Google and Amazon are doing the same thing. Maybe it is a moonshot (pun intended), but Musk is hardly alone in the push.
https://www.wsj.com/tech/bezos-and-musk-race-to-bring-data-c...
https://www.nytimes.com/2026/01/01/technology/space-data-cen...
https://futurism.com/advanced-transport/spacex-buying-unfath...
Plenty of defense contractors with classified projects are already publicly listed, so this is not uncharted territory.
Lockhead Martin for example: https://investors.lockheedmartin.com/news-releases/news-rele...
Gives this level of detail:
> Aeronautics classified program losses $(950)
> MFC classified program losses -
It seems very safe from a national security perspective.
I’m not that smart, but if I were, I would be thinking this is an extended way to move the losses from the Twitter purchase on to the public markets.
[1] https://www.axios.com/2023/12/31/elon-musks-x-fidelity-valua...
[2] https://www.reuters.com/markets/deals/musks-xai-buys-social-...
[3] https://www.cnbc.com/amp/2026/02/02/elon-musk-spacex-xai-ipo...
No? ISS isn't exempt from legal systems.
https://www.esa.int/Science_Exploration/Human_and_Robotic_Ex...
The craziest part of those statements is "100 kW per ton." IDK what math he is doing there or future assumptions, but today we can't even sniff at 10 kW per ton. iROSA [1] on the ISS is about 0.150 kW per ton.
[1]https://en.wikipedia.org/wiki/Roll_Out_Solar_Array
edit: iROSA = 33 kW per ton, thanks friends
Building this is definitely not trivial and not easy to make arbitrarily large.
https://www.planetary.org/articles/20170929-spacex-updated-c...
They’d need incredible leaps in efficiency for an orbiting ton collecting and performing 100 KW of compute.
[0] https://en.wikipedia.org/wiki/Electrical_system_of_the_Inter...
You think I'm joking but I'm not. https://spacenews.com/spacex-files-plans-for-million-satelli...
Was ignored on HN but here's an article explaining:
https://arstechnica.com/space/2025/12/after-years-of-resisti...
NVIDIA H200 is 0.7 KW per chip.
To have 100K of GPUs you need 500 ISSs.
ISS cooling is 16KW dissipation. So like 16 H200. Now imagine you want to cool 100k instead of 16.
And all this before we talk about radiation, connectivity (good luck with 100gbps rack-to-rack we have on earth), and what have you.
—
Sometimes I think all this space datacenters talk is just a PR to hush those sad folks that happen to live near the (future) datacenter: “don’t worry, it’s temporary”
https://www.nytimes.com/2025/10/20/technology/ai-data-center...
https://www.jalopnik.com/did-musk-propose-hyperloop-to-stop-...
[1] https://www.cnn.com/2025/10/20/science/nasa-spacex-moon-land...
According to this other source https://www.satellitetoday.com/connectivity/2026/02/02/space...
the filing mentions this
> these satellites would operate between 500 km and 2,000 km altitude and 30 degrees and Sun-Synchronous Orbit inclinations (SSO)
There are a lot of degrees of freedom to optimize something like this.
Spacecraft radiator system using a heat pump - https://patents.google.com/patent/US6883588B1/en
Company website:
https://rdw.com/wp-content/uploads/2023/06/redwire-roll-out-...
And their Opal configuration beats the metric: 5.3 kW for 42.7 kg.
> ROSA is 20 percent lighter (with a mass of 325 kg (717 lb))[3] and one-fourth the volume of rigid panel arrays with the same performance.
And that’s not the current cutting edge in solar panels either. A company can take more risks with technology choices and iterate faster (get current state-of-the-art solar to be usable in space).
The bet they’re making is on their own engineering progress, like they did with rockets, not on sticking together pieces used on the ISS today.
1. China is very concerned about Starlink-like constellations. They want their own, but mostly they want to be able to destroy competitors. That is really hard.
2. Many countries have single ASAT capabilities. Where one projectile can hit one satellite. However, this is basically shoot a bullet, with a bullet, on different trajectories.
3. > Sure, it'd take orbital launch capabilities to lift ... how many bags of metal scrap and explosives?
If I understand orbital mechanics... those clouds of chaff would need to oppose the same orbit, otherwise it is a gentle approach. In the non-aligned orbit, it's another bullet hitting a bullet scenarios as in 2, but with a birdshot shotgun.
My entire point is that constellations in LEO take hundreds of Falcon 9's worth of mass to orbit and delta-v to destroy them, as in-orbit grenades which approach gently. This IS REALLY HARD, as far as mass to orbit, all at once! If you blow up some group of Starlink, that chaff cloud will just keep in orbit on the same axis. It will not keep blowing up other Starlinks.
The gentle grenade approach was possibly tested by the CCP here:
Not that you would want 500+ square meters just for cooling of 200KW
And, mind you, it won’t be a simple copper radiator
https://www.nasa.gov/wp-content/uploads/2021/02/473486main_i...
https://wiki.pvmet.org/index.php?title=Standard_Test_Conditi...
So, a "400W panel" is rated to produce 400W at standard testing conditions.
I'm not sure how relevant that is to the numbers being thrown around in this thread, but thought I'd provide context.
https://www.spectrolab.com/company.html
Twenty-five years after the ISS began operations in low Earth orbit, a new generation of advanced solar cells from Spectrolab, twice as efficient as their predecessors, are supplementing the existing arrays to allow the ISS to continue to operate to 2030 and beyond. Eight new arrays, known as iROSAs (ISS Roll-Out Solar Arrays) are being installed on the ISS in orbit.
The new arrays use multi-junction compound semiconductor solar cells from Spectrolab. These cells cost something like 500 times as much per watt as modern silicon solar cells, and they only produce about 50% more power per unit area. On top of that, the materials that Spectrolab cells are made of are inherently rare. Anyone talking about scaling solar to terawatts has to rely on silicon or maybe perovskite materials (but those are still experimental).
We saw this on a much smalelr scale a decade ago when one of Elon's companies (Tesla) acquired a second one of Elon's companies (SolarCity) because it was broke and owed a ton of money to a third one of Elon's companies (SpaceX).
Elon was forced to go through with his impulsive Twitter acquisition by a Delaware court, an acquisition that was not only secured by a bunch of Tesla stock but also a bunch of Qatari and Saudi royal money. He then mismanaged Twitter so badly Fidelity wrote down its value by at least 80% [1].
So what did Elon do? Raised even more questionable foreign money into xAI, diverted GPUs intended for another of his companies (Tesla) into Twitter and then "merged" Twitter into xAI, effectively using other people's money to bail him out from an inevitable margin call on his Tesla stock.
Interestingly, Twitter was reportedly valued at $33 billion in this deal [2], significantly more than the less than $10 billion Fidelity valued Twitter at. Weird, huh? With a competent government, this would be securities fraud that would have you spend the rest of your life in jail. And even with all that, $11 billion was lost on the deal.
So here we are and it's time for the shell game to be played again. Now it's SpaceX's turn to bail out the xAI investors.
And what is the argument for all this? AI data centers in space. Words cannot describe how little sense this makes. Launch costs (even if the Starship launch costs get to their rosy projections), cooling in space, cosmic rays (and the resulting errors) and maintenance. Servers constantly need parts replaced. You can just deorbit the satellite instead but that seems like an expensive way of dealing with a bad SSD or RAM chip.
[1]: https://www.cnn.com/2024/10/02/business/elon-musk-twitter-x-...
[2]: https://www.cnbc.com/2025/03/28/elon-musk-says-xai-has-acqui...
https://taranis.ie/datacenters-in-space-are-a-terrible-horri...
I don't have any specialized knowledge of the physics but I saw an article suggesting the real reason for the push to build them in space is to hedge against political pushback preventing construction on Earth.
I can't find the original article but here is one about datacenter pushback:
https://www.bloomberg.com/opinion/articles/2025-08-20/ai-and...
But even if political pushback on Earth is the real reason, it still seems datacenters in space are extremely technically challenging/impossible to build.
https://inhabitat.com/worlds-largest-solar-project-sahara-de...
https://www.theguardian.com/business/2009/nov/01/solar-power...
(and a retrospective from 2023 - https://www.ecomena.org/desertec/ )
The article you linked agrees with me. Greatest resolution in the macula which is a span of approximately 6 degrees from the centre.
Sigh...
It is estimated that Starlink is, accounting for 70% - 80% of revenue. Sources: [1] and [2]
NASA is SpaceX's biggest external customer for rocket launch services.
Although NASA is SpaceX’s largest external customer for traditional launch services, the company earns far more revenue from Starlink customers (millions of subscribers). So overall Starlink itself is SpaceX’s biggest revenue generator and de facto largest customer segment.
[1] https://pestel-analysis.com/blogs/target-market/spacex
[2] https://londoneconomics.co.uk/blog/publication/crouching-riv...
Orbit gets you the advantage of 1/5th the PV and no large daily smoothing battery, but also no on-site installation cost, no grid interconnect fees, no custom engineering drawings, no environmental permitting fees, no grid of concrete footers, no heavy steel frames to resist wind and snow loads. The "on-site installation" is just the panels unfolding, and during launch they're compact so the support structure can be relatively lightweight.
When you cost building the datacenter alone, it's cheaper on earth. When you cost building the solar + batteries + datacenter, it (can be) cheaper in space, if you build it right and have cheap orbital launch.
I meant it specifically for figuring out cooling computers in space.
I am pretty sure this is going to be a solvable problem if this is the bottleneck to achieve data centers in space, given that newer chips are much more tolerant to high temperatures.
https://www.marketplace.org/story/2026/01/07/new-ai-chips-wi...
You don't build a rigid shell of course, you build a swarm of free-floating satellites in a range of orbits.
See https://www.aleph.se/Nada/dysonFAQ.html#ENOUGH for numbers.
Just because an idea has some factors in its favor (Space-based datacenter: 100% uptime solar, no permitting problems [2]) doesn't mean it isn't ridiculous on its face. We're in an AI bubble, with silly money flowing like crazy and looking for something, anything to invest it. That, and circular investments to keep the bubble going. Unfortunately this gives validation to stupid ideas, it's one of the hallmarks of bubbles. We've seen this before.
The only things that space-based anything have advantages on are long-distance communication and observation, neither of which datacenters benefit from.
The simple fact is that anything that can be done in a space-based datacenter can be done cheaper on Earth.
[1] https://en.wikipedia.org/wiki/A_Modest_Proposal for the obtuse
[2] until people start having qualms about the atmospheric impact of all those extra launches and orbital debris
https://en.wikipedia.org/wiki/Jack_Dorsey#Twitter
Still, don't make the mistake I did, which was to read the above comment to mean "he put more money in at the time of the buyout", since he was called an "investor in X".
When it all goes bankrupt, they can pay off the bonds for x¢ in the dollar and own SpaceX.
Perhaps if the gov could organize a little better, they'd make sure SpaceX owed lots of taxes and put themselves in front of the queue for ownership and screw other creditors (especially foreign).
Edit: looks like the US military doesn't spend that much on SpaceX: https://londoneconomics.co.uk/blog/publication/crouching-riv...
He's all over the Epstein files and his daughter has publicly verified that the timing works out and the emails are probably legitimate.
https://www.threads.com/@vivllainous/post/DUMBh2Vkk8D/im-jus...
We know datacenters in space - sound plausible enough - yet not practical - hence they're potential pure play - also you can have massive solar in space - unlimited space -- etc -- all true -- but how economical / practical is it ?
yet we know on earth - to power the whole earth with solar - only a fraction of the land is needed. Hell it's even in the Tesla Master Plan v3 docs [1] - current limitation being storage & distribution
so all you - are now witnessing to the greatest scam ever pulled on earth.
[0]: https://www.youtube.com/watch?v=BzAdXyPYKQo [1]: https://www.tesla.com/ns_videos/Tesla-Master-Plan-Part-3.pdf
https://www.nvidia.com/en-eu/data-center/dgx-h200/?utm_sourc...
Power draw is max 10.2 kW but average draw would be 60-70% of that. let's call it 6kW.
It is possible to obtain orbits that get 24/7 sunlight - but that is not simple. And my understanding is it's more expensive to maintain those orbits than it would be to have stored battery power for shadow periods.
Average blackout period is 30-45 minutes. So you'd need at least 6 kWh of storage to avoid draining the batteries to 0. But battery degradation is a thing. So 6 kWh is probably the absolute floor. That's in the range of 50-70 kg for off-the-shelf batteries.
You'd need at least double the solar panel capacity of the battery capacity, because solar panels degrade over time and will need to charge the batteries in addition to powering the gpu's. 12 kW solar panels would be the absolute floor. A panel system of that size is 600-800 kg.
These are conservative estimates I think. And I haven't factored in the weight of radiators, heat and radiation shielding, thermal loops, or anything else that a cluster in space might need. And the weight is already over 785 kg.
Using the $1,500 per kg, we're approaching $1.2 million.
Again, this is a conservative estimate and without accounting for most of the weight (radiators) because I'm too lazy to finish the napkin math.
A very high end desktop pulls more electricity than the whole JWST... Which is about the same as a hair dryer.
Now you need about 50x more for a rack and hundreds/thousands racks for a meaningful cluster. Shaded or not it's a shit load of radiators
https://azure.microsoft.com/en-us/blog/microsoft-azure-deliv...
The backing table is on page 8. Falcon 9 is (was, in 2018! It’s only cheaper now.) at $2700/kg to LEO. No one else is below $4k, except… Falcon Heavy.
> Despite a revenue drop from $5 billion in 2021 to roughly $2.7 billion in 2024, the EBITDA margin surged from 13.6% to 46.3% due to drastic cost-cutting measures and restructuring
In situ manufacturing. You just have to send enough to build the thing that builds the factory.
Not only did Elon not found Tesla[0], but many employees have described the "babysitters" or "handlers" who are responsible for making him feel like his ideas have been implemented, so that his caprice and bluster don't interfere with the actual operation of the company.
To give him his due, he's a phenomenal manipulator of public opinion and image, and he certainly has invested a lot of his emerald-generated wealth into numerous successful ventures - but he himself is not a positive contributor to their success.
[0] https://autoworldjournal.com/is-elon-musk-the-founder-of-tes...
Discussed earlier: >>46087616
The idea itself was proven by NASA with the DC-X but the project was canceled due to funding. Now instead of having NASA run it we SpaceX pay more than we'd ever have paid NASA for the same thing.
DC-X test flight: https://www.youtube.com/watch?v=gE7XJ5HYQW4
It's awesome that Falcon 9 exists and it is great technology but this guy really isn't the one anyone should want in charge of it.
Sources: https://www.science.org/doi/10.1126/science.aee8001 https://www.statista.com/chart/33709/tesla-byd-electric-vehi...
But I think there's solutions to the waste heat issue
https://www.nasa.gov/centers-and-facilities/goddard/engineer...
To be fair, he later added this:
>in a later interview with students from The University of Edinburgh in 2018, he referred to the premise of the Dyson sphere as being "correct and uncontroversial".[13] In other interviews, while lamenting the naming of the object, Dyson commented that "the idea was a good one", and referred to his contribution to a paper on disassembling planets as a means of constructing one.
Sources are in: https://en.wikipedia.org/wiki/Dyson_sphere
https://taranis.ie/datacenters-in-space-are-a-terrible-horri...
It's not comparable to any data center.
Datacenters in space, on the other hand, are a terrible idea because of the laws of physics, which will not get "solved" anytime soon. But don't take it from me, listen to this guy with a PhD in space electronics who worked at NASA and Google:
https://taranis.ie/datacenters-in-space-are-a-terrible-horri...
* https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/
* https://en.wikipedia.org/wiki/Tiangong_space_station
* "Technically challenging", a nice way to say "impossible"
* "I’m not that smart, but if I were, I would be thinking this is an extended way to move the losses from the Twitter purchase on to the public markets."
* "ISS radiators run on water and ammonia. Think about how much a kg costs to lift to space and you'll see the economics of space data centers fall apart real fast. Plus, if the radiator springs a leak the satellite is scrap."
* "5,000 Starship launches to match the solar/heat budget of the 10GW "Stargate" OpenAI datacenter. The Falcon 9 family has achieved over 600 launches." [nerdsniper]
* "No, we just "assume" (i.e. know) that radiation in a vacuum is a really bad way of dissipating heat, to the point that we use vacuum as a very effective insulator on earth."
* "World's Best At Surfing A Temporary Hyperinflation Wave is not a life goal to really be proud of tbh""Radiators can shadow each other," this is precisely why I chose a convex shape, that was not an accident, I chose a pyramid just because its obvious that the 4 triangular sides can be kept in the shade with respect to the sun, and their area can be made arbitrarily large by increasing the height of the pyramid for a constant base. A convex shape guarantees that no part of the surface can appear in the hemispherical view of any other part of the surface.
The only size limit is technological / economical.
In practice h = 3xL where L was the square base side length, suffices to keep the temperature below 300K.
If heat conduction can't be managed with thermosiphons / heat pipes / cooling loops on the satellite, why would it be possible on earth? Think of a small scale satellite with pyramidal sats roughly h = 3L, but L could be much smaller, do you actually see any issue with heat conduction? scaling up just means placing more of the small pyramidal sats.
The solar panels used in space are really lightweight, about 2 kg / m² [1], it's like ten times lighter weight than terrestrial panels. Still they need load-bearing scaffolding, and electrical conductors to actually collect the hundreds of kilowatts.
Water can't be made as lightweight though.
https://healthpolicy-watch.news/the-human-cost-one-year-afte...
For what it's worth, this project plans to use Tesla AI5/AI6 hardware for the first launches.
> ... we'd need a system 12.5 times bigger, i.e., roughly 531 square metres, or about 2.6 times the size of the relevant solar array. This is now going to be a very large satellite, dwarfing the ISS in area, all for the equivalent of three standard server racks on Earth.
https://taranis.ie/datacenters-in-space-are-a-terrible-horri...
The gist of it is that about 99% of cooling on earth works by cold air molecules (or water) bumping into hot ones, and transferring heat. There's no air in space, so you need a radiator 99x larger than you would down here. That adds up real fast.
>Factories on the Moon can take advantage of lunar resources to manufacture satellites and deploy them further into space.
I love how he goes from "the raw material is there" to "we will build high-tech supply chain to process them", just like that, magically.
https://i.imgur.com/wLJ60Vj.jpeg [I think you should be more explicit here in step two]
Also, https://xkcd.com/1724/
Edit: Formatting
The tale of computers is even more absurd. The first programmable, electric, and general-purpose digital computer was ENIAC. [1] It was built to... calculate artillery firing tables. I expect in the future that the idea of putting a bunch of solar into space to run GPUs for LLMs will probably seem, at the minimum - quaint, but that doesn't mean the story ends there.
https://blog.google/innovation-and-ai/technology/research/go...
You could argue that it doesn't really count though because it was only turing complete in theory: "A Colossus computer was thus not a fully Turing complete machine. However, University of San Francisco professor Benjamin Wells has shown that if all ten Colossus machines made were rearranged in a specific cluster, then the entire set of computers could have simulated a universal Turing machine, and thus be Turing complete."
It's trivial to understand why this is all hype if you pay attention to physics, as another commenter suggested earlier.
https://en.wikipedia.org/wiki/Stefan%E2%80%93Boltzmann_law
Assume you're radiating away the heat for a single B200 (~1kW), and the max radiator temp is 100C, you find A = ~3m^2.
So that's 3 square meters per GPU. Now if you take into account that the largest planar structure deployed into space is ~3k m^2 (https://investors.lockheedmartin.com/news-releases/news-rele...), you're looking at 1000 GPUs.
That's a single aisle in a terrestrial data center.
Cost to deploy on earth vs satellite is left as an exercise to the reader.
No, I put confidence my ability to do a web search, pretty rare skill nowadays ;)
You'll see that none of these are Elon/spacex, hopefully?
https://medium.com/@cognidownunder/google-just-announced-the...
https://www.ycombinator.com/companies/starcloud
https://www.informationweek.com/it-infrastructure/lunar-data...
Why do you feel this kneejerk reaction to defend Elon and his companies? You'll never be him. He doesn't care about you. He'd use you for reactor shielding for an uptick in Tesla share price without a second's hesitation. This is cultish behavior.
Do you have any idea who you're defending? I'll give you just one example. A right-wing influencer named Dom Lucre uploaded CSAM to Twitter, a video. But he didn't just upload it. He watermarked it first so had it on his computer and then postporcessed it. It was I believe up for days. This was apparently a video so bad that mere possession should land you in prison. And the fact that the FBI didn't arrest him basically tells you he'd an FBI asset. After taking days to ban him, Elon personally intervened to unban him. Why? Because reasons.
And this is the same man who it's becoming clear was deeply linked with Jeffrey Epstein, as was his brother [1].
Bringing this back to the original point: this is why Twitter lost 80% of its value after Elon acquired it. Advertisers fled because it became a shithole for CSAM and Nazis.
As for "basically no downtime" that's hilarious. I even found you commenting the classic anecdote "it was fine for me" (paraphrased) on one such incident when Twitter DDOSed itself [2].
Your cultish devotion here is pretty obvious eg [3]. I'm genuinely asking: what do you get out of all this?
[1]: https://www.axios.com/local/boulder/2026/02/02/kimbal-musk-j...
[2]: >>36555897
[3]: >>42836560
[0] https://developer.nvidia.com/deep-learning-performance-train...
Click the "Large Language Model" tab next to the default "MLPerf Training" tab.
That takes 16.8 days on 128 B200 GPU's:
> Llama3 405B 16.8 days on 128x B200
A DGX B200 contains 8xB200 GPU's. So it takes 16.8 days on 16 DGX B200's.
A single DGX (8x)B200 node draws about 14.3 kW under full load.
> System Power Usage ~14.3 kW max
source [1] https://www.nvidia.com/en-gb/data-center/dgx-b200
16 x 14.3 kW = ~230 kW
at ~20% solar panel efficiency, we need 1.15 MW of optical power incident on the solar panels.
The required solar panel area becomes 1.15 * 10^6 W / 1.360 * 10^3 W / m ^ 2 = 846 m ^ 2.
thats about 30 m x 30 m.
From the center of the square solar panel array to the tip of the pyramid it would be 3x30m = 90 m.
An unprecedented feat? yes. But no physics is being violated here. The parts could be launched serially and then assembled in space. Thats a device that can pretrain from scratch LLaMa 3.1 in 16.8 days. It would have way to much memory for LLaMa 3.1: 16 x 8 x 192 GB = ~ 25 TB of GPU RAM. So this thing could pretrain much larger models, but would also train them slower than a LLaMa 3.1.
Once up there it enjoys free energy for as long as it survives, no competing on the electrical grid with normal industry, or domestic energy users, no slow cooking of the rivers and air around you, ...
Then you have to also count the Z3 which predates the Colossus by 2 years.
https://wikipedia.org/wiki/Golden_Dome_(missile_defense_syst...
https://www.amazon.com/dp/162040592X
Télégraphe Chappe was a semaphore system using flags. It was not an electrical telegraph, nor was it binary.
Still, dropping a pod into the sea makes more sense than launching it into space. At least cooling, power, connectivity and eventual maintenance is simpler.
The whole thing makes no sense and is seems like it's just Musk doing financial manipulation again.
https://news.microsoft.com/source/features/sustainability/pr...
oh, we'll sure find a way to weaponize that energy for example - just imagine all those panels simultaneously turning their reflective back in a way to form gigantic mirror to focus reflected solar energy on your enemy, be that enemy in space or on the Earth/Moon/Mars ground. Basically space-scale version of 'death ray scyscrapper' https://www.businessinsider.com/death-ray-skyscraper-is-wrea....
Back in the day the Star Wars program was intending to use nuclear explosions to power the lasers, i guess once all that solar for AI gets deployed in space we wouldn't need the explosions anymore.
Interesting that such space deployment can deny access to space to anybody else, and that means that any competitive superpower has to rush to deploy similar scale system of their own. Space race v2.
However I'm curious how many solar panels you would need to power a typical data center. Are we talking something like a large satellite, or rather a huge satellite with ISS-size solar arrays bolted on? Getting rid of the copious amounts of heat that data centers generate might also be a challenge (https://en.wikipedia.org/wiki/Spacecraft_thermal_control)...
It's a political problem, not a tech problem
One matching what you say; the other saying they're up significantly, e.g. https://finance.yahoo.com/news/byd-overtakes-tesla-world-lar...
I do not know what to make of this.
However, it is unimportant, as the main concern for your argument should be all Chinese brands combined rather than any specific brand. Unfortunately, given I'm seeing two narratives that seem to be mutually exclusive for BYD, I don't think I can trust web searches to tell me about all brands combined either.
However, even that is unimportant, as my point was more focused on the price and value for money, how Chinese models compete on AI for less cost; even to do badly in this regard (which they might or might not be given the mutually incompatible news stories I've seen) is less a narrative about Chinese market failure and more of a demonstration that hardly anyone really cares about the AI in the first place.
3 times the area of the heat dissipating surface compared to solar panel surface brings the satellite temp down to 27 deg C (300 K):
> There is to little matter in space to absorb excess heat.
If that were true the Earth would have overheated, molten and turned to plasma long ago. Earth cools by.... radiative cooling. Dark space is 4 K, thats -267.15 deg C or -452.47 deg Fahrenheit. Stefan-Boltzmann law can cool your satellite just fine.
> You'd need thermal fins bigger than the solar cells.
Correct, my pessimistic calculation results in a factor of 3,...
but also Incorrect, there wouldn't be "fins" thats only useful for heat conduction and convection.
In your opinion, how credible is this story?
> [Kenyan Economist] Shikwati: … for God’s sake, please just stop.
> SPIEGEL: Stop? The industrialized nations of the West want to eliminate hunger and poverty.
> Shikwati: Such intentions have been damaging our continent for the past 40 years. If the industrial nations really want to help the Africans, they should finally terminate this awful aid. The countries that have collected the most development aid are also the ones that are in the worst shape. Despite the billions that have poured in to Africa, the continent remains poor.
https://www.aei.org/carpe-diem/kenyan-economics-expert-devel...
[0] Well, technically in favour of the grain! Pun not initially intended: https://en.wikipedia.org/wiki/Against_the_Grain:_A_Deep_Hist...
For a 230 kW cluster: 16 x DGX (8x)B200; we arrived at a 30m x 30m solar PV area, and a 90 meter distance from the center of the solar array to the tip of the pyramid.
1 GW = 4348 x 230 kW
sqrt(4348)= ~66
so launch 4348 of the systems described in the calculation I linked, or if you insist on housing them next to each other:
the base length becomes 30 m x 66 = 1980 m = ~ 2 km. the distance from center of square solar array to the tip of the pyramid became 6 km...
any of these systems would need to be shipped and collected in orbit and then assembled together.
a very megalomaniac endeavor indeed.
EDIT: found it on the Internet Archive:
https://web.archive.org/web/20251208110913/http://english.sc...
I will come back and give you my opinions.
I'm sorry, but what? Not only has it had multiple half days of downtime, two full days+, but just two weeks ago had significant downtime.
https://www.thebiglead.com/is-x-down-twitter-suffers-major-o...
[0] https://www.foxbusiness.com/business-leaders/spacex-boss-elo...
A better orbit might be Sun Synchronous (SSO) which is around 705 km, still not "deep space" but reachable for maintenance or short life deorbit if that's the plan. https://science.nasa.gov/earth/earth-observatory/catalog-of-...
And of course there are the LaGrange points which have no reason to deorbit, just keep using the old ones and adding newer.
I'm not a rocket scientist, but how do they plan to dispose of all the waste heat? The ISS carefully maintains its temperature, and it's not running racks-full of servers.
edit to add: this guy, who is a rocket scientist, explains exactly why it's a terrible idea, and yes, heat management is one reason. https://taranis.ie/datacenters-in-space-are-a-terrible-horri...
What good is compute if you can't interface with it? This is where we are now: https://en.wikipedia.org/wiki/Deep_Space_Optical_Communicati...
SpaceX may be leading in short-range (few hundred km) space-to-space data transfer but there is a long way to go for terabit/s deep-space links.
There were five separate flights to service the Hubble telescope. It was designed from the beginning to be repaired and upgraded.
https://www.youtube.com/watch?v=3VJT2JeDCyw
If these things were so safe the rich should build them next to their homes.
Tesla isn't even in the top 15 auto manufacturers by volume? The largest manufacturer Toyota produces 9x the cars Tesla does. Tesla is also on a multiyear sales drop with no sign of sales improvement.
The top 15 car makers produced 70 million cars, to Tesla's 1.7m. They have no enormous volume, at all.
https://en.wikipedia.org/wiki/List_of_automotive_manufacture...
If Tesla's stock traded in line with its competitors, its a $30-40B company. The hype around future growth (now completely off the charts) is the only reason the stock price is out of line with reality. There is no reason to expect Tesla's sales figures to improve going forward, in fact, they will continue to decrease.
> Tesla throws off cash which allows the flywheel to keep spinning
Tesla had a profit of $3.8b in 2025 (this is a 46% drop from 2024 and a second year over year drop). It's revenue was $94b (also less than 2024), which places it 12th among auto manufacturers. It's profit is 6th, which is a decent margin compared to legacy makers, but as mentioned above, the profit is plummeting as Tesla struggles to sell cars. It's revenue among all global companies is not even in the top 100.
It does not "throw off cash", the business is in a tailspin.
>They've constantly proved their naysayers wrong at every turn in time
Musk has been promising full self driving mode is within six months to a year away. He first made those claims in the mid 2010s? Do Tesla's have full self driving mode in 2026?
There is a decade long trail of failed claims from Musk and Tesla.
In 2019, Musk predicted 1 million Tesla robotaxis on the road by 2020. How many Tesla robotaxis are on the road in 2026? Fifty? One hundred? It's a rounding error compared to the claim that they'd have a million in 2020...
Musk said in 2019 that he believed Tesla vehicles were not traditional depreciating assets and instead could appreciate because they contained future-value technologies, especially Full Self-Driving (FSD): “I think the most profound thing is that if you buy a Tesla today, I believe you are buying an appreciating asset — not a depreciating asset.”
In fact, Tesla's are among the worst depreciating vehicles on the market today, their depreciation compares to the low end car market of Nissan, Hyundai and other low quality manfacturers.
Elon projected 250-500k Cybertruck sales per year. In reality, they sold 38k in 2024, and just 16k in 2025.
>They've constantly proved their naysayers wrong at every turn in time
[citation needed]
Because according to Bob Taylor, who initially got the funding for what became ARPANET:
> Taylor had been the young director of the office within the Defense Department’s Advanced Research Projects Agency overseeing computer research, and he was the one who had started theARPANET . The project had embodied the most peaceful intentions—to link computers at scientific laboratories across the country so that researchers might share computer resources. Taylor knew theARPANET and its progeny, the Internet, had nothing to do with supporting or surviving war—never did.Yet he felt fairly alone in carrying that knowledge.
> Lately, the mainstream press had picked up the grim myth of a nuclear survival scenario and had presented it as an established truth. When* Time magazine committed the error, Taylor wrote a letter to the editor, but the magazine didn’t print it. The effort to set the record straight was like chasing the wind; Taylor was beginning to feel like a crank.
* https://www.goodreads.com/book/show/281818.Where_Wizards_Sta... § Prologue
> Taylor told the ARPA director he needed to discuss funding for a networking experiment he had in mind. Herzfeld had talked about networking with Taylor a bit already, so the idea wasn’t new to him. He had also visited Taylor’s office, where he witnessed the annoying exercise of logging on to three different computers. And a few years earlier he had even fallen under the spell of Licklider himself when he attended Lick’s lectures on interactive computing.
> Taylor gave his boss a quick briefing: IPTO contractors, most of whom were at research universities, were beginning to request more and more computer resources. Every principal investigator, it seemed, wanted his own computer. Not only was there an obvious duplication of effort across the research community, but it was getting damned expensive. Computers weren’t small and they weren’t cheap. Why not try tying them all together? By building a system of electronic links between machines, researchers doing similar work in different parts of the country could share resources and results more easily. […]
* Wizards § Chapter 1
The first four IMPs were UCLA, SRI, UCSB, and Utah. Then BBN, MIT, RAND, System Development Corp., and Harvard. Next Lincoln Laboratory and Stanford, and by the end of 1970 Carnegie-Mellon University and Case Western Reserve University.
It was only "later in the 1970s" that command and control was considered more (Lukasik):
* https://en.wikipedia.org/wiki/ARPANET#Debate_about_design_go...
But the first two people who get the project going, Taylor and Herzfeld, were about the efficient use of expensive computer resources for research. Look at the firs >dozen sites and they were about linking researchers: the first DoD site wasn't connected until 3-4 years after things go going, and there was nothing classified about it. MILNET didn't occur until 1984:
Depends on the deserts in question and knock-on effects: Saharan Dust Feeds Amazon’s Plants.
* https://www.nasa.gov/centers-and-facilities/goddard/nasa-sat...
Helping vegetation in one place to grow may hinder it somewhere else. How important this is still appears to be an open question:
* https://www.nature.com/articles/s43247-020-00071-w
I'm not sure if humans are wise enough yet to try 'geo-hacking' (we're already messing things up: see carbon dumping).
[1]: https://www.reuters.com/business/autos-transportation/tesla-...
Reports in North Virginia and Texas are stating existing data centres are being capped 30% to prevent residential brownouts.
The author uses the power capacity of the ISS's solar panels as a point of comparison, but SpaceX has already successfully deployed many times that capacity in Starlink satellites[1] without even needing to use Starship, and obviously the heat dissipation problem for those satellites has already been solved so there's little point in hand-wringing about that.
The author also worries about ground communication bandwidth, claiming it is "difficult to get much more than about 1Gbps reliably", which seems completely ignorant of the fact that Starlink already has a capacity much greater than that.
The only unsolved technical challenge I see in that article is radiation tolerance. It's unclear how big of a problem that will actually be in practice. But SpaceX probably has more experience with that than anyone other than perhaps NASA so if they think it can be done I don't see much reason to doubt them.
Ultimately I think this is doable from a technical perspective, it's just a question of whether it will be economical. Traditional wisdom would say no even just due to launch costs, but if SpaceX can get Starship working reliably that could alter the equation a lot. We'll see. This could turn out to be a boondoggle, or it could be the next Starlink. The prospect of 24/7 solar power with no need for battery storage or ground infrastructure does seem tempting.
[1]: https://www.reddit.com/r/spacex/comments/zzwpue/with_starlin...
Here's Sundar talking about doing it by 2027: https://www.businessinsider.com/google-project-suncatcher-su...
The only intellectual dishonesty is “blaming it on the libs” argument. Ignoring the partisan arguments, the platform was quite literally being used by users to undress women and produce CSAM. [1] Just one of the many examples where you can argue the platform is toxic.
[1] https://www.reuters.com/legal/litigation/grok-says-safeguard...
So 3 years ago they managed to get to 10% of the power budget of one data center by accident, using satellites not explicitly designed for that purpose, using a partially reusable launch platform with 1/10th the payload capacity of Starship. My point is they've already demonstrated they can do this at the scale that's needed.
> A single Starlink satellite is using power in the order of watts
Then why does each satellite have a 6 kW solar array? Re-read that post I linked; the analysis is pretty thorough.
> Don't their current satellites have like 100Gbps capacity max?
Gen 3 is reportedly up to 1 Tbps ground link capacity, for one satellite.[1] There will be thousands.
> Do you have any idea how many 100Gbps routers go into connecting a single datacenter to the WAN? Or to each other (since intrahall model training is table stakes these days).
Intra-satellite connections use the laser links and would not consume any ground link capacity.
You're also ignoring that this is explicitly being pitched as a solution for compute-heavy workloads (AI training and inference) not bandwidth-heavy workloads.
The most efficient design and the most theoretically convincing one are not in general the same. I intentionally veer towards a configuration that shows it's possible without requiring radiating surface with an area of a square Astronomical Unit. Minimizing the physics and mathematics prerequisites results in a suboptimal but comprehensible design. This forum is not filled with physicists and engineers in the physical sciences, most commenters are programmers. To convince them I should only add the absolute minimum and configure my design to eliminate annoying integrals (for example the heat radiated by earth on the satellite is sidestepped by simply sacrificing 2 of the triangular sides of the pyramid to be mere reflectors of emissivity ~0, this way we can ignore the presence of a nearby lukewarm earth). Another example is the choice of a pyramid: it is convex and none of the surfaces are exactly parallel to the sun rays (which would result in ambiguity or doubt, or make the configuration sensitive to the exact orientation of the satellite), a more important consequence of selecting a convex shape is that we don't have to worry about heat radiated from one part of the satellite surface, being reabsorbed by another surface of the satellite (in view of the first surface), a convex shape insures no surface patch can see another surface patch of the satellite. And yes I pretend no heat is radiated by the solar panel itself, which is entirely achievable. So I intentionally sacrifice a lot of opportunities for more optimal design to show programmers (who are not trained in mathematical analysis, and not trained with physics textbook theorem-proof-theorem-proof-definition-theorem-proof-...) that physically it is not in the real of the impossible and doesn't result in absurdly high radiator/solar panel area ratios.
To convince a skeptic you 1) make pessimistic suboptimal estimates with a lot of room for improvement and 2) make sure those estimates require as little math and physics as possible, just the bare minimum to qualitatively and quantitatively understand the thermodynamics of a simple example.
You are asking the right questions :)
Given the considerations just discussed I feel OK forwarding you to the example mini cluster in the following section:
It describes a 230 kW system that can pretrain a 405B parameter model in ~17 days and is composed of 16x DGX B200 nodes, each node carrying 8x B200 GPUs. The naive but simple to understand pyramid satellite would require a square base (solar PV) side length of 30 m. This means the tip of the pyramid is ~90m away from the center of the solar panel square. This gives a general idea of a machine capable of training a 405B parameter model in 17 days.
We can naively scale down from 230 kW to 700 W and conclude the square base PV side length can then be 1.66 m; and the tip being 5 m "higher".
For 100 such 700 W GPU's we just multiply by 10: 16.6 m side length and the tip of the pyramid being 50 m out of the plane of the square solar panel base.
[1] https://www.forbes.com/sites/paultassi/2024/11/22/elon-musk-...
There's also a YC startup "Starcloud trains first AI model in space using Nvidia hardware" https://www.proactiveinvestors.com/companies/news/1084176/st...
>the satellite successfully ran Google’s open large language model Gemma and trained NanoGPT on Shakespeare’s works, generating responses in the style of the playwright.
https://oag.ca.gov/news/press-releases/attorney-general-bont...
https://www.reuters.com/legal/litigation/grok-says-safeguard...
https://www.nytimes.com/2026/01/09/technology/grok-deepfakes...
https://www.vogue.com/article/grok-deepfakes-trend-essay
https://www.the-independent.com/tech/ai-grok-twitter-fake-im...
https://techpolicy.press/the-policy-implications-of-groks-ma...
https://www.rollingstone.com/culture/culture-features/grok-s...
The French raided the X offices in Paris.
> https://www.theguardian.com/technology/2026/feb/03/french-he...
> It said the alleged offences it was investigating now included complicity in the possession and organised distribution of child abuse images, violation of image rights through sexualised deepfakes, and denial of crimes against humanity.
40% isn't much in the grand scheme of things, but maybe they can reach higher reduction with more research/materials. Mass and power are pretty cheap for spaceX, so shipping more solar panels and a heap pump might not be a deal breaker.
Would e.g. a reduction of 90% in radiator area change the overall picture on the overall feasibility? I think not, it would still be ludicrous, but I'd be happy to be proven wrong.
But even if share price is the metric for success, 33.6% over 5 years is like 6% compounded annually, which is okay I guess? [0]
[0] https://www.investopedia.com/magnificent-seven-stocks-840226...
https://www.imo.org/en/mediacentre/pressbriefings/pages/imo-...
I wouldn't be too surprised by beamed power being used on Mars, because that planet has global dust storms during which nowhere on the surface is getting much light, but it doesn't make as much sense here: because of the atmospheric window, you either use 0.4µm-to-10µm-wavelengths or 10cm-to-10m-wavelengths* with not much in between, µm means lasers and the mere possibility you may have included lasers powerful enough to be useful means everyone else will demand something similar to the IEA nuclear inspection program or will put similar lasers on the ground and shoot them upward to destroy those satellites, while cm-wavelengths means each ground station is a *contiguous* roughly 10km diameter oval.
Given the expensive part of large-scale PV has shifted from the PV itself to the support structures they're on, the ground station ends up about the same cost as a same-sized PV installation, and because that's just the ground station this remains true even if all the space-side components are zero cost. Normal ground-based PV also has the advantage that it doesn't need to be contiguous.
It is also possible to use a purely-ground-based method to transfer power from the other side of the world; a cable thick enough that the resistance is only 1 Ω the long way around is already within the industrial capacity of China, but the same geopolitical issues that would make people hostile to foreign beamed power satellites also makes such a cable a non-starter for non-technical reasons.
* https://en.wikipedia.org/wiki/File:Atmospheric_electromagnet...
*googles* Mid Jan this year? Yes, I was focusing on my German language course for the entire month. Only online here to relax.
> It wasnt close to fsd.
Except it was. Failure modes make Tesla's FSD a level-2 system, not even level 3: https://abc7news.com/post/mercedes-beat-tesla-become-1st-off...
Almost all businesses are more cautious than Musk, that doesn't tell you the systems are actually lower performance. The certification shows where they're at after all the smoke and mirrors, and where Tesla's at just isn't very impressive these days.
This difference isn't just a Euro/US split, most US companies are also more cautious, so same goes for Waymo who have been maintaining their slow-and-cautious approach despite what Musk keeps promising with Tesla, and operate actual robo-taxies in more cities than Tesla does.
> Ask your local llm for the earnings of a $.20/.30 per mile autonomous vehicle
I mean, I can do that in my head because 100,000 miles/year is a lot of driving even at motorway speed, and 1e5 times any cost per mile is trivial mental arithmetic, and even at 30¢/mile it still doesn't get you $50k/year/car.
30,000 miles/year is more likely, given constraints about when people most need vehicles and the relative fraction of time spent on motorways vs. urban areas, at which point 30¢/mile gets you more like $9k/year.
Also, crucially, 30¢/mile is what Waymo are already claiming as its operating cost. The reason this matters is that the moment anyone has competition on this (e.g. should Tesla actually do what they've been promising is 6-18 months away for the last decade), they don't corner the market and don't get to charge that much just because it's cheaper than a human Uber driver, they're facing off against other robo-taxi people with the same advantages who are, today, already operating in more places than Tesla are and without as much political stigma. Basically, when you get two competitors like this, it looks like the market for software and prices tend to costs; everyone in transport then only makes a profit when the demand exceeds supply, like this Monday in Berlin when my partner had to spend half as much on one single taxi ride as a monthly Deutschlandticket because of a strike action, but this kind of thing does not a business plan make.
They highlight the exact reliability constraint I was thinking of: that replacing failed TPUs is trivial on Earth but impossible in space. Their solution is redundant provisioning, which moves the problem from "operationally impossible" to "extremely expensive."
You would effectively need custom, super-redundant motherboards designed to bypass dead chips rather than replace them. The paper also tackles the interconnect problem using specialized optics to sustain high bitrates, which is fascinating but seems incredibly difficult to pull off given that the constellation topology changes constantly. It might be possible, but the resulting hardware would look nothing like a regular datacenter.
Also this would require lots of satelites to rival a regular DC which is also very hard to justify. Let's see what the promised 2027 tests will reveal.
