can now coordinate all this stuff in a manner that makes sense , the near term use cases, and by near term, I mean the next five to 10 years sounds like there are a lot more defense, emergency response and air traffic control related. But do you think this is gonna evolve to [00:28:00] where there'll be some commercial applications where enterprises will actually start to leverage this as well? Uh. Rob: A thousand percent. 1000 Let's talk about orbital data centres ODC We don't. Assume at all that ODCs are going to replace terrestrial data centres but right? But they will absolutely augment terrestrial data centres So the conversation is to work out collaborations with the hyperscalers. Right. And we're having those conversations now I've worked in data centers, I've worked around data centers. I've worked with data centers as a client user facing instances in those cloud services, it, it's always problematic. You know, we had a hard time at, at both SOM and Nimble, getting enough instances ready for our clients in the morning because those servers are busy doing other things. And the other things they're doing are LLMs for, for artificial intelligence cryptocurrency, mining rendering, et cetera, et cetera, et cetera. So there's all this backend processing stuff that doesn't need a user facing [00:29:00] component. So the conversation is just give that to us. Things that you don't need for, several hours. Just, just pass that up to us. We'll do that processing for you and we'll send you back the results. And then you can use those servers that you're tying up, doing things that you don't need to be tied up doing for user facing instances so you get more clients in on your end. It's an interesting little addition to what's going on right now. If only there was a company that was really big in data centres that was also involved in space launches. Rob: Yeah. Yeah. If only, So when you look at how this industry is going to develop over the next 10 years, what do you think are going to be the big drivers behind it? Is it gonna be the technology advancements, the prices coming down, or is it gonna be a massive demand for these types of solutions? Rob: nature abhors a vacuum, right? So, the more compute you provide. And the more storage you provide, the more you'll need it, right? The best way to paraphrase it is, we used to have these computers that had like 640 k of memory and, [00:30:00] you know, people were writing compact codes, so it would run in 640 k of memory. Now memory's not a problem. You get computers with 32 gigs, 64 GB regularly being shipped out. Right? But what has happened is not that, oh well, we've got more memory than we need. What has happened is we don't have enough memory. You just, you're on a gaming computer with 64 gigs, then you don't have enough memory. You don't have enough memory because these programmes have expanded to take advantage of all that power that you've suddenly given them. That'll happen here. Right? And so commercial applications. We'll find the way, right? And if, if you suddenly tell them that, look, you can put up this for, for pennies on the dollar, you can put up this sensor in orbit that does a thing that you want to have done or does long-term storage, or is doing, you know, high intensity compute, and we'll give you the computers to do that. You're off to the races, right? And, and by the way, five tiles. So one tile generates we're projecting about 450 tops, of, compute power. Five of those tiles technically meet [00:31:00] the definition of a supercomputer, and because they're networked together and because of the way they work, off you go, right? So, so now you've got this situation where you have unlimited computing. If you need more compute, you just launch more tiles and, and you attach those on and you can get access to it whenever you want. So what do you think is gonna happen I think it's gonna be fantastic, but let, let's look out 10 years and let's say that the industry has completely taken off. What are the critical success factors along the way over this 5 to 10 year period? What needs to happen to make this market really get to the next level? Rob: We gotta get the operating system right. It, it is a project that will never end. We'll always be manipulating it, making sure that it works. 'cause you're not sending up a service tech to repair a tile or to move tiles around or what have you. The operating system on the tiles has to work out of gate, and then as we move forward and we expand to new and different clientele and, and larger and larger collections of tiles, the [00:32:00] operating system has to be able to do the right thing to manage those tiles as though it was a person, right? And so, that's the tech that, that, that really needs to progress in our space. And then on the ground it's keep keep the costs coming down. You know, let's, let's get down to a couple hundred dollars a kilogram, right? As those costs drop dramatically and they are, we can start to do more and more ambitious things. We as a people, right, as a, as a humanity, right? We can start to do more and more ambitious things if we can get into space easier. This has been fascinating and I love the fact that I feel like I actually have a decent understanding now of where this is at. Rob: You're a smart guy, Charles. I've known you for a long time. Yeah, but I'll be honest, I was struggling with what some of the use cases were going to be, and now I've got a better understanding of the defense side the air traffic side as well as the emergency response side. And I also see how when these prices come down, you know how it [00:33:00] actually doesn't replace an existing data center? It augments it. So these things are complimentary. And let's face it, like we said, there's a couple companies who could benefit on both sides of having data centers and space launches. I want to close out by asking a few more fun questions because you've got a breadth of experience over the years, especially on the startup side. So you've gone through this time and time again. What advice do you have for entrepreneurs or someone that's thinking about becoming an entrepreneur? Rob: If you're risk averse, don't do this I don't mean that to be as a negative, but it does take a toll. You're always hustling for your next dollar and you're always hustling to make your, your minimum viable product. And you're always hustling to, to, to catch up with. The bigger players have been around for a while, so you have to be internally prepared for that. The positive side of that, if you wanna call it a positive side, is it becomes very addictive. So when I left government work and I moved into entrepreneurial work, first of all, it was a culture shock. Once you got into the groove, you really [00:34:00] can't stop it. It really feels like, you know, it it like eating another peanut and, and, you know, I was semi-retired, I was out and they just, when you're out, they reach and they pull you back in. It tickles a part of the brain that maybe you didn't know that you had. Right. So the intellectual curiosity, the business side of it the maneuverability of it the knowing when to hit the gas and when to quit, right? All of that stuff that is dormant in your lizard, hind brain comes up. So the best advice I have for people are doing this stuff is if you want to do this, you should do this. And if you have any hesitation at all, the world is a big place. There's a, there's a lot of very interesting things that you could do. The next question is how you keep yourself motivated during this process. You mentioned you're going to be doing some demos in 2026. You're not really gonna deploy until 2027, but you're looking at some longer development cycles. So along this journey, what keeps you motivated day to day? You know, is it the intellectual [00:35:00] challenge of it? Rob: . That, that's part of it. It, it's also shorter than you think. We're talking about putting up a data center in, you know, six years. Right. That's a short timeframe. And when I worked at JPL the first time around, mission lifecycles were 20 plus years, right? Like I said, I was the image processing specialist on Galileo. I never saw those images delivered. I only worked on test data because I, I left before Galileo completed its mission. People stay at those positions for years, right? And rewardingly. So, but the, the concept of going from a 20 year project cycle to a few years is incredibly compressed. And that in and of itself is the intellectual challenge for me personally. It is good to keep things in perspective. I mean, five, six years sounds like a long time, but I think five years ago we were in the middle of a pandemic. The final question is what should we be looking to see from Sophia space in the next 12 to 18 months? Rob: Demos, so in orbit demos next year and [00:36:00] then activating one of our one of our collaborations for 2027. So we'll be putting up tiles tile satellites that are doing the work that is required by our collaborator. . Rob: So you'll see that and then near the end of 2027, you're gonna see us doing production facilities. The interesting thing to keep in mind here, and the thing that separates us from our, our brethren at, at JPL is that we're not building a special purpose device. We're Dell computer in space, so basically we're, we're building computers. And you and I have a good friend that, built a business in, in in data centres in Asia. And I was talking to him about this and he's like, you can't be in a better business than making the machine itself. Right. And so we WE'LL be rolling these things off an assembly line, the same way that you roll out servers off an assembly line and being sold to clients giving them all the tools to make it work the way that they need it to work. Here's your computers. Go crazy. So we're not building [00:37:00] special purpose devices. We're not building something to look for life on, on your robo. We're not building something to wander the surface of Mars. We're not building a thing to do a thing. We're building computers. And so once you start hearing that we are selling these tiles which will be in the late 2027, early 2028 timeframe, that's when you know things are changing. And the second part of the final question is, what about you personally? What are you most looking forward to in the next 12 to 18 months? Rob: Oh, oh yeah. Lighting these things up. . Rob: , This business as well as every other business I've ever been in because I'm always on that edge. Looking over there are the naysayers. You can't do this for blah blah blah we know of course we're not idiots. We, know, we know about all that stuff. And getting to that point of going, turning this thing on. Getting the first blip that's going to be satisfying not only from a business and commercial milestone and an intellectual and [00:38:00] commercial milestone, but from a personal Bahaha told you milestone. So, so all that stuff I'm looking extremely forward to I love it. The whole thing you're most looking forward to is the, I told you so moment, and I don't blame you because when you're at the forefront of technology and trying to change people's perspective or preconceived notions on a topic, the best thing at the end is like, I knew it was gonna work all along. Rob: It's the best, it's the best feeling, and, and it, it's happened several times in my career and it's great. So, Rob, thanks so much for taking time out of your busy schedule to talk to us today. This has been absolutely fascinating and thanks for being so patient with my very naive questions because I'm not an expert. Rob: Oh, please, CRA: now I feel like I'm, I think I. Rob: no, no, no, no, no, no, no, no, no, no, no, no. And neither you nor your listeners or. 90% of the people out there. It's, it's, once you grasp that, it's, it is a computer in space. It's all, it all makes complete sense. Rob, I wish you all the best and hopefully you'll keep us updated on your [00:39:00] journey and the progress, and hopefully we'll catch up for a drink pretty soon. Rob: That would be tremendous. Looking forward to it, Charles. Cheers. Bye.