Wouldn’t it be easier to address the issues (power and water consumption, noise pollution) that lead people to object to data centers in their communities?
That’s not correct. The article is about data centers in space. The title of the article is “Why Orbital Data Centers Are Harder Than Silicon Valley Thinks”. The digression to other forms of computing in space seems irrelevant to whatever points about orbital data centers he may be trying to make, for or against.
ETA: You may be thinking of a different article about the thermodynamics of computing in space, but that’s not the article you linked and I was talking about what you actually linked to.
Just on the off chance some of you missed itt, there’s the parallel conversation going on in FQ with some of the same participants on the Space Data Center Issue:
Again, only on the chance people aren’t aware while we’re having one of our common breakouts of enlightened speculation in the Pit. 
But while we’re here - most of the speculation is da fuq - given the AI boom, almost all the components spoken for, scheduled out over the next year+, and that’s just the standard, commercial stuff with no hardening for space applications. So as for Elon’s claims, 99.99999% bullshit, because he may well have options on the basics, but no one is building space ready AI components on the scale he’s talking about right now. The leadup time alone isn’t something that’s going to be fixed by throwing even his level of assets around.
Fascinating article about the sheer colossal scale of some of the hyper-scale AI data centers currently being built, which require whole new approaches to engineering:
Modern AI data centers often use rack-scale systems, such as the Nvidia GB200 NVL72, which occupy a single data-center rack. Each rack contains 72 GPUs, 36 CPUs, and up to 13.4 terabytes of GPU memory. The racks measure over 2.2 meters tall and weigh over one and a half tonnes, forcing AI data centers to use thicker concrete with more reinforcement to bear the load.
A single GB200 rack can use up to 120 kilowatts. If Hyperion meets its 5-gigawatt goals, the data-center campus could include over 41,000 rack-scale systems, for a total of more than 3 million GPUs.
Demands on the power grid aren’t necessarily a problem. Meta’s Hyperion 5 GW data center in rural Louisiana will have its own set of three gas turbine generators. But CO2 emissions will still be a problem. The three turbines together will emit as much CO2 as the entire country of Latvia, or the same as about two million cars. To be fair, another data center they’re building in Ohio will be powered by a nuclear plant.
Naturally, Musk’s xAI data center in Memphis is the least efficient, currently powered by dozens of inefficient smaller gas turbine generators.
Are you saying there is no need for computers in space? Or that there is no need for additional computing in space applications? Because I read the article twice, know people who launch sensor payloads, and spent about 20 years of my life developing better computing for space applications. It is a shame that was all wasted effort.
Where on earth did I ever say such a thing?
I’m just objecting to the way that I think you mischaracterized the article, which is primarily and explicitly about large-scale AI data centers in space, as one might note from the huge title in big block letters and from the bulk of the content. It sounded to me like the author was trying to justify such orbiting data centers as helping to further the development of improved computing for satellites and spacecraft, and faster bandwidth capabilities, both of which are of course very important but it’s a bullshit justification for the data centers.
I’ll try one last time…
The article is explicitly about doing a systems engineering assessment of what is required to put the level of computing required for data centers into orbit and maintain them. The author is an engineer and shows what would be required and concludes that it can be done but is economically infeasible and doesn’t begin to compete economically with data centers on the ground. But he is enthusiastic about increasing the amount of compute on orbit, for specific niche needs.
However, as I did once as part of a 6 month study on a satellite that required 50 MW onboard (not for compute is all I can say about the application), he shows how to close on the technical aspects of the system because frankly it is definitely not impossible with current technology. But a 100 MW data center in orbit would cost on order of 100 billion dollars (extrapolated from my study), and that’s a non-starter, even for the richest man on the planet.
I’m beginning to understand why @Stranger bugged out
Apparently it’s because we’re all idiots.
Look, there’s no need to get your shorts in a knot. We apparently have a different interpretation of what the article was supposed to be about. Perhaps your background made that one tiny bit at the end of the article resonate with you. But I believe that in the absence of journalistic malfeasance (e.g.- “click-baiting”) the title of an article should be a pretty good indication of what the article is supposed to be about, don’t you? Does IEEE Spectrum do a lot of click-baiting? I may not be a spacecraft engineer, but I can read.
The question of compute capability in space for specific niche needs was a small bit at the end of the article – to me it was almost like an afterthought, and had little to do with the main topic, and certainly not a justification for the staggeringly enormous cost of such a foolhardy venture.
I guess my own bias is that these freaking billionaires are chasing a pipe dream with their orbiting data centers. The ones on terra firma are fine with me, provided that they’re environmentally responsible. Some are better than others, but so far, none are good. It’s really not acceptable to have a single data center emitting as much as 10 million metric tons of CO2 annually.
Naturally, on a pro-rated basis Elmo’s facility in Memphis is the worst of the lot.
Then why the fuck are you bringing up data centers if that’s not what it’s about?
It seems like you can’t even make up your own damn mind. And then argue about other people being idiots when you’re posting like you’re having an episode or something.
As you yourself point out, the article talks about datacenters in space being impractical, we are talking about datacenters being impractical, and then you rant about how we don’t get it.
If you think we need to lecture us about the importance of computers in space, well no fucking shit Sherlock. Even a car uses a computer these days. Next you’ll argue that the sun is hot, or that it can get dark at night.
Which of course is why night is the best time to perform your mission to the Sun. Your ship will be much less melty then. 
Why don’t astronaut people think of this if they’re so smart.
Well, not in space.
‘I always thought space was dark and cold’ he remarked vaguely.
‘Forgotten the sun?’ said Weston contemptuously.
– Out of the Silent Planet
Shhhhh, remember the first rule of nocturnal solar astronaut club
Space is a vacuum. It can’t be hot or cold. It can only be bright or dark.
There’s a temperature associated with that. That’s why an object will radiatively cool when above that temperature. (And radiatively warm when below.)
The underlying aspect is that radiant heat is a form of light energy. The hotter an item is, the brighter it will be. A hot object will radiate light in the infrared range, but a really hot object will radiate in the visible range. We can see it ourselves when hot objects are red and really, really hot objects are white. That’s all from light energy that’s leaving the object and cooling it off in the process. This process works even in a vacuum where convection/conduction isn’t feasible. However, if the object is in a location where it’s getting hit with a lot of light, then it will gain energy from that light. Eventually it will reach equilibrium between the light it’s radiating away into space and the light that is absorbing from space.
The “air density” in LEO is too low for its temperature to have much effect on objects in orbit, but the radiative heat from a certain largish ball of gas does heat objects.
Technically one could say empty space can’t be bright or dark, either, in the classical sense, only matter can, via blackbody radiation that depends only on its temperature.
However, although space can’t have a temperature in the classical sense which is defined by the energy of molecular vibration, since there are for practical purposes no molecules in empty space, it can have a temperature in the sense of quantum energy. So whether empty space can have a temperature has one of these tricky “yes and no” answers. Deep space has a fairly uniform temperature of 2.7 Kelvin (around --450°F, or --270°C) due to the Cosmic Microwave Background, the residual heat from the Big Bang.
At the time of the “last scattering” about 380,000 years after the Big Bang, when electrons bonded with protons to form the first hydrogen atoms, photons could travel freely for the first time and the universe became transparent, IIRC the quantum temperature of space was around 5000°F. Then, as now, the CMB heat is comprised entirely of photon energy.
A vacuum would have no temperature, but much of space isn’t a true vacuum, just a relative one. The space where we travel in (especially in our planetary orbit as proposed for datacenters) is full of stuff; particles, radiation, etc. And those things have a temperature.
You have to get very far, far from any stars to find anything approaching a true vacuum.