That's not a new problem that no one has dealt with before. The ISS for instance has its External Active Thermal Control System (EACTS).
It's not so much a matter of whether it's an unsolvable problem but more like, how expensive is it to solve this problem, what are its limitations, and does the project still makes economic sense once you factor all that in?
It's worth noting that the EACTS can at maximum dissipate 70kW of waste heat. And EEACTS (the original heat exchange system) can only dissipate another 14kW.
That is together less than a single AI inference rack.
And to achieve that the EACTS needs 6 radiator ORUs each spanning 23 meters by 11 meters and with a mass of 1100 kg. So that's 1500 square meters and 6 and a half metric tons before you factor in any of the actual refrigerant, pumps, support beams, valve assemblies, rotary joints, or cold side heat exchangers all of which will probably together double the mass you need to put in orbit.
There is no situation where that makes sense.
-----------
Manufacturing in space makes sense (all kinds of techniques are theoretically easier in zero G and hard vacuum).
Mining asteroids, etc makes sense.
Datacenters in space for people on earth? That's just stupid.
Your calculations are based on cooling to 20c, which is exponentially harder than cooling to 70c where GPUs are happy. Radiators would be roughly 1/3 the size of the panels for 70c.
I get that vacuum is a really good insulator, which is why we use it to insulate our drinks bottles. So disposing of the heat is a problem.
Can't we use it, though? Like, I dunno, to take a really stupid example: boil water and run a turbine with the waste heat? Convert some of it back to electricity?
What do you do with the steam afterwards? If you eject it, you have to bring lots of it with your spacecraft, and that costs serious money. If you let it condensate to get water again, all you did is moving some heat inside the spacecraft, almost certainly creating even more heat when doing that.
It's a good question, but in a closed system (like you have in space) the heat from the turbine loop has to go somewhere in order to make it useful. Let's say you have a coolant loop for the gpus (maybe glycol). You take the hot glycol, run it through your heat exchanger and heat up your cool, pressurized ammonia. The ammonia gets hot (and now the glycol is cool, send it back). You then take the ammonia and send it through the turbine and it evaporates as it expands and loses pressure to spin the turbine. But now what? You have warm, vaporized, low pressure ammonia, and now you need to cool it down to start over. Once it's cool you can pressurize it again so you can heat it up to use again, but you have to cool it, and that's the crux of the issue.
The problem is essentially that everything you do releases waste heat, so you either reject it, or everything continues to heat up until something breaks. Developing useful work from that heat only helps if it helps reject it, but it's more efficient to reject it immediately.
A better, more direct way to think about this might be to look at the Seebeck effect. If you have a giant radiator, you could put a Peltier module between it and you GPU cooling loop and generate a little electricity, but that would necessarily also create some waste heat, so you're better off cooling the GPU directly.
However there are workarounds. People are talking like the only radiator design is the one on the ISS. There are other ways to build radiators. It's all about surface area. One way is to heat up a liquid and then spray it openly into space on a level trajectory towards a collecting dish. Because the liquid is now lots of tiny droplets the surface area is huge, so they can radiate a lot of heat. You don't need a large amount of material as long as you can scoop up the droplets the other end of the "pipe" and avoid wasting too much. Maybe small amounts of loss are OK if you have an automated space robot that goes around docking with them and topping them up again.
The ISS consumes roughly 90kW. That’s about *one* modern AI/ML server rack. To do that they need 1000 m^2 of radiator panels (EACTS). So that’s the math: every rack needs another square kilometer of stuff put into orbit. Doesn’t make sense to me.
And what happens every time a rack (or node) fails? Does someone go out and try to fix it? Do we just "deorbit" it? How many tons per second of crap would we be burning in the upper atmosphere now? What are the consequences of that?
How do the racks (or nodes) talk to eachother? Radios? Lasers?
What about the Kessler Syndrome?
Not a rocket scientist but 100% agree this sounds like a dead end.
Communication is a well-understood problem, and SpaceX already has Starlink. They might need pretty high bandwidth, but that's not necessarily much of a problem in space. Latency could be a problem, except that AI training isn't the sort of problem where you care about latency.
I'd be curious where exactly they plan to put these datacenters... In low Earth orbit they would eventually reenter, which makes them a pollution source and you'd have no solar power half the time.
Parking them at the Earth-Sun L1 point would be better for solar power, but it would be more expensive to get stuff there.
Seasons mess that up unless you're burning fuel to make minor plane changes every day. Otherwise you have an equinox where your plane faces the sun (equivalent to an equatorial orbit) and a solstice where your plane is parallel to the sun (the ideal case).
Heat exchanger melts salts, salts boil off? Some kind of potential in there to use evaporants for attitude/altitude correction. Spitballing. Once your use case also has a business case, scope to innovate grows.
It makes sense to target a higher operating temperature, like 375K. At some point, the energy budget would reach an equilibrium. The Earth constantly absorbs solar energy and also dissipates the heat only by radiative cooling. But the equilibrium temperature of the Earth is still kind of cool.
I guess the trick lies in the operating temperature and the geometry of the satellites.
Asking for a friend (who sucks at thermodynamics:) could you use a heat pump to cool down the cold end more and heat up the hot end much higher? Heat radiation works better the higher the temperature?
The distinction is that what they are doing for Webb is trying to dissipate small amounts of heat that would warm up sensors past cryogenic temperatures.
Like on the order of tens or hundreds of watts but -100C.
Dissipating heat for an AI datacenter is a different game. A single AI inference or training rack is going to be putting out somewhere around 100kW of waste heat. Temps don't have to be cryogenic but it's the difference between chiselling a marble or jade statue and excavating a quarry.
That's a solution for minuscule amounts of heat that nevertheless disturb extremely sensitive scientific experiments. Using gold, no less. This does not scale to a crapton of GPU waste heat.
Anything is possible here, it's just there's no goddamn reason to do any of this. You're giving up the easiest means of cooling for no benefit and you add other big downsides.
It's scifi nonsense for no purpose other than to sound cool.
Getting better at creating and erecting solar panels & AI datacenters on earth is all well and good, but it doesn't advance SpaceX or humanity very much. At lot of the bottlenecks there are around moving physical mass and paperwork.
Whereas combining SpaceX & xAI together means the margins for AI are used to force the economies of scale which drives the manufacturing efficiencies needed to drive down launch etc.
Which opens up new markets like Mars etc.
It is also pushing their competitive advantage. It leaves a massive moat which makes it very hard for competitors. If xAI ends up with a lower cost of capital (big if - like Amazon this might take 20 years horizon to realize) but it would give them a massive moat to be vertically integrated. OpenAI and others would be priced out.
If xAI wants to double AI capacity then it's a purely an automation of manufacturing problem which plays to Elons strengths (Tesla & automation). For anyone on earth doubling capacity means working with electricity restrictions, licensing, bureaucracy, etc. For example all turbines needed for electricity plants are sold years in advance. You can't get a new thermal plant built & online within 5 years even if you had infinite money as turbines are highly complex and just not available.
Hmm, Elon really did run that flywheel pretty well. He built the Roadster to drum up some cash and excitement so he could develop the Model S, then he used that success to do the Model X, and then he expanded capacity to develop the 3 and Y, and he reinvested the profits to develop the Model 2, finally bringing EVs to the masses, displacing ICEs everywhere, and becoming the undisputed leader of both EV and battery manufacturering.
Oh wait, that didn’t actually happen, because he got distracted or something? He doesn’t really have battery capacity worth writing home about, the Chinese are surpassing Tesla in EV manufacturing, and Waymo is far ahead in self-driving.
The amazing space computation cost reduction process sounds rather more challenging than the Model 2, and I’m not sure why anyone should bet on Elon pulling it off.
Not sure how you can say that. Nothing lasts forever, especially in the face of Chinese market dumping, but for a while there Tesla really was the undisputed king of EV manufacturing, that flywheel is how he got there, he did release all the patents because he said from day one he didn't anticipate or aim for 100% market share for Tesla and assumed there'd always be lots of EV manufacturers in future. All that sounds like - mission accomplished?
As for Waymo being ahead, maybe today. But Waymo's tech stack is largely pre-DL, they rely heavily on unscalable techniques like LIDAR and continuous mapping. Tesla is betting big on the "scale up neural networks" model we know works well and their FSD can drive everywhere. They're perhaps behind Waymo in some ways, but they're also in different markets - Waymo won't sell anyone a self driving car and Tesla will. I wouldn't count them out. Their trajectory is the right one.
> I’m not sure why anyone should bet on Elon pulling it off.
PayPal, SpaceX existing at all, then doing reusable rockets, Tesla, FSD, large scale battery manufacturing, Starlink, X ("he can't fire 80% of employees it'll crash immediately"), robotics, training a SOTA LLM so fast even Jensen Huang was shocked ... the man consistently pulls off impossible seeming things in the face of huge skepticism. How many examples does it take before people start taking the guy seriously? Infinity examples?
Paypal is in no way a Musk creation, no one makes that claim and in fact they got rid of him quite quickly.
X has plummeted in value, and is worth a fraction of what he paid for it? How is this "pulling it off" by shrinking the user base, revenue, etc? While we don't have publicly audited figures, they announced a net loss for the first three quarters of 2025, while it posted profits prior to his purchase.
FSD isn't even real? Why would you cite a feature that doesn't actually exist as an example of "Elon pulling it off"? He promised FSD would be available over a decade ago, and it's still not real.
> How many examples does it take before people start taking the guy seriously?
I'd personally settle for real examples, and not the false ones cited above.
Tesla invested into the first Lotus roadster - and put that cash into the S then the X. Used that cash to build the worlds largest factories and make the 3 & Y which sold at enormous volumes - so large in fact that the S & X are now tiny single percentages of sales which is why Tesla is stopping manufacturing them now.
Tesla is one of the very few vehicle manufactures which makes a profit manufacturing vehicles. Tesla throws off cash which allows the flywheel to keep spinning.
Tesla is now operating fully autonomous rides. They've constantly proved their naysayers wrong at every turn in time. What the Chinese are doing in battery tech is irrelevant to US vehicles as they will never be allowed to sell in the US which is Teslas largest market.
The model 2 has the possibility of being profitable at insanely low purchase price which has the potential to completely disrupt the economics of US sales in such a way that legacy auto could well be bankrupt in 5-10 years. Who will be making Waymo's vehicles then?
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.
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
Hey remember that time someone had their Tesla running down the highway and the superior self-driving capability failed to see an 18 wheeler that crossed the road and the person was decapitated and there are videos of that complete with blood spray?
There is nothing we need on Mars other than science. It's not a market because there isn't money to be made outside of what is required to do whatever economically useless but scientifically valuable efforts we can convince people to fund.
We can't build an independent colony we can't live there any time soon. Arguably it may never make sense to live there.
1. Mankind never systematically lived in caves; that's just where remains and rock paintings are more likely to have survived.
2. Farming didn't evolve from a vision of "let's stay in one place, so let's find a way to do it"; it evolved from the gradual application of accumulated practical knowledge under real constraints until eventually it was possible to stay in one place. If Paleoelon had somehow convinced early humanity to abandon hunter-gathering and settle into a sedentary life because he had a vision for new markets around farming it would have led to the earliest famine.
While what you say is mostly correct, the lifestyle switch to farming was determined not by some random gradual accumulation of knowledge during the previous million years, but by accelerated accumulation of knowledge during a few thousand years at most, which was caused by the dwindling hunting resources, which forced humans to abandon the lifestyle that they had for a couple million years and switch to a lifestyle where the staple food consisted of plant seeds, with anything else providing much less of the nutrient intake. Only after a few more thousand years, raising domestic animals allowed the return to a more diverse diet.
Switching to a farming lifestyle was certainly not done by choice, but to avoid death by starvation, as we now know that this has caused various health problems, especially in the beginning, presumably until experience has taught them to achieve a more balanced diet, by combining at least 3 kinds of plant seeds, 2 with complementary amino acid profile and 1 kind of oily seeds for essential fatty acids (the most ancient farming societies have combined barley or einkorn or emmer wheat with lentils or peas or a few other legumes less used today and with flax seeds).
Yes, your description of how farming and sedentary lifestyle progressed is much more accurate than my somewhat clumsy attempt. My intention was to emphasise that such a transformative event in human history did not take place thanks to visionaries going against the grain [0] , but rather through a long and complex process.
Datacenters in space are a TERRIBLE idea.
Figure out how to get rid of the waste heat and get back to me.