This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Starship bet update
A few years ago I made a series of bets about Starship making it to orbit with other posters, last rounded up here:
The last one is a real nail-biter. When I heard about the SpaceX IPO I first thought it's time to call it a day. My model for my predictions about Elon was that he has a hype-compulsion, making wilder and wilder promises to get money out of investors, and as it becomes clear he won't be able to reach the hyped up goal, at some point they will get fed up with him. So when the news of the $85.7 billion came out, I figured that even if I do win, it will be on a technicality - maybe they won't pull it off by end of this year, but this sort of money will surely be enough to get them over whatever humps they run into on the road.... Then again maybe not! It also turned out that they have $41.3 billion in accumulated losses since their founding, and have burned $4.3 billion on AI in Q1 2026 alone, so maybe I will lose on a technicality instead, where they will indeed get to orbit by end of year, but will be dragged down by the unprofitable parts of the company.
I now believe that such a "loss on a technicality" is a pretty likely outcome, precisely because of the IPO. Like I said last year, if my bet was with Elon, he probably could have ordered the damn rocket to be put in orbit, just to prove a point, and while I'm lucky enough to have made my bet with internet randos instead, the IPO changes the dynamics such that he will be very tempted to do such things just to prove a point. Currently 95% of SpaceX stock held by insiders is locked up and it will be gradually released over the course of the year. Stonks are largely guided by hype, hype is generated with media articles (such as "SpaceX makes history with Starship orbital launch!!!11"), so while a frivolous orbital launch would make little sense before, it could make a lot of sense now. There's already talk of Starship 14 being orbital, and I fully expect them to schedule it just before one of these unlock dates.
That said, it's not over until it's over! Just because they might want to do it, doesn't mean they'll pull it off. This whole bet is starting to feel like an episode of Wacky Races.
I haven't been following Starship progress over the last 12 months, and all your bets are essentially bets about timing, which is contingent on uninteresting factors like the political environment, Elon's newest distractions (attention and finance wise), stochastic problems and causes for caution, so I won't comment on them. Ignore if you're in this for the pure love of the game.
But bets aside, you make a categorical prediction:
Have you elucidated your logic anywhere?
I'm afraid you have a case of Musk Derangement Syndrome. I see it a lot on X. Musk has a lot (as in, millions, a significant percent of X population) of extremely annoying fanboys of the lowest castes – crypto bros boosting #grok who got rich off $DOGE pumps, bots, edgy right-wingers, desperate $TSLA investors who are literally, well, invested in his success. He is obnoxious himself, prone to making false promises, grandiosity and loathsome behavior. So there's a reactionary cohort that naysays everything he does. But isn't this beneath human dignity to let that influence the judgement of the technical project such as Starship?
Starship, at this point, essentially can't not work. We know of no compelling reason why it won't, and a plethora of reasons why it will. Exactly a decade ago, there was vigorous skepticism that Falcon program can work. Russians in particular, being pathologically proud of Soviet space industry, dunked on the idea of rocket reusability with our typical overwrought literary wit, which hopefully can evoke some cringe in you today:
It goes on for a while but the conclusion is obvious already: Falcon is Another American Grift, the metal will get le tired, defect inspection will be prohibitively costly, the construction is suboptimal modulo reusability, and anyway the first landed unit didn't qualify for reuse, so QED. Coming from an engineer by training, this all sounded persuasive to my engineer friends at the time. To me, it sounded like status anxiety. It sounds quaint today, when Booster B1067 has a record of 35 launches, when Falcons provide the majority of LEO lift capacity for the planet, when the shortest turnaround is a bit over a week, and the safety track record of Falcon has exceeded that of Soyuz, painstaikingly built over half a century. The metal seems really vigorous and not tired at all. My understanding is that Elon's hypothesis was: all of the industry was thinking too small, these paranoid quality standards and laborious procedures are mostly downstream of cost ker kilogram to orbit, you can just do propulsive landing well enough that the vehicle takes negligible damage, and this unlocks a whole different regime of unit economics; and this is a mere issue of engineering. Seems like he was just correct. Then Starlink happened. Similar dismissals, similar outcome, SpaceX acquires the perfect demand sink and revenue stream and can seriously invest into what is functionally and economically near-equivalent to a reusable SSTO with 100+ tons of payload. But you know Starship's pitch, of course, and how it renders SLS and all other alternatives obsolete. Mars or Moon – in the context of full reusability with these payloads, does it even matter? These are mission details, what is important is what kinds of missions you can begin to plan at all at $1000/kg to LEO, at $100/kg, at $50/kg… and, much as I loathe to agree with @Shakes, the military can come up with quite a few. «Spy catellites» is thinking too small, for sure. On the civilian side, the space compute idea will genuinely work too, given political and logistic problems with terrestial datacenters in the US – and the objections to it are more motivated thinking, not solid engineering or bottom-line costs analysis; and this can trivially become another Starlink. You can start to actually think about microgravity manufacturing, as well. There is a lot to do in space, once you can get there cheaply. The last Starship feat that I've watched was the chopstick capture, it looked like they're really close to maturity. It can take a year or 5 years, but the probability of Elon running out of capital on the way there in the American system is… remote. So what's the actual crux? You say it's not scaleable and cite an article about Raptor production from 2021. They're on Raptor 3 now, all the concerns in that email are, far as I can tell, obsolete. Do you have some physics-driven argument as to why Falcon works but Starship does not? I am confident that you don't, because I've never seen any and apparently neither have SpaceX's investors, for all the hate Elon gets.
There is another strong reason to think that Starship can work. We had more ambitious designs in the 20th century, and today other companies are doing similar things. New Glenn works, 9x4 will haul 70 tons, and although they've had a setback with explosion on the pad, Bezos will see to it that they recover, they have their own constellation program that adds urgency, and will need heavy lift capability. More saliently, LandSpace has a pretty well-validated engine of roughly Raptor 2 class, and plans to use it in a Starship-class rocket somewhere after 2030; this far they've been fast-following SpaceX at a crazy pace, they've started in 2015 and have actually put the first methalox-powered rocket in orbit (3 years ago), so I'm optimistic about this schedule. Within a month they will likely make their second attempt at landing ZQ-3, which is basically a Falcon-9 with Starship characteristics (steel body, methalox). The first one failed in Dec 2025, but it was close and Elon himself said it's potentially better than Falcon. If they succeed, no doubt this boosts Elon's standing with the government and military again, because that'll make China the second power with reusable rocketry, and we can't allow a reusable rocket gap, can we? And if Starship doesn't work, then the gap is extremely likely - China can weld steel cylinders at scale and mass produce engines like nobody's business, like look at their shipbuilding or the recent pace of fighter jet delivery (they make ≈100 J-20s per year now, which above the total F-35 program output in 2024, though 2025 was a big year for LM with 191; and recall that J-20 is a massive twin-engine). They have something like 20 private companies competing for the launch provider market. On the state side, CASC's CZ-10B likely does its own launch and barge landing (very interesting mechanism by the way, initially explored by the US, abandoned) this week. CASC has a whole family of partially reusable Falcon-esque rockets in the pipeline (10A, 12A, 12B, maybe 8) and a very Starship-like superheavy CZ-9. They even have plans for space-based solar and compute. Regardless of how all this goes (I'm personally bearish on Chinese rocketry aside from LandSpace), it obviously bolsters Elon's narrative. In light of this, I don't even think the speculations about future Democratic hostility are convincing – the US has strong bipartisan support for any anti-China and arms-race-with-China initiative; Biden tightened the screws of Trump-1's trade war, Trump-2 didn't touch Biden's export controls. So Starship will almost certainly keep being funded and the only thing that can kill it is physics.
In sum, I'd like you to spell out your bear case that survives these objections.
P.S. SpaceXAI (what a lousy name) has just released a frontier LLM, I can vouch for it being genuinely on the same tier as Anthropic/OpenAI's latest (Fable/5.6 excluded), and with Chinese open source costs. Elon: «Grok groks engineering. Next month’s release will be another step-change improvement, as we close the loop on solving real-world engineering problems at Tesla, SpaceX, Neuralink and Boring Company.»
I have seen enough of his empty promises, but it does feel qualitatively different, an unexpected closing of the gap. He's still got it.
P.P.S. China Academy of Launch Vehicle Technology's Long March 10B's first stage has just been recovered by their clever net capture barge, making China the second nation with reusable rocket technology. This, of course, is bullish for SpaceX.
The reason the bets are timed is that I wanted them to be resolvable within a reasonable timeline. I made the original bets 3 years ago, and 2 of the 3 users I made them with no longer seem to be posting here, so I think it's fair concern. I agree that it's possible for me to win them due to uninteresting factors, which is what I called a "technical" win in the top level post.
Closest I got was here. It's not a specific prediction about Starship, it's a general prediction based on the hype-cycle of his products / companies, and it boils down to:
This was about Tesla, but I get the feeling that Starship is SpaceX Roadster/Semi/Optimus, where Elon bit off more than he can chew. It's mostly based on instinct though, but in my defense, I've made a few long-shot predictions on my instinct, on this very forum, that turned out to be true.
Your specific arguments for why Starship can work all sound reasonable to me, but they don't sound different to me from arguments for why Cybertruck could be a good truck, why FSD could drive safer than human drivers, why optimus could be a great humanoid robot, etc. I'm not arguing for physical impossibility, I'm arguing against the "make insane marketing promises, and let the techies figure it out" management style.
Maybe. I know exactly the type of people you're talking about, and I admit I was influenced by them. On the other hand, my "I don't have MDS" argument is that I don't actually want to win these bets. I want to lose them, and lose spectacularly. A world where I get btfo'd is by far better than the one where I win, and the reason I'm betting the way I'm betting is because it sounds too good to be true.
Didn't he buy Cursor, and these guys were the ones who figured it out? It certainly shows a lot of political / business acumen, but I didn't get the impression that that's the sort of "it" he's supposed to have.
I don't challenge your reasoning for making resolvable bets, my problem is that they don't have much relation to the interesting question. This is the usual forecasting problem.
All of these efforts being meh (so far) is not very informative. Cybertruck is just a goofy car, there's only so much you can achieve by making a big electric pickup with edgy body panels. It adds very little to Tesla's current position. Car people are somewhat insane in paying so much attention to car models. FSD works, Waymo is reportedly great, so Tesla robotaxi also could work. Optimus is a legitimately good robot, it's just not the time for robots yet, and China is way ahead of Elon on the entire robot supply chain except high-end chips (both the brain and external compute). Starship is a categorical breakthrough in space logistics, which is the one area where Elon is far ahead of the competition already. There is no way for others to overtake him on any reasonable timeline.
He did, and no doubt their data has contributed a lot. But the base model is in-house, and I see that RL was done on xAI's stack. This is impressive because the original xAI team has completely fallen apart, there was the impression that xAI has become a mere compute provider for Anthropic. He has crashed and rebuilt a near-frontier lab from the ruins. This suggests at the very least good capability for delegation outside the hardware domain.
In absurdly compressed timeline. Something like 3 months. Even with homegrown meth labs I can't see how they can deliver in such a short time.
More options
Context Copy link
I just wanted a way to get out of the "would the Enterprise win against a Star Destroyer" type nerd slapfights. I agree it was far from optimal, but it was the only thing I could come up with, and no one actually stopped to raise that particular objection at the time. The bet-takers thought Elon is great, so he will surely crack this trivial problem in no-time, which was part of the problem I was meaning to highlight.
Well, hold on. Waymo has a very different approach to Tesla. It's armed to the teeth with sensors - cameras, radar, lidar, audio... the works. I think they even do a high-definition scan of the cities they deploy in. Elon claimed he could crack it with cameras and AI alone. Using one as an example for why the other could work seems wrong. What's more, he was constantly promising that all the hardware for a better-than-human FSD is already inside each and every Tesla, and it's just a question of working out the kinks in the software. That soon (next year, next year, next year... no matter which year we were currently in), at the press of a button, every Tesla would become fully autonomous. Recently they said some of the older hardware might not be enough, and that they will need to upgrade it, and more recently still they gave up on even that idea. People paid a pretty steep price for a feature that never arrived.
Did they make some great leap recently that I'm not aware of? I think every other robotics company I've seen came up with something more impressive. It even looks like they're remotely controlled during their demos.
Well, but if it's not time for robots, why is he acting like this is the very thing that will take the entire company to an entirely new level? Transform human society, even. Shouldn't his car company focus on cars, and leave humanoid robots as a niche R&D project?
Maybe. Like I said your arguments for why it can work sound reasonable to me, but the problem is they sound similar to why FSD can work. I also saw what I think is a very similar next year / next year / next year dynamic with it. He used to do these semi-regular all-hands meeting at Starbase, where he'd talk about Starship (I think I linked to them in the old post I referred to in the previous comment), and from what I recall they started off with saying Starship V1 will be able to take 100t to orbit, a few years later that V1 could take (100 -X)t to orbit but V2 will be able to take 100t, and a few more years after that, that V1 could take (100 - X - Y)t to orbit, V2 (100 - X)t, but V3 will be able to take 100t.
Maybe they'll finally crack it, but it looks like "fake it 'till you make it" to me. I used to work for a guy like that, that would make insane promises to clients, and than expected me to deliver. If I have MDS, it might just be PTSD that Elon is triggering by reminding me of the experience.
About SpaceX being ahead of the competition: yes, but this used to also be the case for Tesla, and now BYD overtook it. There are other companies who are still in the nipping at the heels phase, but it's not obvious to me why it would stay this way given how difficult Staship development is proving to be, and how far they already got with their own rockets of similar class.
Optimus is great. Very good hands (some Chinese stole them), powerful, iterates quickly, and more importantly I can trust Elon to mass produce it, as he mass produces lots of things. Sure it's a bit quaint compared to the Chinese robotic supply chain and scaling potential of Unitree and UBTech and others. But regulatory barriers will all but ensure that Western markets heavily go to Elon.
The problem with Optimus and with Tesla taxis is the same: it's a bet on the exponential, and you don't know your exact location. Elon's theory of victory for FSD is that good enough AI will make do with human-level sensorium; arguments about lidars being expensive are of course nonsense, the costs of lidars can fall like costs of any other component. He's obviously correct on the substance; the question is what does it take for "good enough", how much more data, pretraining and onboard processing? He keeps discovering that the answer is "more than you have". But at some point, very likely it just works and Waymos become overengineered toys.
I can't muster the outrage. His corporate governance experiments have trivial explanations, and he'll have the cash to burn on it all.
I think you miss other variables changing.
The problem is they haven't gotten far. If rocket reliability requires exploiting Wright's law, Elon is very much ahead.
Well, at least as far as Starship is concerned, neither has SpaceX.
If. All the Falcons they produced didn't seem to help them get to a running start with Starship.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
I disagree, that is the primary "it" that he has, and the most important one by far. It doesn't really matter how elegant your product is, how technologically advanced, how innovative. The ability to create a solid business plan and actually execute it (eventually) is the critical factor. SpaceX is the best example: none of the technologies involved are especially novel: keralox rocket engines, aluminum alloy rocket bodies, carbon fiber fairings, etc. The actual innovation of SpaceX was figuring out a way to build and operate these things in a way that made them profitable at scale. This is an unpopular view, I know, due to the implications about whig history, but is something I think is true of all the great industrialists.
Elon himself is very well aware of this. I sometimes say that he's spiritually Chinese, which is lost on his greatest admirers in the West, who imagine his success is due to some brilliant insights. No, it's similar to what guys like Lei Jun do, just with American capital scale and more chutzpah. It's maybe the most potent recipe there exists.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Is the point of datacenters in space to keep them out of Yudkowski's airstrike range? Or, with less cheek, to keep them away from regulator?
The bull case for data centers in space is that the bottlenecks for getting data centers constructed are going to be
a) permitting / regulation
b) power generation / storage
Data centers in space have the potential to solve both of these bottlenecks, in exchange for introducing the manyfold engineering challenges of getting them into space (which other posts have already described in great detail).
I'm not entirely sure I believe this thesis, but it's at least fairly plausible to me that the engineering challenges can be overcome while NIMBYism gets ever more dysfunctional and terrestial power generation remains a bottleneck.
It has nothing to do with airstrikes though; any serious nation already can or soon will be able to strike assets in LEO all the same, not to mention that if push comes to shove the owners and operators of data centers in space are very much going to be within the reach of nation states.
SSO, but you're still correct.
The bear case for data centers in space is that the engineering challenges need to be overcome in the right order. I know the anti-data-center craze is literally crazy right now, but earth-bound solar+battery-powered passively-cooled centers are probably no more likely to be strangled by red tape than orbital centers. The orbital centers may look cheaper in back-of-napkin calculations right now, but only because battery prices haven't yet crashed as far as solar panel prices have and chip prices are so high that you want to run everything on a 100% duty cycle. If battery supply improves enough, or chip supply does (or if chip demand falls), the numbers change.
Solar prices are already creeping up as China has ended subsidies and the global demand is surging. The Chinese will do all they can but at the end of the day PV panels have scarce physical inputs (like silver). The lowest realistic price for batteries that I've seen was something like $15/kWh (if Sodium-Ion works out at scale). That's about $300000 for year-round battery+solar 1MWh supply (given seasonality and losses), probably more. Plus immensely more costly solar installation (lower area efficiency, overbuilding due to day-night and seasonal cycles, weather protection, land)…
Might as well just yeet it into orbit.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
In just a few short years, SpaceX went from "it's impossible to land an orbital rocket booster, and SpaceX's attempts keep exploding" to "here's a hilarious video making fun of all the hardware we blew up developing our incredibly successful rocket booster recovery system."
In just a few short years, SpaceX went from "an LEO constellation with thousands of satellites is insane" to being the most lucrative internet service provider in the world.
I have very little doubt that SpaceX is capable of turning "Starship is impossible and keeps exploding" into "here's a hilarious video making fun of all the hardware we blew up developing the most capable orbital launch system ever flown."
I kinda want to buy a piece of SpaceX because I think it's cool, but I'm also guessing that there really is a huge bubble right now and I'd be stupid to buy in.
Man, I wish I had the slightest idea about investments.
I also feel like SpaceX is a bubble, but I also thought that Bitcoin was a bubble.
More options
Context Copy link
It's a huge gamble. In 20 years everyone will either agree that it was a huge bubble or that it was a huge opportunity, but without hindsight it's really a huge gamble. I would have been thrilled to buy in during most of their previous private fundraising, but by IPO time the huge upside possibilities they have were clear to everyone, and IMHO that upside is now more than adequately priced in. I'm not planning to directly buy any shares unless there's a big unwarranted dip, and even then my motivation would likely be "I want to help ensure their employees' options are worth what they deserve" more than "I'm using the Kelly criterion to maximize expected utility of my portfolio".
Lesson 1 is very short and yet very effective: "Diversify". Get a NASDAQ-tracking index fund, and you'll have a small piece of SpaceX, but you'll also have enough other weakly-correlated-with-SpaceX stocks that you won't have to worry about losing big if things go bad for SpaceX in particular, just about downturns in tech as a whole or the stock market as a whole.
There's something to be said for the fun of wild gambles over sensible investments, of course. I made my first ETrade account around age 20, picked two stocks, and was so excited by the one that quadrupled that I didn't feel too bad when the other went bankrupt. (And even that was a mini-lesson in diversifying! Imagine if I'd bought the same two in serial rather than in parallel!) Just remember that high-variance gambles, even positive-expected-value ones, are the sort of thing you want to do with disposable income when you're 20, not with base retirement savings or the kids' college funds when you're 40.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
You won't see this because you've blocked me, but this isn't true and several people have tried to do the math. Here's the most recent. To sum up, cooling even a 1MW space datacenter (tiny by terrestrial standards) would require a radiator of 2000 square meters. The article doesn't discuss solar panel sizing, but using star cloud's numbers of 400 W/sqm we're talking about 2500 sq m before we consider redundancy. And that's just power and cooling for a single MW. In fact, it seems to me that it's the space DC boosters who refuse to engage on this and show their work.
This is an interesting report but I want to highlight one section in particular:
We're not really debating if space compute is feasible, we're just debating if it will scale. Given that it's technically possible and useful in niche functions, small space datacenters are probably going to be made. Whether bigger ones will follow is an open question.
Indeed. It's quite clear that you can compute in space. My contention is that it will not be cheaper than terrestrial computing (contra Elon), large satellites (100+ MW range) will be infeasible (given the technology under discussion), and small satellites will not be useful.
More options
Context Copy link
More options
Context Copy link
What I want to know about space datacenters is what's the plan for repairs and maintenance. Or is the idea: "YOLO, just deorbit it, and launch a new one"?
Gosh, I hope so. Imagine if sending objects into orbit became so cheap that just launching a new datacenter took roughly the equivalent amount of money and effort as shutting down a terrestrial server, fixing/replacing the broken part, and then turning it back on. By that point, the Futurama joke about landing on the Moon in less time than it takes to count down from 10 could be real. But that was the year 3,000, which still leaves a large range of time between now and then when rocketry will get that good.
More options
Context Copy link
For small-scale nodes, i.e., AI1-sized, I suspect that the node would just either operate in reduced capacity or be retired. They probably would not be designed for serviceability. On the flip side, properly engineered, there would be few moving parts, and low risk of environmental damage. Something like 99.8% of Starlink satellites are operational with a median age of a little over 5 years, so it doesn't seem implausible on first impression that AI satellites would experience similarly low rates of failure.
I dunno, man. Space Routers sounds a lot more simple than Space Datacenters...
They're almost entirely microchips in either case. The main complication is the need for circulating coolant through passive panels, but this is not an exotic field of space engineering at all.
Yeah, but I can hook up my router and let it accumulate dust until approximately forever (I never had one break), my desktops were also quite reliable, but I sometimes had to get a new part. The SysAdmin guys taking care of the servers at the companies I worked for, OTOH, were always running around and tinkering with shit. Pointless busywork? Upgrades that aren't going to be a part of equation here (but if so, isn't that a pretty big downside for putting these things in space?)?
Heh, this is my day job. Most of the time, the tinkering is replacing very old stuff with less old (or, ideally) new stuff. Sometimes stuff that was deliberately underprovisioned for business reasons that has to be upgraded later. And if it's a business office, physical networking often needs to change to suit the needs of the office workers. But often as not the old stuff coming out has been operating continuously for many years on end with ~0 maintenance and still works. I routinely pick up enterprise gear from work for my home that was retired and removed in perfect working order, but is no longer supported, surplus to requirements, or replaced with something more capable and more efficient. Usually if there is a failure, it's a spinning hard disk or a cooling fan; eliminate those and enterprise gear is generally pretty bulletproof and service lifetimes of 10 years or more are not uncommon at all. You observed that desktop PC hardware is already fairly reliable, and that's after having every cost cut to the absolute bone. Enterprise gear largely avoids those cost tradeoffs for reliability.
True that upgrades in the satellite model are precluded, but after almost a decade of service they probably wouldn't be upgrading in any case. In an industrial datacenter, there is a lot of infrastructure in the form of buildings, facilities, and power distribution that makes ripping the racks and replacing a sensible "upgrade" path. In the satellite model, what infrastructure there is, is largely degradable (solar panels and mechanical components), so there's not much benefit to upgrading them. Additionally, since each node is self-sufficient, its entire lifespan can be monetized without sacrificing efficiency, at decreasing revenue rates over time, unlike in a data center where there's a constant need to cycle in new hardware as soon as possible to maximize electrical efficiency. Compute per watt efficiency just matters a lot less when your electrical cost is 0.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Serious question, do you think that all of these engineering organizations who are going all in on this concept haven't done the math themselves?
I was skeptical about the cooling issue myself but I did the math myself and it turns out that if you're willing to run your chips a little hot you can get away with less radiator surface area than solar panel surface area. It's simply not the issue that people seem to think it is.
The math not mathing hasn't stopped repeated attempts at solar roadways (and solar railways, etc.). Sure, you can technically cover the road with solar panels and get some amount of power out of them, but for far less money you can get far more power by just building them in an empty field or the desert etc.
Similarly, you technically can run a datacenter in space, but it gives you effectively zero advantage over a terrestrial one (or even exotic ideas like building them out on the ocean or in Antarctica) while adding tremendous amounts of cost and complexity. I mean, I guess you get twice the amount of power per square meter of solar panels, at only an order of magnitude or two higher delivery and installation costs. And you don't really escape terrestrial legal jurisdictions any better than you would have by building the DC on a container ship out in the Pacific or something.
It's a retarded idea being pushed because it hypes normies and other retards (like venture capitalists).
Zero serious engineering organizations were involved in any of the various "solar freakin roadways" proposals. I give a lot more credence to SpaceX, who have a decades-long history of actually delivering on serious technical challenges previously considered outlandish under conventional wisdom.
In both of these places you are adding bandwidth/latency challenges, and Antarctica is forbidden from commercial exploitation.
I absolutely believe SpaceX can make a space datacenter work. I'm questioning the practicality and economics of the entire idea, not whether the engineering needed for it is possible.
More options
Context Copy link
What do you mean by "serious engineering organizations"? There were several of those built in Europe (Germany, France, and the Netherlands, off the top of my head), and they were, predictably, all boondoggles, but I assume they were built by "serious engineering organizations". Now, it was all most likely corrupt political deal-making, but someone "serious" put their name on it.
I mean organizations with a track record of delivering on ambitious projects.
Does the Colas Group count?
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Antarctica is closed to economic exploitation by international treaty. It's the complete opposite of escaping terrestrial legal jurisdiction. As for oceanic datacenters, they have a lot of technical disadvantages compared to orbital components. Large vessels would be preferable for stability, security, and navigational control, but power generation becomes impractical unless you permit commercial maritime nuclear reactors, which seems unlikely, or plan to have LNG refueling tankers visit every few weeks, which is expensive. A constellation of small solar powered vessels scales a lot less conveniently in the ocean, given communications constraints from the ground, security difficulties, and the scale of ordinary maritime maintenance that is necessary.
So, I think you are wrong - there are other advantages. In orbit, security is a non-issue, environmental degradation is minimal, solar power is abundant, communications are easier, and in general the floor cost per node is lower, meaning that scaling down incurs fewer penalties.
Space is also covered by various international treaties, and nations on Earth would be quick to update any such treaties to cover datacenters even more strictly if companies were using such datacenters to flaunt the law of countries. And at least in their current state, these treaties basically make satellite operators subject to the laws of their host nation.
Long distance oceanic LNG shipping is about 3 to 5 cents per KG. I imagine it would be cheaper when you're just shipping out to international waters from the coast. Starship's most optimistic projections for price per kg to low Earth orbit is a little under $100 per kg, but could end up closer to $1,000 per kg.
As opposed to the cost of sending people into space to fix the space datacenter? Most space DC proposals I've heard have actually proposed not having any human maintenance at all because of how expensive it would be, instead opting to add extra redundancy for essential components and just writing it off when a GPU or PSU fails.
Please, please, please show me your math for this. Even use the $100 per kg to LEO price if you want to make itas favorable to your argument as possible. I would be willing to bet you $100 donated to the charity of the winner's choice if you math shows it being cheaper than a container ship (or oil platform type structure) 370 KM off the coast (i.e. in international waters) of some LNG processing hub.
People didn't do some very basic pricing math with Solar Roadways and similar grifts. Solar Roadways would cost a metric ton more per km than just using asphalt, and a metric ton more that just sticking solar panels in a field or the desert or on a rooftop somewhere etc., while being a worse road surface and producing less power.
Space datacenters face similar economic disadvantages, and none of the proponents seem to be saying anything about the financial math here. I personally find the discussion of stuff like "How are you going to cool it?" irrelevant and a distraction, except inasmuch as they affect the cost. Cooling the datacenters is absolutely feasible, but it definitely complicates the engineering (and drastically increases the amount of material that has to be launched) far above and beyond what would be needed for a nomal terrestrial datacenter or some of the other exotic options I mentioned.
First, you have to determine what is a minimally-viable node in each context. SpaceX is proposing essentially a single-rack node with 120KW power at a mass of about 2 tons. Let's assume that the same Nvidia racks would be used in an oceanic platform, so we can disregard the silicon costs. Starlink satellites are constructed at scale at a cost of about $1M/ton, so a reasonable cost estimate for Starmind satellites is about about $2M per satellite. Add $200K in launch costs and we're about $2.5M/node up front, with ~0 ongoing costs.
If we assume that solar arrays are impractical for oceanic data processing, the minimum viable node would have to be some kind of hull with active station-keeping and enough fuel storage to fuel a diesel generator and station-keeping for extended periods between refueling (30 days?). It starts getting sketchy here, but working with requirements of about 25 tons fuel capacity, it seems like you're looking at a 30-40 meter DP1 vessel. I couldn't find costs for new construction, but listings for similar class vessels decades old are around $3M (e.g., https://maritimesales.com/DAB17.htm), so that seems like a reasonable conservative estimate. And this the up-front cost only. Assuming it's autonomous, it will still need monthly fuel deliveries, regular PMCS and overhauls on engines, gensets, and thrusters, other seaworthiness maintenance like painting, cleaning, and lubrication, and you can expect substantial wear and tear and damage from environmental forces. Fuel replenishment alone is going to be at least $30K/month. And this is all for a single Nvidia rack!
Now of course as you start scaling up, the economics shift, but my point was that one of the advantages of orbital deployment is the ability to scale node sizes down.
If you're scaling things down to a single rack, then for $2.5 million I'll gladly stick it in my basement (which has gig fiber internet, I'll upgrade to the 2.5 gig plan if they'd prefer for that kind of money) and handle all maintenance for them.
If the point of space datacenters is being able to do them at a very small scale, there are a million better and cheaper options that don't run into the same sort of political/NIMBY resistance that big datacenters have. There's plenty of vacant office buildings with good internet and electrical hookups that would be far cheaper and easier to maintain than chucking a rack into orbit.
The economics of this literally make no sense, there's no point in doing this stuff at a small scale because you lose all the benefits of economies of scale.
More options
Context Copy link
More options
Context Copy link
Space datacenters don't have to be cheaper than ground datacenters, they just have to be cheaper for processing space data than ground datacenters. The latency and bandwidth of space-to-grond will make space datacenters naturally attractive as space scales.
That's maybe a trivial compared to the actual core question, which is whether space datacenters can make sense for processing ground data. I think that's an open question. But the economics will become known as space datacenters are built to process space data.
What are you smoking? Firstly, there's hardly any space data other than scientific probes by various space agencies, and communication satellites operated by both public and private entities. The latter, by their very nature, need to communicate with the ground.
Secondly, the latency difference would depend largely on where both the datacenter satellite and the source of the space data are located. In LEO, space datacenters would be less consistent in their latency than ground stations because they're orbiting the Earth roughly every 90 minutes (or you'd have to be constantly passing the data around to different space datacenters to keep latency somewhat consistent, but this data transfer would likely kill any gain you got from the latency reduction and then some).
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
It's puzzling to respond to a report with numbers worked out with an appeal to authority and a "it's not an issue". Puzzling, but perhaps typical of this discourse.
Please show the math.
SpaceX gives values of 150kW of compute power and 110 m^2 of radiator panels (220 m^2 of total radiating surface area since they are double sided).
A typical radiator has an emissivity of about 0.95.
So, per Stefan-boltzmann:
150000 = 220 * 0.95 * 5.67e-8 * T^4
T^4 = 12.6579e9
T = 335K or 61.85 C
If you run your chips at 80C (not unreasonable!), that's a 20 K delta from the chip to the radiator, which seems very reasonable to me? My gaming PC does about 40C from my chips to ambient and I'd expect SpaceX to be able to engineer a better thermal solution than me.
Of course you can make the radiator smaller by making the satellite smaller. But you lose any economies of scale by having a big cluster of compute, which is presumably why Starcloud is targeting a massive DC.
My point is not that you cannot run a computer in space, obviously. My point is that small DCs are unlikely to be useful and large DCs are unlikely to be feasible.
It doesn't matter much for inference, which is what the satellites are for. Heck, you can run inference at home on your PC if you want to fork out for the hardware.
Indeed, you can run inference on the ground. Exactly my point.
More options
Context Copy link
More options
Context Copy link
I think I get it now
You think a "space datacenter" is a monolithic piece of hardware, a single satellite bus that's carrying 100% of the compute that's intended to be operated in orbit
You do understand that they're going to launch more than one of these, right?
I am indeed aware of that. There's advantages to having the compute collocated, which is why StarCloud is doing it that way.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Numbers presented without showing the math can be dismissed just as readily.
That said assuming you construct the radiator as a "fin" so both sides are exposed to space you would need a 50m x 20m structure or maybe 5 10m x 20m structures mounted along a truss to ensure 2000 square meters of radiating surface. That actually sounds pretty workable.
For reference the primary solar panels on the ISS are 36m x 12m a piece and there are 8 of them.
As explained, it's a simple application of the Stefan-Boltzmann law.
2000 sq m is for a small DC. StarCloud proposes a radiator of nearly 8 sq km.
Show your math then.
I did.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
I've unblocked you just to respond to this, though I don't remember why I blocked you in the first place.
In short, even if we assume 2000 square meters, this is nothing. Ascend 950 SuperPod has an area of 1000 square meters for the actual scale-up compute unit that works as one GPU.
«Redwire Q-Rad Deployable Radiator (commercial, TRL 5-6): 3.5–4.9 kg/m² areal density. Source: Redwire radiator datasheet lists; brackets our StarThink V1 assumption as plausible near-term path.» Let's say 4. This is just 8 fucking tons. This is peanuts. At $100/kg it'd be merely $800000 for delivery, 8% of Starship capacity, or four more big Starlinks. 1 megawatt of compute costs… let's see, a modern-ish GPU that draws 1 kW in a rack can go for $20K at least, and actually we'll see prices creep towards $50K. Well there you have it, $20 million as the floor (and GPUs are just ≈40% of BOM). 1 megawatt is 8760 MWh/year. Google tells me wholesale electricity in the US is like $45. Almost $400K a year of free power. None of this matters of course, when inference margins are >80% even with hardware depreciation, and all that matters is deploying as fast as possible, as much as possible.
You don't know the relevant numbers in any of the involved verticals, and for some reason (unfathomable to me) you want to believe that the numbers support your (quixotic but perplexingly popular) case against compute in space. They don't.
Uh, okay. Manhattan has an area of 20 square miles. What's the relevance to space radiators?
Hmm. What is StarThink?
I think it's quite clear that mass is not the concern I raised wrt the size of the radiators, which is why my post did not say anything about mass. Huge radiators cause drag and are vulnerable to micrometeorites. Redundant cooling loops can mitigate micrometeorites, but then you need bigger radiators.
But since you brought it up - the radiator you posted is solid state, SpaceX says they will use circulating liquid radiators, same for starcloud. I expect this is because circulating liquid radiators scale more than passive radiators like the one you are looking at. Just a guess on my part, though.
Now, StarCloud says each data center will be 5GW. They are cagey about how exactly they will dissipate 5GW, but with a simple water/glycol loop we are looking at 50,000 kg/s of flow which is massive. All those pumps will of course also need to be cooled.
On top of that, the radiators need to be filled with coolant. Back of the envelope math, again for a simple coolant loop, suggests perhaps 100,000 tons of coolant in the radiator, so we're looking at $10B just to get the coolant up there across hundreds of launches and nothing else. They do not go into the math of this in their whitepaper, probably because single phase liquid coolant is totally infeasible.
Tedious bluster. Please save it for Twitter.
It matters because it deflates the context-free appeal to "omg 2000 square meters". Ok. 2000 square meters is just 40 * 50 meters. Is this supposed to be a lot?
It's not clear, because mass is the only interesting concern there is. Drag, too, is an issue of mass (for ion thruster fuel). Your link says: "And that’s the best-case scenario. Additional problems are hidden in the low Earth orbit environment itself. Space exposes radiators and their coatings to a chemically hostile brew of ultraviolet light and atomic oxygen, quite the opposite of a clean-room environment. Over a LEO satellite’s typical 5-year lifespan, these elements degrade the radiator’s surface properties and lower its ability to shed heat. … Including this degradation in the model reveals that as the radiator degrades from a “fresh” state to an “end-of-life” state, the physics demands a further penalty. To maintain that same 60 °C operating temperature for the GPU chips, the required surface area jumps from about 1.4 square meters per chip to nearly 2.0 square meters. In other words, the physics tax rises by 40 percent. Therefore, you must launch at least 40 percent more radiator mass, endure higher atmospheric drag, and sacrifice valuable launch volume just to survive the degradation of the thermal coating."
Or you can simply launch a little higher. No matter how you cut it, it's all ultimately about mass.
Huge solar panels cause drag and are also vulnerable to micrometeorites. This issue, just like the radiator issue, is negligible. Neither solar panels nor radiators lose function quickly from random point damage. At, say, 500 km the lifetime of an inference node with several thousand square meters of total area can be a decade. How much economic value does a decade of compute with free power provide? That, cost per kilogram to orbit, and costs of hardware are all that matters.
Since you dislike X, I'll cite it again. NVIDIA CEO JENSEN HUANG: 1GW AI FACTORY ON NVIDIA ARCHITECTURE COULD COST NEARLY $100 BILLION
So, maybe $450B for that 5 GW you talked about. Cooling alone is likely a fifth of that. I guess popular reporting can create the impression that Americans are actually standing up tens of gigawatts of capacity without problem, like so much coal plants in China. This is not, in fact, happening. Most Blackwell compute is still not operational. All this space math only matters in relations to costs on Earth.
I hope I didn't mislead anyone into believing that 2000 square meters is a megastructure. Nevertheless, most people have never seen an Ascend 950 so I don't think that helps contextualize anything for anyone. 2000 sq m is fairly large for a space radiator - the ISS has only about 400 sq m.
Perhaps. And yet, Starcloud plans to operate in LEO. I assume they aren't totally retarded and have thought through the choice of orbit. It's difficult to have a discussion about this when you ignore the details from the actual proposals in favor of advocating for stuff they aren't doing when it's convenient. Either the people working on this are smart and have chosen the best parameters for this, or they are stupid to the point that the internet peanut gallery can do better and therefore aren't going to succeed. You must pick one.
If you're pumping coolant through a tube that's open to vacuum, you're going to have some problems.
I don't really understand what drives a man to repost second hand all caps claims. I'm not even saying that he didn't say this, but surely you must understand that this is simply not convincing to anyone?
Space based DCs also fail the "not currently happening" test, so this part is a wash.
You know what, fair enough. Let's ignore Starcloud since this is primarily about SpaceX. They've just issued a concrete design: Starmind
• 150 kW peak compute payload
• 120 kW average compute payload
• 70 kW per ton
• Wingspan: 70 meters
• Deployed height: 20 meters
• 110 m² deployable liquid radiator
• Redundant pumping loops
• Integrated micrometeoroid shielding
• 150 kW solar array
• 250 W/m²
• High-speed laser links interconnect satellites and beam AI results back to Earth through Starlink. Low-latency, high-bandwidth connection
• SpaceX-manufactured solar technology from Bastrop, Texas
So, that's 917 square meters of radiator per 1 (sustained) megawatt, and more importantly 70 kW of capacity per ton, at SSO. I see Starship has the theoretical capacity of 40-60 tons to SSO, let's say 50. At, say, $200/kg that's $8M to deliver 2.8 MW of compute. As per Jensen, 1 MW can go for $100M. There's plenty of slack in this. Even if Jensen is off by an order of magnitude, the "getting it into space" part is almost a rounding error and can make straightforward sense given terrestial/political constraints.
More options
Context Copy link
I was too lazy to de-caps it, and I hope that people of this forum will find the issue of the costs of 1 gigawatt of capacity on Earth more salient than the funny detail about all caps.
Problems of space compute have straightforward engineering solutions, the costs of which can be estimated. Whether these solutions are worth the cost depends on the costs of building the same capacity on Earth. So arguments about radiator area, micrometeorite damage or coolant mass are kind of… weightless unless grounded in comparison to the baseline.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
This. Basically, space is the last place you want to put your data center. Putting your computers basically anywhere else, be it in high altitude balloons, the summit of Mt Everest, the Mariana Trench, Point Nemo, downtown Manhattan, on harnesses worn by stray cats, the surface of the Moon, the rectal cavities of cybertruck drivers, Antarctica, Gaza (to just brainstorm a few not-so-good ideas) is going to be much less of a hassle than LEO.
While solar power is plentiful in space, computing turns the energy consumed into heat, and radiative cooling is not very efficient, especially if you want your chips to run at 400K and not 4000K.
There are also other minor objections (e.g. if satellite data links would scale to backbone ranges, we would not rely on expensive undersea cables instead, how do you service your equipment? and Kessler syndrome makes you extremely vulnerable to sabotage), but cooling is the big one.
It is not that computing in space is impossible per se (every cubesat does some, after all), it is just that it is extremely painful compared to computing dirtside.
As an analogy, there is no reason why sex in the vacuum of space should be impossible, one could certainly design pressurized space suits which have docking ports in the correct places. It is just that we already have much more convenient places to have sex, including space stations, beds, parachute jumps, submarines, mini-golf prop houses, presidential offices, fields of nettles, BDSM dungeons and many more. Until we saturate these environments, there will be little economic demand for space suit sex (beyond the novelty value).
I busted out one of my old textbooks and found the equation to compute the area required to radiate a given amount of energy in a vacuum. "A = W / (E * SB * T ^ 4)" where A is the radiator area in square meters, W is the wattage, E is a value between 0 and 1 representing the efficiency of your radiator, "SB" is the Stefan–Boltzmann constant, and "T" is your operating temperature in Kelvin. So lets do some back of the napkin math...
A = 1,000,000 / (0.25 * 5.67e-8 * 400^4)
A = 1,000,000 / (0.25 * 0.0000000567 * 25,600,000,000)
A = 1,000,000 / (0.25 * 1,451.52)
A = 1,000,000 / 362.88
A = 2,755.73 square meters.
Per my reply to @sarker above I actually think that sounds pretty doable. For reference the ISS has just over 2,500 square meters of solar panels.
I am not saying that it can not be done. I am just saying that it does not seem cost-effective. The ISS costs 100G$ for development and running over a decade, so lets call it 10G$/year. Of course, the solar panels (a modest 120kW) are likely not the most expensive part, and mass production would drive the costs further down.
Still, back dirtside I can get a 1MW peak solar plant for less than 1M$ (to generously allow for the lower efficiency as compared to LEO; excluding land prices), which is possibly less than the ISS spent on space-certified bolts to mount their solar arrays.
As a further complication, transporting electric energy is a lot easier than transporting heat. If your compute is all in one place, you will want convection cooling, which means pumping some fluid to heat exchangers. I am not sure what the ideal fluid for space cooling is, actually. With water, you would have to build pipes to handle the vapor pressure of about one atmosphere, which will likely be heavy. And if space junk punctures your heat exchanger (which is a concern with 50mx50m panels), that will quickly lead to a loss of operating fluid for that loop.
As an alternative, you could spread out your electronic components evenly over the area of your radiator. However, your H200 (TDP 600-700W) will take about two square meters of radiator for cooling, so you will want at least heat pipes instead of relying just on conduction.
Or you could double down on fluid pumping and use a heat pump, so you can run your radiator at higher temperatures than your electronics. The coefficient of performance for cooling is T_C/(T_H-T_C), so if you want to run your radiator at 800K, you will need as much energy for your compressor as for your electronics, for an 8x reduction in required radiator area. Most refrigerants have a critical temperature (beyond which the refrigeration cycle does not work) lower than 800K, R-110 comes close with a critical temperature of 700K. Of course, the critical pressure is 39 atmospheres, so you would require massive pipes per Barlow's formula.
As a kicker, one 1MW-facility would cover 1/50000th of humanity's data center needs.
Google's AI claims that electricity costs are about 10-20% of the TCO of a data center (and only 60% of the operating expenses). This means that even if Musk shipped your solar panels to LEO free of charge (or even provided them for free altogether), all the hassle with radiators and comms and lack of equipment replacement options means that it would very likely not be worthwhile.
More options
Context Copy link
Note that your value for Emissivity is very low. Radiators with emissivity around 0.95 are common, which cuts your radiator area nearly 4 times, and the total compute power per satellite given by SpaceX is about 150 kW, which further reduces radiator area by about 6.6 times.
More options
Context Copy link
More options
Context Copy link
OK Bill Clinton.
I tend to agree that building compute in places where the most efficient electricity option is hamsters in wheels is probably an easier engineering challenge than cooling datacenters in space.
More options
Context Copy link
I think you're greatly exaggerating. Deep ocean is a much more hostile and inaccessible location than LEO by almost any possible metric I can think of except, possibly, the energy cost of reaching it. The Moon is much further away, requires much more Δv, and isn't even sunny for half the time. Antarctica is extremely energy-poor and is unavailable for commercialization in any case.
At 400K, your panels should be able to reject over 1KW per m² to deep space, continuously. That's actually pretty efficient! You can do better with air cooling of course, so long as you don't care about environment heating at all, but that's also at some energy cost.
Dirtside computing can be infinitely painful, depending how uncooperative governments want to be with regulations and lawsuits. At least in LEO there's limited jurisdiction.
To be fair, LEO is also not always sunny. For a simple equatorial orbit, you are in the shadow of the Earth almost half of the time.
Of course, Sun-synchronous orbits (which use the precession due to the Earth's shape to adjust the orbit by about one degree a day, so you can always ride the terminator line) exist, but they also tend to have worse radiation exposure, and so far nvidia has not built a radiation-hard variant of the H200.
More options
Context Copy link
More options
Context Copy link
Why do you think you can just say this and not show the math for radiative cooling? The prose about stray cats and sex in a vacuum is cute, again, very Russian-engineer-coded, but the boring reality is that a Starlink satellite is substantially made of, well, chips, which do computations, and it dissipates just fine with a primitive one-sided radiator on the hull. How do you imagine anything ever works in space? How does ISS work? Do you believe that 20 kW is workable but 120 is where physical limits kick in? Care to show this? For example:
This is an engineering question. And your objection is the «Mars has radiation, bet you never thought about that eh» tier smug dismissal, it's plainly disrespectful and incurious. I suspect that you thought of that one too, well, I recommend to read on Suncatcher.
Other items are also trivial.
The fact is that
the US cannot compete with China on power generation in the medium term due to political schizophrenia, pathetic industrial base outside some bloated military supply chains and third world logisticsat sufficiently low cost per kilogram to orbit, yeeting inference nodes into one makes straightforward economic sense. Freed from gravity, atmosphere, moisture and hail hazard, solar panels become like 50 times more effective per unit of mass (likely more because you can move to lighter substrates). You don't need batteries with 24/7 noon. You don't need cabling. You don't even need a lot of structure.You have it entirely backwards. Having sex in spacesuits is what we have been doing all this time, running electronics in the wet dirt. Carbon life is made for Earth. Metals prefer the orbit and vacuum.
More options
Context Copy link
This is at least a solvable design problem, if not a trivial one --- terrestrial temperatures being generally comfortable compared to the extremes of hot and cold in space. It's a bit different for LEO because the Earth is a big object in view, but otherwise the Sun is hot and dark space is very cold.
The entire planet sits in (mostly) thermal balance between solar radiation, terrestrial energy, and radiative cooling to space. No particular reason a satellite can't do that too, although again not as trivially as "slap a heat sink and fan on it" that works down here.
More options
Context Copy link
More options
Context Copy link
I agree. I actually also agree with the main thrust of his post, but orbital datacenters make zero sense unless you’re wrongly thinking “space = cold” instead of “space = vacuum”. Power needs could theoretically be solved by using nuclear reactors instead of solar panels (which is still pretty impractical compared to just… building a reactor on Earth) but the absolutely ludicrous size of the necessary cooling radiators (and what happens if/when that radiator gets hit by a micrometeor or a piece of debris? how easy is it to repair? how long can you wait?) makes it a non-starter, barring some borderline-magitech advancement in cooling that would surely also make it easier to build on Earth. Cooling and especially power are the limiting factors of datacenter construction, above the raw land requirements.
Maybe there’s a future case for datacenters on the moon, using some sort of geothermal-esque cooling system with boreholes? I imagine the underground temperatures of the moon are pretty damn cold, I bet we could use it as a heat sink. But there a whole lot of steps to cover before there’s any benefit at all to doing that instead of just building a normal datacenter.
I think the only real economic case for what we’d recognize as sci-fi-level space development is mining, whether that’s helium on the Moon or rare earths from asteroids, etc. I think this would require launch economics to get vastly cheaper before anything could come of it, but it could potentially take off as both a sovereign and zero-pollution (on Earth anyway) means of acquiring certain resources. I think it’ll happen eventually. But not very soon. Near-future space development will be all about communications, GPS, and surveillance — perhaps with a bit of weaponization thrown in to deal with the surveillance.
I'm not convinced that the sizes are that "ludicrous", especially if Starship actually delivers on even a fraction of it's promise.
We're not talking about orbital megastructures here, we're talking about something on the scale of the ISS
More options
Context Copy link
Correct me if I'm wrong, but the radiator for a spaceship itself is basically just a big piece of metal, right? It's what the condensing stage of the cooling unit dumps heat into to reject it? As an HVAC tech I'll say it should work fine if it has a hole in it, just like plenty of condensers on earth work fine with hail damage on the fins or dirt on them. Not ideal, but fine.
Typically there is a heat exchanger on the exterior of the hull with coolant that is circulated between the interior components that need to be cooled and the exchanger. These are the matte white panels you see around the base of the service module on the Apollo and Soyuz capsules. On the Shuttle the heat exchangers were on the inside surface of the cargo bay doors which is why you never see the bay closed while in orbit. Meanwile most satellites as well as the ISS have radiator fins mounted perpendicular to thier solar arrays such that when the solar panels are oriented towards the sun the radiator will either be in shadow or edge-on
More options
Context Copy link
More options
Context Copy link
I am sincerely curious: are you a conspiracy theorist? Do you think Musk, Jensen Huang, Google and everyone else are in on the joke, just peddling a physically nonsensical project because they know that the target audience (VCs) has the intuitions of an illiterate Ghanaian child? Is this the great blessing of living in a nation with a perfected cognitive sort – almost everyone can be clueless, but anyone can make 6 figures?
But space really is cold, by the way. 2.7K. It's not like your Stanley "vacuum" that has room temperature. Radiative cooling doesn't work when the radiated heat radiates right back at you. You've never actually touched cold vacuum, and yes it is a meaningful notion. In the vacuum of space, you radiate and lose energy like a long-wave infrared heater, and very quickly die. The cartoons are correct on this account, they just conflate "vacuum of space" and "absence of air".
More options
Context Copy link
Well, space is a vacuum, yes, but the radiative heat sink is extremely cold. People have suggested testing cooling in a vacuum chamber to prove the infeasibility, but this misses the critical factor that the vacuum chamber walls are not ~2 K, and the efficacy of radiative cooling scales by the differences in temperature to the 4th power.
But, see, we've been to space, and had to cool shit down up there. It would seem like NASA has the table to know exactly how much radiator surface is needed for every heat load and the heatload is theoretically calculable based on existing chip design. I suspect that the datacenter inside the satellite would need to be redesigned down to a very small level due to the lack of a cooling medium, but 'how much radiator do we need to get rid of heat' doesn't seem like something we'd debate without really knowing the answer- it's not exactly the drake equation of building datacenters in space.
More options
Context Copy link
More options
Context Copy link
Have you actually done the math on this, or has someone else? My understanding is that it's totally doable with relatively modest radiators; I'm open to this guy not knowing what he's talking about, but all I've seen from the other side is sneering.
The bigger issue is probably that chips are designed to work immersed in air, which conducts heat away. In orbit, you'd have to either build a pressurized system or redesign the chips to have a different kind of active cooling. It's getting the heat to the radiator that would be the problem.
We already know how to fluid-cool chips and get the heat to a radiator.
Indeed, I've had a fully submerged mineral oil PC rig since before Covid.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
The economic case for asteroid mining is also dubious as far as I can tell.
Indeed, and I have seen people discuss that a niche for space datacenters might be processing satellite data up there rather than beaming it down (for latency and availability reasons). Perhaps, presumably you wouldn't AI-scale DCs for that so it might be more feasible.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link