You wouldn’t say untouchable if you’d seen me with a slingshot.

You could burn the ground stations down, but they're all around the world, and all it takes is 1 to route the data to via the inter satellite laser links. You'd definitely add latency though taking out ground stations near you.

To run one rack of GPUs they will be about 100 square meters of solar panels and 30 square meters of radiators. That's about 130 kilowatts of compute power. To get a gigawatt multiply that by 6,000. Musk is claiming he wants to send terawatts up to there.

Yeah it's tricky, given the pushback people are showing against datacenter construction, the issues around power, the vulnerability to attack... The space data center idea is starting to become more attractive to big tech over time, not less.

Initially, everyone agreed that the numbers just don't add up, there's no way it could be economical. But then these terrestrial datacenters started getting a lot of bad press, everyone was concerned about the environmental impact, about the strain on the energy grid, about water usage. All of a sudden the space plan is seeming less crazy.

and if they're in LEO, they'll fall back down to earth. Oh, and there's more likelyhood that bits get flipped from all of the high energy particles up there.

You're thinking about the right things here, these are major issues! But, they can largely avoid serious problems here. Over 500 km, there's a lot less drag and these aren't just going to fall out of the sky. And under 1000 km the earth's magnetic field still mostly protects them from high energy particles. They will still definitely be more exposed to high energy particles causing random bit flips, but it's going to be in a totally manageable range.

Giant AI datacenters will truly never be an efficient or wise use of resources, at least in my opinion, it's just a largely stupid endeavor. But for better or worse, I expect we will be seeing some of these space based data centers in the near future. I guess we'll just have to wait and see how economically sustainable they are...

Just FYI the rockets won't be expensive if this plan ends up working.

The only way this works beyond a tech demo, is if Starship works.

If Starship works as planned (big if), the marginal cost to orbit will round down to the cost of fuel to launch over the lifetime of the rocket.

Starship being fully reusable and working is going to be a mind boggling change in what access to space looks like and means.

you can only vent heat in infrared

Yeah, when you heat something, it emits infrared. Kind of the nature of hot things.

There's only a rack worth of compute on each satellite. There is nothing monstrous to deal with individually. The plans are to put up a very large quantity of these which will sum up to a large amount.

That probably makes training on them less effective, and that would likely remain a land based thing, but for inference, any given person probably doesn't need more than 1 rack of compute to themselves, and if they do, they'd probably be using something land based.

Begrudging upvote

no worries about the temperature if the ship collides with a spec of spacedust :)

Yes, but the infrared radiators for a single megawatt are apx the size of 4 tennis courts. Until there's a space elevator, that's a tremendous amount of resources to get into orbit.

Apparently there are a lot of concerns, of which, one is venting heat. Also the change of heat as a satellite dips in and out of sunlight.

I didn't think about radiation, which I should have since it's a not-small problem for human astronauts.

Heavy CMEs are going to be the AI equivalent of Purple Haze LSD.

With our current tech, a Mars mission pretty much 100% certain a one-way trip.

Common elon, wheres your "move fast and break stuff" spirit? Lets get you suited up and on a rocket asap!

I'm still waiting on how he's going to manage CMEs during spaceflight.

NASA timed its manned high-space missions to low-ebb periods, and there was still some risk of astronauts being baked to a crispy golden brown.

Getting to mars is nine months, rather than a few days to the moon. CMEs be coming.

Exactly. Once the data center in space idea inevitably flames out, he'll come up with some new bullshit "vision" that the rubes will lap up just as eagerly.

or failing to only have sons

Soo, he thougt his genes are "stronger" or something like it? This would be another failing to understand how things work.

Yep. Figuring out data centers in space is pretty much a prerequisite for any sort of significant transorbital flight. Unless we want to be limited to dinky little 2-3 person capsules that fit in a garage, efficient cooling in space will be necessary.

Hilarious prank though. Send him into space, and then "accidentally" forget some of the heatshield.

Just read a comment about exactly this a few days ago. Anyone working on space components know that a very hat problem to work on is how to keep things cool because, well - having the fucking sun shine on things most of the time directly without the ozone layer and air to dissipate heat in and way does make things go hot.

Also, you forgot a very important thing a data center needs: people that work there. That is one of the reasons we currently build data centers so close to urban sprawls in Arizona instead of somewhere in the middle of Alaska or in polar regions. There must be personnel working onsite all the time.

Also it will need repairs. And probably more often than not due to the radiation.

It's funny to me because the moon would be a much better choice, if latency wasn't an issue. Don't get me wrong, if your aim is uptime, reliability and ease of upgrades/use then terrestrial based is your only option.

But if your goal was off Earth, the moon has a number of advantages that free floating in orbit lacks. You can place the servers themselves underground in lavatubes or excavated caverns as well as install heat-sinks (or even seal a cavern, pressurize it and use what ever gaseous medium for extra heat capacity and thermal transfer) then place a series of solar on the surface with battery support (remembering that lunar nights are 2 weeks long) to generate all your power needs, as well as take advantage of the moon's tidally locked nature to have your transmission and receivers always pointed at the Earth, or relay satellites in orbit.

Such a project could coincide with NASA or other agencies or enterprises looking to establish settlements on the moon for research or mining purposes, so you could have short or long term settlements with technicians on hand to troubleshoot any problems that arise or even get semi regular (every 18-24 months) deliveries of components for repairs and upgrades for your system, something that long term would be a lot simpler logistically and more stable than attempting a satellite or constellation of satellites any Earth orbit.

The plan is nonsensical for a number of reasons, namely the cost and long term dedication, but it makes a hell of a lot more sense than trying to put an energy intensive project in orbit and hope the lowest bidder parts last more than 6 months.

People forget that black body radiation is to the 4th power of the temperature difference. I assure you if your server cluster is at 1000 Kelvin cooling will be the least of your worries

Well, you're thinking about the right things for sure! Power, cooling, Internet connection, all super important.

But as it turns out, terrestrial solar panels are actually heavier than space solar panels, like a lot heavier. Shielding against radiation is actually less of a challenge than shielding against weather and acorns. The ISS updated it's solar array a while back and its new ones were deployed by rolling them out, they were a thin enough film to actually roll up for launch and then roll out once attached to the ISS.

Also, these solar panels won't have blackouts. With a sun synchronous orbit, these panels simply won't be in shadow basically ever. The one exception would be during solar eclipses.

As for cooling, you do probably need radiators with active cooling systems. Generally you need radiator panels at a ratio of about 1/5 of the area of your solar panels, but this can vary with different designs. But that's a linear relationship, cooling doesn't become unfeasible at some point it just remains a design concern. The more power you generate, the more radiators you need. It's just part of the cost of power and it scales linearly with the area of your solar array.

Would the water need to be desalinated if it’s only going to be cooling the equipment?