Posted on 09/25/2026 3:45:14 AM PDT by Libloather
Elon Musk is making a sky-high prediction for artificial intelligence. The SpaceX CEO says computing will eventually move almost entirely off Earth.
“The amount of compute in space will obviously round up to 100% of all compute,” Musk wrote on X in response to a video about Google’s Project Suncatcher.
The project explores whether solar-powered satellites carrying Google’s AI chips could work together as an orbital data center. In the video Musk responded to, researchers describe satellites flying in tight formation and linking their chips to handle larger computing tasks.
The idea is still in the testing stage. Google has said it plans to launch two prototype satellites with partner Planet by early 2027 to test the hardware in orbit. The X post sharing the video claimed Google would launch chips on a SpaceX Falcon 9 next week, but Google’s published plan does not confirm that timeline. Google says the satellites could tap into near-constant sunlight to power AI chips. Its first planned test involves just two prototypes, leaving a long way to go before Musk’s vision of moving nearly all computing into orbit could become reality.
(Excerpt) Read more at foxnews.com ...
That means datacenters on earth are already obsolete.
And when Starship goes orbital it will take 60+ of the V3 birds up that make V2s look like dial-up modems the new V3 DC will send 5G
—
They could be usable in a matter of weeks after disbursal.
“Yes radiation is the only way and it’s very effective the hotter the radiator”
Semiconductors don’t get very hot. I’m not saying it can’t be done at all, but I need to hear why it’s better to put datacenters in space than on earth.
Recall the datacenter they’re building in West Virginia. It will use as much energy as Los Angeles. Actually, generating that much energy with photocells doesn’t sound like a great idea either. Use nuclear for that? That also generates a lot of heat that needs to be dissipated.
Going to space is rather obvious. The power will be “free” and the cooling will be “free”. The only big computing done on earth will be time sensitive, where it is imperative there is no time delay that you have shooting to and from space.
Not a chance. The lift capacity to get the literal multiple square km of solar panels into orbit is prohibitive. Then there’s the prohibitive expense of maintenance and repair, as well as the cooling problem. Not gonna happen.
Villa Straylight.
I’ve speculated before that I think Musk is following the example of the fictional Tessier-Ashpool family…
… but also it makes sense. Abundant solar for power, easy cooling, lots of space. Just gotta get all the parts up the gravity well, which is why SpaceX went public.
You may well be a “GenX Polymath” but you aren’t conversant in military technologies.
WHERE WILL THE “COOLING WATER” COME FROM?
Maybe in a future world with no conflict? Lots of luck with that.
Elon Musk has a long track record of making bold predictions that don’t pan out, some bold enough to constitute misleading investor information and securities fraud.
Believe his extravagant claims at your own risk.
I know enough to know that I am not an expert.
I do know that people are serious about this, even though I cannot understand why.
I had a long conversation with the google bot about this, and still don’t understand the cost/benefit ratio.
even on earth I have trouble with random bit flipping even with Ecc Rdimm
I knew many rocket scientists and satellite designers back in the day, including the designer of Teledesic if anyone remembers that, and a bunch of computer scientists from Dec too.
I still cannot see this being useful for anyone except a Govt entity or for HFT.
However I know that people are going to do this anyway, I heard one of our new hires from space-x talking about it.
Not sure what kind of chips they will be using, I suspect Arm, one of my friends is a voting member on RISC-V International’s Technical Steering Committee. I will ask him what he thinks next time I see him.
And people wondered why we wouldn’t just be able to pull the plug on skynet.
AI data centers have two problems: Energy and Cooling.
Cooling is taken care of in space. It’s perfectly cold up there.
Energy is a problem of solar arrays. I don’t know how large and efficient they can be at scale in space, and whether the lift to get them there is economical. So the jury’s out on that one … to me.
I hold a Geointelligence cert at the post grad level from the NGA and a clearance to match it.
Plus a master’s in remote sensing one of 3 master’s plus a PhD and double post docs.
I only say what is allowed in open space.
Again Elon is backed by teams of people with similar pedigree.
The FCC just took up his 100000 sat request they will grant it with a rubber stamp.
When not if Starbase LA is at its full capacity it can do 30 flights per day. That works out to 600,000+ V3 sized birds at 2000kg each per year.
No one anywhere can shoot down numbers like that. You couldn’t shoot down the 10000 up right now even if you expended every orbital and sub orbital booster stack , let alone the kill vehicle manufacturing rate which is dozens per year not 10K. Every starship will lob 60 into orbit and at 30 per day yeah you can put up a 100k in a couple months.
Space is big too really big.
You can stack 1 million VELO birds from 300-450 km in five shells and each bird will have over 500 square km of area around it and 30km between them.
At mid latitudes you would see 15000 birds in a 40 degree come over your head. Each V3 XAi based bird is a 3 PETABYTE really and 150kw GPU stack.
Again fundamentally changed.
FR is full of people who don’t realise Elon is the general directing an Army of master’s and PhD level engineers, mathematicians and scientists. He just does every one else plays catch up.
The physics is rock solid a V3 bird does 150kw with arrays not much larger than the V3 starlink birds and radiators edge on seeing the absolute zero cold of deep space 24/7 in fact the radiator area is 1/3 the solar panel size or really close to it. Why again space is bloody cold and they plan to reject heat at 140F so 600w/sqm double sided. When the birds go into eclipse the struggle is keeping heat IN not dumping it minus 250F nearly instantly as you cross the terminator. Yeah he got this can’t wait to see the daily starships let fly.
Monday they put up the first V3s and this kicks off the FCC licence for them 30,000 worth then 100K then a million or more.
You can stack 10 million or more at 30km spaced shells from 400km to 500km the higher you go the larger the area of each shell.
SpaceX already solved the automated traffic control AI with V2 starlinks they maneuver on their own and do avoidance via AI too.
Very impressive pedigree- but how well will those thousands of satellites do against a few nukes in space. Hard gamma tends to fry things.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.