SpaceX wants to fill Earth orbit with a million datacenter satellites
- Reference: 1770314151
- News link: https://www.theregister.co.uk/2026/02/05/spacex_1m_satellite_datacenter/
- Source link:
SpaceX [1]applied for permission to launch as many as one million satellites for its proposed orbital datacenter constellation on January 30, and the FCC's Space Bureau has already [2]accepted [PDF] the application for filing just five days later, though perhaps just so it can collect enough public comments to shoot the ridiculous idea down before it gets beyond the Muskian brainstorm stage (we inquired, but didn't hear back).
According to Harvard astrophysicist and space object cataloger Jonathan McDowell, there are a paltry 14,518 active payloads in Earth orbit as of the end of January, 9,555 of them belonging to Starlink. Earth's orbit is [3]already crowded with fewer than 15,000 active satellites circling the planet, and tens of thousands more tracked objects when defunct spacecraft and debris are included, while large satellite constellations are [4]increasingly interfering with astronomical observations from the ground and [5]from space .
[6]
Increase the number of active satellites by nearly 6,800 percent and the risk of Kessler Syndrome - a situation where Earth orbit becomes so crowded a chain reaction of collisions creates so much space debris that portions of Earth orbit become unusable - starts to look quaint.
[7]
[8]
"One million satellites are going to be a big challenge for astronomy, especially as they are in higher orbits which is worse for us," McDowell told The Register by email when asked about Musk's orbital datacenter plan. SpaceX's [9]application [PDF] to the FCC calls for its datacenter constellation to be in multiple orbital shells in altitudes between 500 km (310 miles) and 2,000 km (1,242 miles).
"I think a constellation like this will absolutely be required to have a fleet of tow-truck satellites to remove failed ones to avoid Kessler," McDowell added. Companies like Astroscale are developing such satellites, McDowell said. He also noted that the Chinese have [10]demonstrated satellites with such capabilities.
[11]
As of last October, Starlink was estimated to be deorbiting and burning up roughly one to two satellites a day, which McDowell told us at the time could be a [12]serious problem for the planet's health. Those concerns have yet to be resolved, McDowell told us in our most recent conversation.
Is Musk's mega datacenter constellation even feasible?
Musk said in a Monday blog [13]post announcing SpaceX's acquisition of xAI (and, by extension, [14]financially-troubled post-Twitter social media platform X) that the 1M satellite orbital datacenter constellation is part of his bizarre dream to scale SpaceX to "make a sentient sun to understand the Universe," and while that sci-fi dream didn't make it into the FCC filings, one of Musk's other delusions did.
"Launching a constellation of a million satellites that operate as orbital data centers is a first step towards becoming a Kardashev II-level civilization," SpaceX said in its filing.
Kardashev II-level civilizations are able to harness the full power of their star, often theorized to be achieved by a Dyson sphere that completely envelopes a star at a habitable orbit.
In order to become a Kardashev Type I–level society, Earth would need to harness most of the energy available on the planet, a theoretical milestone often cited in discussions of long-term technological growth, before leaping to a Type II civilization capable of exploiting stellar-scale power.
[15]Musk is messing with the Cosmic Dawn. Will alien hunters save the day for all mankind?
[16]Orbital datacenters subject to launch stress, nasty space weather, and expensive house calls
[17]Bezos plan for solar powered datacenters is out of this world… literally
[18]Satellite slinger AST reckons newer birds won't outshine stars in night sky
Given that this is one of Musk's fever dreams, concrete technical details of the satellite constellation remain sparse. SpaceX only noted in its filing that the satellites would rely on "high-bandwidth optical links" and claims the system could deliver on the order of 100 kW of power per metric ton allocated to computing, with traffic routed through SpaceX's Starlink network for transmission to authorized ground stations.
Also featuring prominently in the FCC application is SpaceX's [19]struggling Starship super rocket, which has yet to do much more than occasionally perform a test flight between [20]mishaps .
[21]
"With Starship's ability to deliver unprecedented tonnage to orbit for AI compute, the capacity for intelligence processing in space could surpass the electricity consumption of the entire U.S. economy," SpaceX said in its application, describing the craft as bringing "inherent efficiencies" to its ability to launch the constellation via Starship and scale it as demand requires.
In classic Musk fashion, there's a lot in this proposal riding on assumptions and technology that doesn't appear to actually exist yet.
"I think it's unclear at this stage whether it's feasible or not - but SpaceX seem to think it is," McDowell told us.
It could all be a pipe dream, or it could be a sign that Earth's orbit is about to get a lot more crowded and hazardous. The FCC is now soliciting public comments on SpaceX's application with a deadline of March 6.
SpaceX didn't respond to questions for this story. ®
Get our [22]Tech Resources
[1] https://fccprod.servicenowservices.com/icfs?id=ibfs_application_summary&number=SAT-LOA-20260108-00016
[2] https://regmedia.co.uk/2026/02/05/public-notice-spacex-ai-sats.pdf
[3] https://www.theregister.com/2024/05/01/space_insurer_record_loss/
[4] https://www.theregister.com/2023/07/10/starlink_satellites_leaking_astronomydisturbing_em/
[5] https://www.theregister.com/2025/12/03/space_telescopes_photobombed_leo_comms_satellites/
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aYUhEhk8N3exCOs62g921wAAANU&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aYUhEhk8N3exCOs62g921wAAANU&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aYUhEhk8N3exCOs62g921wAAANU&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[9] https://regmedia.co.uk/2026/02/05/spacex-orbital-dc-sat-narrative.pdf
[10] https://spacenews.com/an-object-is-now-orbiting-alongside-chinas-shijian-21-debris-mitigation-satellite/
[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aYUhEhk8N3exCOs62g921wAAANU&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[12] https://www.theregister.com/2025/10/06/starlink_vaporizes_satellites_daily/
[13] https://www.theregister.com/2026/02/03/musk_spacex_xai_merge_hallucination_burning_man/
[14] https://fortune.com/2026/01/27/x-uk-twitter-financial-filings-revenue-decline-58-percent/
[15] https://www.theregister.com/2025/07/23/opinion_column_starlink_interference/
[16] https://www.theregister.com/2025/07/25/orbital_datacenters_subject_to_all/
[17] https://www.theregister.com/2025/10/03/bezos_space_dcs/
[18] https://www.theregister.com/2025/04/28/ast_satellites/
[19] https://www.theregister.com/2025/09/22/nasa_starship_artemis_doubts/
[20] https://www.theregister.com/2025/11/21/spacex_super_heavy_mishap/
[21] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aYUhEhk8N3exCOs62g921wAAANU&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[22] https://whitepapers.theregister.com/
Re: I idly wonder
Obligatory [1]what if?
[1] https://www.youtube.com/watch?v=JqFSGkFPipM
Re: I idly wonder
"what the possibilities of bringing down any that are over the UK might be ?"
Easier to go after the ground stations.
Re: I idly wonder
Don't zap them, tax them. $1m per satellite per year "orbit rental" should come in handy, shared among the overflown countries.
"SpaceX wants to fill Earth orbit with a million datacenter satellites"
...cue more unforeseen satellite collisions and even more space debris. In addition to periodic meteor showers, we might end up getting quite a few more bright satellite reentry showers.
Well yes
A recently published report on this basically says that the Kessler Syndrome is now pretty much inevitable. So let’s make it worse!!!! Yay!!!!
You can take this job and shove them
"The FCC is now soliciting public comments ..."
Hey FCC, we'll be taking down your names and volunteering you for going up there and cleaning out those orbits you filled with junk. Oh, and you must provide your own transportation for this assignment.
> unforeseen satellite collisions
Some of those collisions might even be accidental. Not a lot of sovereign space in (near-) space. Maybe the Dogey donkey will contract out defence to Mom, her kill bots always do well.
"...cue more unforeseen satellite collisions and even more space debris."
I have the feeling that it won't be one launch per orbital data center. Adding the numbers for propellant, likelihood of BURG's (Blowed Up Real Good) on the launch pads taking out a whole launch facility and that sort of thing is something I'd rather leave to Randall Munroe. I should send in the idea.
The amount of Earth debris would be stunning. The cost would be...... wait for it.............. astronomical. Serviceability would be zero. Sans a robot that's far more advanced than what's around now to make repairs, launching people into the metallic soup that will be Earth orbit would be a non-starter.
I'm going to build an underground bunker on the lot next door after I buy it. Any suggestions about depth?
I'm a little confused by this idea. I've been led to believe that shedding heat in the vacuum of space is harder than in an atmosphere. Have I just been misled about that issue?
I've been led to believe that shedding heat in the vacuum of space is harder than in an atmosphere. Have I just been misled about that issue?
No, you have it right. In space you can't just dump waste heat into the nearest body of water (conduction) or depend on the atmosphere to carry it off(convection). You need to radiate it. Which implies a substantial dark surface to radiate from. And that surface has to be protected from direct sunlight lest it rapidly become a lot warmer and start pumping heat back into your data center.
That's not just a problem for these Muskian fantasy vehicles BTW. It's also likely to be an issue for the nuclear power plants some folks want to put on the moon.
You also have to have chips to put up there too.
Current smallest node that is radiation hardened seems to be around [1]12nm , whereas nVidia's Blackwell GPU are on TMSCs 5nm process.
[1] https://www.baesystems.com/en-uk/article/bae-systems-advances-rh12-storefront-with-new-radiation-hardened-circuit-technology-for-space-community
"It's also likely to be an issue for the nuclear power plants some folks want to put on the moon."
The best thing on the moon would be to use the heat from a reactor to warm up habitats. The problem there is the lack of data on how compacted the regolith is and how hard it would be to move that heat from a reactor that's likely a fair distance from any human occupied structure. I don't expect there will be too much shielding as that's very heavy. When the Apollo astronauts tried to take some cores, they found that just below the loose regolith, it was nigh on impossible to get the coring tool to go down. The jagged grains interlock something fierce. Perhaps that would make road building a matter of sweeping a path and fusing the little bit of loose material to the substrate. Lots to do and a reason why I don't see adding complication through trying to go to the south pole as being a good thing. It can be saved for later after the more utilitarian research is done.
I got to play a bit in a big box of JSC 1A stimulant a number of years ago. It's NOT sand. It will also take a load of time to work out how to deal with it.
Say again
Did you play in a big box of stimulant? Really? I can believe it wasn't sand, that's not usually very stimulating.
vtcodger has given good answer.
My own fag packet calculations, extrapolating from the weight/mass of the International Space Station's cooling system, suggests that 5 rocket launches would be needed just to put the cooling system in orbit for a 1MW data center.
You might reduce that by dispensing with the "center" bit and have CPUs glued to the heat radiator (thus avoiding the need for any heat-exchanger liquids) so that one satellite hosts a handful of servers, but that then just increases the engineering problems since everything needs to be attached to the cool plate - the CPUs, the RAM, the SSD, the electrical wiring and so on. Everything on a terrestrial server that is in air needs to have its cooling considered.
That's doable, of course, but it means that current server manufacturing can't be used: custom factories (or at least custom assembly lines) to build custom servers would be needed.
This makes my...
Biz plan to develop Kevlar-infused umbrellas more marketable.
Back to plan B
Let's build more terrestrial data centers. We can power them with fossil fuels and use local water resources to cool them.
It is utter non-sense to have orbiting data centers, functioning off-grid using solar power and radiating heat in to the vacuum of space.
Besides, aliens will probably take over the DC's anyway >> icon excuse.
Re: Back to plan B
"We can power them with fossil fuels and use local water resources to cool them."
Put them in Iceland. Run them from geo-thermal power and use the waste heat to keep roads de-iced, domestic hot water hot, homes heated, etc. A wee bit of latency to move the data is no big deal for eShopping and consumer applications. The same goes for commercial data analysis. An extra 5 seconds ain't gonna kill most applications. Where it might, users can pay a premium.
Re: Back to plan B
Just 2 problems with that,
Have a look at the blue lagoon power plant, it has 25 foot high earth walls all around it now due to a rather pesky volcano deciding to erupt over the past couple of years and on Iceland... thats a small volcano.
Now try putting your data center further east... nice ice caps for cooling water, plenty of thermal heat from the volcanos under the ice caps for power generation.
then the volcano erupts, your data center has now over heated just before a glacial outburst flood washes the remains into the north atlantic.
How about we don't do what is essentially patent squatting on what is a scarce international resource for what is on-the-face plainly fantastical.
How do they plan to radiate kW of heat by millions of satellites without blocking out the sky by thermal radiators? Heat regulation is a complex and highly technical problem that engineers have worked on since the beginning of the space age. And the US is going to give them permission based upon a shrug?
This is as fanciful as his terraforming Mars, which is also silly due to physics. I would like to be proved wrong, but I feel pretty safe in that assertion. I don't expect the laws of physics to change anytime soon. The reason why underwater data centers are far more attractive is the exact opposite of what space can offer. Just got to grift that money from the US government while equally daft people are making political decisions.
How do they plan to radiate kW of heat by millions of satellites without blocking out the sky by thermal radiators?
Very much tongue-in-cheek but maybe that would help mitigate the affects of global warming by reducing solar incidence on Earth.
"Very much tongue-in-cheek but maybe that would help mitigate the affects of global warming by reducing solar incidence on Earth."
At the point where solar irradiance comes into play, there would be serious consequences when it comes to food production.
Re: reducing solar incidence on Earth
Orbital data centre fantasies often involve sun synchronous orbit. The satellites never enter Earth's shadow so there is no down time / need for batteries. Likewise the satellites never cast a shadow on Earth.
If you want to global cooling wash these fantasies then move them to Sun-Earth L1. They would cast a shadow on Earth continuously. They would not end up burning in Earth's upper atmosphere. A Kessler cascade would not affect things in Earth orbit. The down sides are latency, comms power, the cost of getting there and even more difficult on site maintenance. Those problems are like all the other space data centre problems: no need to look for solutions until after investors have been fleeced.
I've spent my entire career as an ASIC designer worrying about mW of power. I find MW datacentres offensive, and the idea of putting them in orbit is bonkers. I doubt even SpaceX can manage millions of orbits without having failures to avoid collisions, given that some of those satellites will be non-responsive.
Kessler syndrome looking quaint? Just the opposite I'd have thought.
IPO Pumping
SpaceX is going public so expect regular Tesla like BS promises from Musk to pump the stock valuation.
would it even work?
I refer the honourable members to the recent problems experienced by airbus at 30,000 feet. How much random bit flipping are you going to get in medium earth orbit in the full force of the solar wind?
Flying junk
Imagine the obsolescence of those data centres in five years.
Re: Flying junk
And that, ladies and gentlemen, is that. It's not even worth talking about daily maintenance.
Re: Flying junk
Five years? That would be amazing for them to work that long. The orbital shells being thrown about would be longer than that, but the hardware, with no repairs, might be more difficult to keep going.
It's generally older and larger scale semiconductors that are used for space applications so that means any data centers would be based on decades old processors. The trade off for some extra rad-hardness seems a wee bit problematic. Racks of Z-80's? I've got a few tubes I could donate.
Kardashev nonsense
Shouldn't Musk pay Roddenbery, or whoever created the Kardashev typology, royalties?
Re: Kardashev nonsense
Unfortunately Nikolai Kardashev died in 2019.
There might be a few problems
This scheme seems unusually nutty even for Musk. But I should point out that these hypothetical flying data centers don't have to be in Low Earth Orbits. If they are farther out, there's more room. But I'm not sure data centers are going to work that well in the Van Allen radiation belts. And the further out the satellites are, the greater the latency.
All of your orbits now belong to us
US space grab.
No you can't park your satellite here, this space is reserved.
Re: All of your orbits now belong to us
How come the FCC authorises everybody's space usage? Doesn't the rest of the planet get a say in the matter? Aren't there another 194 countries who might wish to have some access through for their own projects? Did they foolishly agree a treaty where the FCC gets to decide?
Re: All of your orbits now belong to us
"How come the FCC authorises everybody's space usage?"
Getting the permissions for radio frequencies/bandwidth is one of the steps. The FCC only has authority in the US, so it's not "everybody" although there are treaties.
It's a bit pointless to move forward if there's no way to move the data back and forth.
Is Musk's mega datacenter constellation even feasible?
tl;dr
No.
Utterly impractical
There are huge numbers of essentially insuperable problems with plans of this kind, see [1]this very recent article for an objective analysis.
[1] https://theintercept.com/2026/01/12/data-centers-space-ai/
Ridiculous
I recently read an excellent blog post by "a former NASA engineer/scientist with a PhD in space electronics": https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/
TL;DR: There are huge technical obstacles here, particularly with regard to heat dissipation and radiation hardening. Modern GPUs are very sensitive to both transient bit flips and permanent damage from radiation, and have to dissipate massive amounts of heat.
Even ignoring technical obstacles, the economics make _zero_ sense in comparison to terrestrial datacenters.
Re: Ridiculous
That really is a good article. Here's a clickable link [1]https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/ for those who hate cutting and pasting or using devices where cut/paste are a pain.
[1] https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/
It's not about Space, or Sentience...
By now we should all understand that Musk is playing the three cup game, and juggling unprofitable businesses around by wrapping them up in nonsense 'visionary' shells.
SpaceX, through its deep connection with government contracts, is possibly the only reliable money maker in his whole empire (so long as it is tolerated by the American President and other governments using Starlink for strategic communications). Burying X and xAI in there hides the failures. Adding a nonsense vision to replace the previous nonsense vision of cities on Mars is designed to distract from the massive delays (and possible failures) of the Starship program (the Cybertruck to the F9's Model 3).
And then, with a lot of hype (look! money! visions! sentient suns!) we move towards an IPO before anyone questions why xAI and the Optimus programme are in separate companies, and how chaotic the Artemis project has become.
Focus on the cups. Round and round they go. Where has the pea gone? Is it under here? No, silly - Musk has taken it.
Kessler syndrome won't happen
The orbital DCs will be Fully Self Driving.
The Sound of Drums
It's just dawned on me! Musk wants to be Harold Saxon and the "satellites" are his Toclafane! Oi Musk! You're the master of bullshit, nothing more.
I idly wonder
what the possibilities of bringing down any that are over the UK might be ?
If he can put a million satellites into space. then surely a few hundred motivated nerds can reply with a few hundred perfectly aimed lasers ?
Probably not going to do much, but boy, what a project to work on. Who knows, it may be useful in the future.