Utopia Talk / Politics / Google building data centers in space

Pillz rank	Tue Dec 02 18:41:28 https://ww...nters-space-solar-2027-2025-11 Sundar Pichai says Google will start building data centers in space, powered by the sun, in 2027 Google CEO Sundar Pichai Google CEO Sundar Pichai said the company's next moonshot is space data centers. Justin Sullivan/Getty Dec 1, 2025, 5:33 AM GMT-5 Google unveiled Project Suncatcher earlier this month. It aims to reduce AI's environmental impact by relocating data centers in space, powered by the sun. Google CEO Sundar Pichai said the company plans to begin sending 'machines' to space next year. The great AI space race has begun. Google has been quietly working on a long-term research initiative, internally known as Project Suncatcher, to "one day scale machine learning in space." Google CEO Sundar Pichai told Shannon Bream on Fox News Sunday that Google's goal is to start putting data centers in space, powered by the sun, as soon as 2027. Related video Inside the secretive world of America's AI data centers "We are taking our first step in '27," he said. "We'll send tiny, tiny racks of machines, and have them in satellites, test them out, and then start scaling from there." In a decade, Pichai said that it'll be normal to build extraterrestrial data centers. "At Google, we're always proud of taking moonshots," he said. "One of our moonshots is: How do we one day have data centers in space so that we can better harness the energy from the sun, which is one hundred trillion times more energy than we produce in all of Earth today." Google's cosmic pivot comes amid growing global scrutiny over the power demands of data centers. "There is still much we don't know about the environmental impact of AI, but some of the data we do have is concerning," Sally Radwan, the chief digital officer of the United Nations Environment Program, said in a press release in November. "We need to make sure the net effect of AI on the planet is positive before we deploy the technology at scale." The UN says AI's toll on the environment stems from the extraction of rare materials and minerals needed to build the technology and microchips to power the technology, the massive amounts of electronic waste data centers produce, the water needed to operate and cool data centers, and the greenhouse gases produced by operating data centers. Google's plan is to divert some of that environmental toll off the planet. "In 2027, hopefully we'll have a TPU somewhere in space," he said, referring to the company's custom AI chip, on the "Google AI: Release Notes" podcast last week. Google did not immediately respond to a request for comment from Business Insider.
Sam Adams rank	Tue Dec 02 19:13:11 Seems expensive
Pillz rank	Tue Dec 02 19:17:03 Yes, I m sure murder is gonna hate how much money SpaceX makes
TheChildren rank	Tue Dec 02 19:20:02 i call bs
Paramount rank	Tue Dec 02 19:33:21 Yeah. The US is yet to build proper platform barriers in their subway stations. A proper wall along the border. High speed trains. Etc.
Average Ameriacn rank	Tue Dec 02 19:52:55 MAGA Would never happen under a Kamala presidency! But Trump makes it possible! US Space Force created by Trump! MAGA!
obaminated rank	Tue Dec 02 22:27:38 Good for them.
Seb rank	Tue Dec 02 22:59:57 Seen this idea a few times. It seems outlandish to me. Heat management is a total bitch in space - that and the shit duty cycle would seem to impose more costs than just building something on the ground.
murder rank	Wed Dec 03 00:06:33 "Yes, I m sure murder is gonna hate how much money SpaceX makes" I'm sure Elon will make a killing launching Google's tiny tiny racks. -
Sam Adams rank	Wed Dec 03 05:34:03 ”High speed trains.” Lol trains were cool once. Before airplanes that is. Anyway I hate to agree with seb but ya your trading 4x cheaper solar panels for insane launch costs, terrible cooling, a total inability to repair. There's obviously no direct cost advantage so the only reason to do it is 1) marketing 2) you wanna do shit that earth governments can't interfere with.
Seb rank	Wed Dec 03 13:59:34 Sam: I don't think having the servers in space helps. They'd be deemed as under the control of either the state from where launched, where they are conducting business or where the owning co is based. The last time this came around it was supposedly about latency. Combining the idea that you are never 250km from the lowest regulatory jurisdiction might, might, for some use cases work. But I suspect that will just lead to countries adjusting their regulatory approach. The US or EU isn't going to let a satellite beam child porn into their territory from a satellite and shrug and say "of dear". They'll want access blocked locally and they'll find ways to do that. Is it 4x? Half the time the servers will be in night or depending on orbit not near the demand.
Seb rank	Wed Dec 03 14:01:34 I think people like this idea because it soibds cool, bluntly. Or as green washing. Like data-centre that have got planning permission on the provision of being "green" and have "contracts" for nuclear power in the distant future, but are actually running on gas turbines "on an interim basis" etc.
Sam Adams rank	Wed Dec 03 17:48:13 "Is it 4x? Half the time the servers will be in night or depending on orbit not near the demand. ” True... if the servers are in LEO it's more like 2x.
murder rank	Wed Dec 03 20:50:53 Wouldn't they just power those things with nuclear reactors? -
Sam Adams rank	Wed Dec 03 21:03:15 Nuke plants generate lots of waste heat.
Seb rank	Wed Dec 03 23:31:35 https://ta...errible-horrible-no-good-idea/ Good deconstruction of the whole idea. Murder: I don't think anyone's actually designed a proper nuclear reactor for use in space. RTGs won't cut it. All sorts of potential issues there, not least the sheer weight of the thing. Sub reactors are something like 1000tonnes. Actually assembling a reactor in orbit, difficult. And zero G buggering up cooling and the lack of a heat sink are all massive problems
Pillz rank	Thu Dec 04 00:18:32 Hey, I'm not gonna bother looking this up, so Seborsam can answer it. If we use water to cool reactors on Earth, how the fuck do we plan on cooling reactors in space?
Seb rank	Thu Dec 04 01:40:15 Giant radiators. Whatever you do for the primary and secondary cooling loops (convection doesn't work in space, so you'll need to get very confident you can get decent circulation) which are closed, you can't dump the secondary loop through a heat exchanger into a lake, river, or cooling tower. You run the secondary it through a bunch of pipes to a giant panel that's as close to a black body as you can get facing into deep space and let it radiate the energy away as infra red. It has to be pointing into space, if it's in sunlight, it will get very hot (hundreds of degrees) just from sunlight and be far less effective. Even if it's pointing at earth, it'll get to around 1 degree centigrade eventually just from light and heat the earth radiates. These are incredibly inefficient, so you need a huge area. The largest one in space is on the ISS, is about 50sq m and dissipates about 15kw. A high end GPU plus server and electrical supply probably puts out around 1kw of heat so that's likely your limiting factor here. Basically something the size of the ISS couldn't even run a rack of GPUs, let alone an actual data centre. As for a 20MWTh reactor (low end of a nuclear sub), about 40% of that is waste heat straight away so needs a radiator array 500 times bigger than the ISS ones, and of course all the rest of that energy in data centre scenario ends up as heat also which likely needs to be radiated. The one thing that might help is the waste heat from a nuke will be high temperature and radiative power goes as the 4th power of temperature. So maybe you can design a radiator that can operate at really high temperature with really high temperature coolant but that's likely to get very heavy. Basically running an actual nuclear reactor in a spaceship is pretty hard. Much easier if it's stuck to something big like the moon or an asteroid you can dissipate heat into bits diffusion. Even better if it has an atmosphere and then you use convection. Radiation is just very very shit at moving heat around and that, basically, is why thermoses work.
Seb rank	Thu Dec 04 01:45:06 If you want to write hard sci-fi and you want a nuclear reactor or whatever powered starship with a human crew, you need to think about things like venting coolant into space, big ass radiators, or some complicated hybrid like liquid radiators (a fine spray of viscous liquid between two masts, one with a nosels one that collects and recycles).
Sam Adams rank	Thu Dec 04 08:46:53 Can't vent coolant seb, the mass budget doesn't allow for it on a true hard scifi starship. Ok maybe you can vent coolant for the first little bit of delta v but that's the easy part so who cares. "how the fuck do we plan on cooling reactors in space?” Well that's the real trick isn't it. You don't have much choice other than big radiators. Which of course adds more mass which means you need more power which means even more radiators... Etc etc Those exponential bits of the rocket equations are a real bitch.
Sam Adams rank	Thu Dec 04 08:53:53 Basically to make a real actual starship you need to move around a butt load of plasma with nothing but magnetic fields. This is your fusion core AND your radiators. Preferably you control the moving plasma in such a way that it creates even more magnetic field in such a way that it can move more plasma. Etc etc. Good luck controlling that. Basically it would be a vastly complex set of differential equations balancing your starship on a knifes edge between atomizing itself with the slightest deviation either way.
Pillz rank	Thu Dec 04 10:35:59 So this is just a PR stunt by Google that's gonna be forgotten about until they remind everyone by canceling the project
murder rank	Thu Dec 04 16:31:27 It's not just Google. Bezos and others have pushed the same idea. It's likely a response to the data center backlash all around the US. They are probably trying to tamp that down by pretending that it's only a temporary problem ... with the data centers migrating to orbit in just a few years. -
Seb rank	Sat Dec 06 02:43:28 Sam: Depends. I was thinking of something where a mission might be a very short decisive engagement (not impossible: lasers and it's all over) requiring a relatively short burst of high power. You use radiators outside of combat, but keep them folded for engagement. Can also imagine something like a nuclear salt water rocket. Coolant doubles as reaction mass. Murder: Yup. Like when they were promising they'd be nuclear powered 18 months ago.
Sam Adams rank	Sat Dec 06 19:57:19 Agreed, combat is a place where it's justified to dump coolant. To give a brief speed boost and or protect radiators. But that kind of combat is more in realm of interplanetary cruisers rather than true starships. Starships can irradiate planets for fun so combat for them is an entirely different ballpark. Here's a fun question... What's the maximum effective combat range of a laser in space combat.
Seb rank	Sun Dec 07 15:00:59 Sam: I'm thinking more like the expanse. Hard Sci-Fi on an interstellar scale is very very boring in between the interplanetary bits unless you make up an FTL mcguffin; but it's actually very hard to do that without breaking causality. The expanse does it with gates, but that just makes everywhere effectively interplanetary space. "What's the maximum effective combat range of a laser in space combat." Depends on line-width of the laser source and aperture size and how steady you can point the laser. I don't know the current state of the art. Eventually, potentially some factor involving speed of light, maximum acceleration of the target craft. But at the moment very much the former than the latter). I'm sure atomic rocket has a whole page on this.
Sam Adams rank	Sun Dec 07 21:33:40 ”Eventually, potentially some factor involving speed of light, maximum acceleration of the target craft." Ya that's what I was going for. That happens pretty quick on a fast target. Small fractions of a light second.
Sam Adams rank	Sun Dec 07 21:44:23 And I've seen a decent amount of hard interstellar scifi. You gotta use life extenders or artificial sleep or switch generations(like foundation... Not that that was hard scifi of course but the generations bit). The technology to pause or extend consciousness should be utterly trivial to a proper starship so no worries there from a science point of view.
Seb rank	Sun Dec 07 22:07:52 Sam: IIRC at the moment plausible state of the art apertures and line width and current brightness of sources put ranges around the 100's of km in vacuum (not that they are used in vacuum, so it's more like 10's of km at sea level) - but I'm sure we can improve on that significantly if we were actually trying to designing for long range in vacuum using current technology. Sure, but that's what I mean about the interstellar bit being mostly very boring. All the story and action needs to happen in interplanetary space and in between your ship is cruising with a frozen crew and probably not terribly easy to identify, track or engage. Star trek style interceptions and combat in interstellar space seem outside of plausible known physics.
Sam Adams rank	Sun Dec 07 22:48:01 "Sure, but that's what I mean about the interstellar bit being mostly very boring.” Well ya you'd skip that bit. Like writers already skip the legionary walking 1000 miles or the sailor taking months to cross oceans.
	show deleted posts

Your Name:
Your Password:
Your Message: