Transcript 0:00 [upbeat music] Welcome to the Viro Podcast. Every week, we dig into what's really happening where climate crashes into culture, politics, and the world you actually live in. The stuff that matters, told straight. 0:17 This is Viro. Let's go. Hey. Something different this week. No co-host, no back and forth. Just me, one story, and about twenty minutes of your time. 0:33 We're going deep on data centers in space. And I know how that sounds. I know. But stay with me, because the money is already moving. The rockets are already being built. 0:46 And the people making this bet include the world's first trillionaire, the founder of Amazon and Google. This is not science fiction anymore. 0:55 This is the next fight over who controls AI infrastructure, and it's happening right now above our heads, whether we're paying attention or not. So, what are orbital data centers? How do they actually work? 1:12 Who's building them? How are they powered? And is this genuinely a climate solution or just a very expensive way to create new problems in orbit? That's what we're getting into today. 1:25 No headlines, no bullets, just the full picture. And one quick thing before we dive in. Every time you use Viro AI, you're funding renewable energy and climate restoration. No extra steps, no donations, no guilt. 1:43 Just use it the way you'd use any AI, and something real happens in the world. Link in the show notes. Okay, let's go to space. [upbeat music] A quick note for any brands listening. Wanna showcase your brand with us? 1:59 Partner with the Viro Podcast, and your marketing dollars go directly toward funding renewable energy and climate projects. Simple as that. Link in the show notes. One trillion dollars. Not a billion, a trillion. 2:17 A thousand billions. And I want you to actually sit with that for a second because our brains are not built to process it. A million seconds is eleven days. A billion seconds is thirty-one years. 2:33 A trillion seconds is thirty-one thousand years. Last week, Elon Musk became the first human being in history to have a net worth of one trillion dollars. 2:47 It happened when SpaceX went public, the largest IPO ever, valuing the company at one point seven seven trillion dollars. Musk owns about forty percent of it. And here's the number that really got me. 3:02 The second richest person on Earth right now is Larry Page, Google co-founder, worth two hundred and ninety-one billion dollars. 3:12 Which sounds like a lot until you realize Musk could write Page a check for his entire net worth, his whole fortune, everything he has ever built, and still have over seven hundred billion dollars left. 3:27 Elon Musk could lose a trillion dollars and still be the richest human being alive by a comfortable margin. We genuinely do not have a cultural reference point for this. 3:41 The closest thing I can think of is a medieval king, someone who owned the land, the roads, the ports, and the army. Except Musk also owns the rockets and the satellites and the AI company and the social network. 3:58 So what does the world's first trillionaire plan to do with all of it? He wants to put data centers in space. And not, like, a few. 4:08 SpaceX filed an application with the FCC in January to launch up to one million satellites as orbital data centers. A million. And he is not alone. Jeff Bezos' Blue Origin filed for fifty thousand. 4:24 Google announced a prototype constellation. Startups are already building hardware. This is the episode where we figure out whether that's genius or the most expensive way ever devised to create new problems in orbit. 4:40 Let's start from the beginning. What even is a data center? At its core, a data center is just a building full of servers, computers. Those servers process information. They train AI models. They run cloud services. 4:56 They handle your Netflix stream, store your photos, power every app on your phone. The Internet, essentially, lives in data centers. And they have two massive problems: electricity and heat. 5:12 A single large AI data center can draw as much power as a small city. We're talking gigawatts. And all those chips generate enormous amounts of heat. 5:24 So you need cooling systems running constantly just to keep the hardware from destroying itself. Cooling alone accounts for thirty to forty percent of a data center's total energy use, 5:37 and most of that cooling uses water, billions of gallons of it annually across the industry. Communities in Virginia, Arizona, Texas, 5:48 they've started fighting data center construction because the power draw is too high, the water use is too high, and the local grid simply cannot handle it. 5:58 Some counties have started requiring water use disclosures just to get a building permit. So the pitch for space is: what if both of those problems just go away? 6:10 In low Earth orbit, solar panels receive about thirteen hundred and sixty-one watts of sunlight per square meter. No clouds, no atmosphere to filter it. 6:21 And if you're in what's called a dawn-to-dusk Sun-synchronous orbit, basically always sitting on the boundary between day and night on Earth, you get near-continuous sunlight. The energy is just there, free, permanent. 6:39 No power grid required. And cooling. Here's where it gets interesting. In space, there's no air, so you can't use fans or cooling towers. But here's the thing. 6:51 Hot surfaces radiate heat as infrared energy into the vacuum, passively, indefinitely. No water, no chillers, no cooling towers. The vacuum of space is, in one sense, the most efficient heat sink in the Solar System. 7:10 So on paper, unlimited solar power, passive cooling, no land permits, no community opposition, no water, no grid. That actually sounds pretty good. But there are several catches. Let's go through them. 7:28 First one, heat is harder than it sounds. I know I just said space is a great heat sink, and it is, but getting enough heat out fast enough is the problem. 7:41 Getting ten megawatts of waste heat out of a satellite requires radiator surfaces comparable in size to two football fields. Two football fields on a satellite. 7:55 The chips generate heat faster than the vacuum can passively absorb it at industrial scale. So you need massive physical radiator structures just to dump that energy. 8:06 And you have to build those radiators, launch them, assemble them in orbit on hardware that has to survive the journey. Second catch: radiation. On Earth, the atmosphere acts as a shield. 8:21 It filters out cosmic rays, solar radiation, high-energy particles. 8:26 In orbit, that shield is gone, and radiation causes something called bit flipping, random errors in computing hardware, where a cosmic ray hits a transistor and flips a zero to a one or vice versa. 8:40 On Earth, this is rare enough that error correction handles it fine. In orbit, it's a constant background problem. 8:48 Every processor either needs to be radiation hardened, which makes it heavier, more expensive and slower, or it's running sophisticated error correction software that adds compute overhead and complexity. 9:02 And a solar flare, a big one, can disable an entire constellation all at once. And the third catch is a bit cheeky. You can't call the repair guy. Data center hardware fails constantly. 9:19 On Earth, a technician drives over, swaps the broken server. Done. In orbit, repairs are extraordinarily costly and difficult. 9:29 Hardware that fails stays failed until the satellite is deorbited, which means either you build massive redundancy into every satellite, expensive, or you accept that your constellation is gradually degrading over time. 9:45 Fourth, latency. Now, for AI training, this actually doesn't matter much. Training is a batch workload. You send the data up, the satellite processes it, sends results back. The latency hit is fine. 10:00 But AI inference, actually using a model in real time, like a voice assistant or an agent making decisions, is latency sensitive. The round trip distance from low Earth orbit adds meaningful milliseconds to every query. 10:17 For real-time AI applications, talking to a server in orbit is meaningfully slower than talking to one in a data center forty miles away. And fifth, Kessler Syndrome. 10:30 This one is not a joke, and I want to make sure it lands. 10:33 If SpaceX launches a million satellites, and Blue Origin launches fifty thousand, and Google launches its constellation, and startups fill in the gaps, the density of objects in low Earth orbit becomes a collision risk that compounds on itself. 10:52 One collision creates debris. Debris creates more collisions. Each collision multiplies the debris field. 11:02 The cascade scenario, called Kessler Syndrome after the NASA scientist who modeled it, could render certain orbital altitudes genuinely unusable for decades, maybe longer. And here's the thing. 11:17 Even SpaceX's own IPO prospectus warned investors that orbital data centers, quote, "May not be commercially viable." They buried it in the risk disclosures, but they wrote it down. So where does that leave us? 11:33 Orbital data centers are a real idea with a real engineering foundation, backed by real money and real rockets that faces real problems that have not yet been solved at scale. 11:46 Whether they become the dominant AI infrastructure of the twenty-thirties or a cautionary tale about Silicon Valley's relationship with scale depends almost entirely on engineering execution and on whether the economics actually work out the way the prospectus says they will. 12:05 And here's the thing that gets missed in almost every piece of coverage on this. The reason this race is happening is because AI demand has outpaced the Earth's ability to supply the power to run it. 12:18 That's the actual problem, and the space data center pitch is a supply-side solution to a demand-side problem. Supply-side solutions to demand-side problems in tech have a track record. 12:32 More efficient AI servers didn't reduce AI energy consumption. They made AI more accessible, which increased total usage, which increased total energy demand. The same dynamic could easily play out in orbit. 12:47 Cheaper AI compute drives more AI usage, drives more satellites, drives more launches, drives more rocket fuel burned in the atmosphere. So is there a better answer? 13:01 I think the honest answer is the only actual solution to AI's energy problem isn't to find a cleverer way to power the demand. 13:11 It's to make sure that however AI is powered, on Earth or in orbit, the energy is clean, the environmental costs are fully accounted for, and the people bearing those costs have a voice in the decision. 13:26 We don't have a starship. We're not launching satellites. But at ViroAI, every query you send funds renewable energy and climate restoration on the ground right now with infrastructure that already exists. 13:41 Not in three years, not after the engineering problems are solved. Today. The future of AI might end up in orbit. In the meantime, we'll keep funding climate solutions from down here on Earth. 13:57 Thanks for going deep on this one with me. Let me know if you want more episodes like this: solo, one subject, full picture. See you next week. [upbeat music]