Musk's Plan to Put AI in Space

Kind: captions Language: en You can mark my words. Uh in in 36 months but probably closer to 30 months the the most economically compelling place to put AI will be space. >> That was February 2026. So, by about 2029. Why would AI be cheaper in space? And what has to work for him to be right? >> The availability of energy is the issue. The output of chips is growing um pretty much exponentially, but the uh power of electricity is flat. >> [music] >> So, where how are you going to turn them chips on? >> Um you know, I I >> Magical power sources? Magical electricity fairy? >> His long-term target is a terawatt. >> [music] >> All of the United States uh currently uh uses only half a terawatt of power on average. So, you know, if you say a terawatt, that would be twice as much electricity as the United States currently consumes. >> On the ground, he says, that means power plants, permits, and delays. His shortcut, go where the sun almost never sets. >> You you you're going to get about five times the um effectiveness of solar panels in space versus the ground. And you don't need batteries. Um I almost wore my other shirt, which says it's always sunny in space. Which it is. So, um because you don't have a day-night cycle or uh seasonality, uh clouds, uh or or an atmosphere in space. >> In March, he showed a rough early design for a small AI satellite. >> And then you can see that just the that's just a rough approximation of the the AI the mini version of the AI sat. So, that's roughly 100 kilowatts. Uh it's showing the uh solar panels and the radiator to scale. >> Almost every watt the AI chips use turns into heat. Follow it to the radiator. >> So, uh for some reason there's been a bizarre debate about radiators in space. Uh it's safe to say SpaceX knows how to do heat rejection in space with 10,000 satellites in orbit. Might know a thing or two. Um so you can see the radiator is actually quite small relative to the solar the solar panels. >> So why the debate? The physics. Nvidia's Jensen Huang, also in March. >> We have We're working with our partners on a new computer called Vera Rubin Space One and it's going to go out to space and start data centers out out in space. Now, of course, in space there's no conduction, there's no convection, there's just radiation. And so we have to figure out how to um uh cool these systems uh out in space, but we've got lots of great engineers working on it. >> 100 kilowatts is more heat than the International Space Station's main external cooling system was built to handle. So how big must the radiator be? In June, Musk gave SpaceX's own numbers. >> The assumptions here are uh 250 W per square meter for the solar array and um about 1,400 W per square meter for the radiator. So the radiators These are double-sided radiators are radiating both sides. They're uh oriented knife edge to the sun and uh and and it's 1,400 W per square meter is a very achievable goal. >> And how would you design a chip for space? Musk's answer? >> Well, I guess you want to have it designed to be um more radiation tolerant and run at a higher temperature. Uh so you could um you know, roughly if you increase the um operating temperature by 20% in degrees Kelvin, you can cut your radiator mass in half. Um so look right running at a higher temperature is is helpful in in space. >> Next problem, getting all of that hardware up there. >> So, in order to get to the terawatt uh of computer per year, uh you need about 10 million tons to orbit per year, and at 100 kW per ton. >> In June, Musk put SpaceX at about 2,500 tons a year, roughly 90% of the mass sent to orbit. Starship is designed to close that gap, 100 tons of flight, fully reusable. >> So, I would expect to be at least >> [music] >> sort of 5 years from now, a few hundred gigawatts per year of AI in space. Uh and and rising. >> Even 100 gigawatts would take some 10,000 Starship launches a year. That's about one every hour. >> Yeah, I mean, that's actually a low rate compared to airlines. Uh like like aircraft aircraft >> So, when? Here's Musk's target from June. >> This So, this is not a this is not a promise what we'll do. This is what we what what what we are going to try to do, and think we probably can do. Um which is to get to roughly uh an annualized rate of a gigawatt per year by the end of next year uh in terms of space uh AI compute. Um and then aspirationally scale that by an order of magnitude per year. >> Not everyone is convinced. OpenAI's Sam Altman. >> I honestly think the idea with the current landscape of putting data centers in space is ridiculous. It will make sense someday. >> [applause] >> But if you just do like the very rough math of launch costs relative to the cost of power we can do on Earth, to say nothing of how you're going to fix a broken GPU in space, and they do break a lot still, unfortunately. Um we are not there yet. >> So, could AI really get cheaper in space? Musk says yes, if it all works. Near constant sunlight, hotter radiators, and rockets [music] that fly like airliners. Whether all of that beats Earth on cost still has to be shown. And SpaceX lists one more limit itself, chips. >> But the the current industry it's it seems like it's going to get to maybe around 100 gigawatts a year of of AI compute. But that doesn't answer the question of well, how do you get to a terawatt? That's why you need the terrafab. >> That's terrafab, a chip factory planned by SpaceX and Tesla. Can they build it? That's our next story.