The classic idea is to release them in the same orbit, but in the opposite direction. Any that hit will do so at orbital velocity, and as they will come around again and again they’ll have multiple chances to hit even with poor aim.
Right, but that takes a lot more energy, because you not only have to get the impactors up, but you have to get them to orbital speed.
It’d depend on how high up your target is. For low earth orbit, the orbital speed is the vast majority of the energy. Anyone who’s making any serious effort at rocketry at all can get the height. As you get higher, though, the potential energy rises and the kinetic energy at the top decreases. For a geosynch satellite, say, an opposing orbit might make sense (though would inflict a lot of collateral damage on everyone else’s satellites, too).
I was watching the news at the dentist office this week. They had a segment where a reporter visited one of Google’s data centers. The reporter had to scan her keycard 6 times and do a facial scan.
Well, usually in such a hypothetical the target is a military satellite, and in a war a tactic of “destroy all the assets you can” is pretty common. Especially when there’s nothing alive there to be killed.
Not sure why somebody would blow up an orbiting “data center”, but presumably their reasons would influence how much is the way of collateral damage they card about.
Destroy all of the enemy’s assets you can, sure. But preferably without getting every other technological nation in the world into the war against you, too. Plus, of course, probably some of your own assets are up there, too.
We had a similar thread in November when Google announced a moonshot project to explore datacenters in orbit:
Here is the announcement:
Which also has a link to their whitepaper that outlines the challenges (radiation, heat, cost, networking, etc.). It includes cost projections as well.
Google’s strategy here is to use a single project or goal to further development in multiple areas. Even if the goal ends up failing, the hope is their is collateral innovation.
In any case, datacenters in inaccessible places like space or the ocean will be running commodity HW for common use-cases. It could be tasked with batch processing corporate data overnight, for example. The cutting edge stuff would stay local-ish.
To be fair, a typical non-datacenter Google office has badge scans at most doors and elevators.
Are there office buildings that don’t require badge scans to go between areas? I thought that was standard in any business that needs information security.
Bumping this thread with a relevant article from IEEE Spectrum:
I skimmed the beginning of that article. The most interesting bit, I thought, was this, “At ABI Research, where I work as an aerospace analyst, we did a rough total-cost-of-ownership comparison between a data center on Earth and one in space. It showed that the cost to launch and run a GPU in space for a year is at least an order of magnitude higher than the same feat in a terrestrial data center.”
So what is the great advantage that has people like Elon Musk proposing such things?
The great advantage is that it gets people excited for Musk’s rocket company.
Hey, nobody said there had to be an advantage for anyone else.
The author is saying the TCO doesn’t make sense in 2027 while Google’s paper say they are targeting mid-2030’s and launch costs of $200/kg.
I wish he went into a little more detail on his numbers. His numbers are more optimistic than in Google’s paper, but his conclusions are more pessimistic. And it is not clear that he amortized the launch costs over the lifespan of the system.
The article is interesting though and goes into good detail on the cooling. It is a little contradictory though. It starts off saying “TCO doesn’t make sense, applications don’t make sense, and cooling won’t work”, but it ends up with “here are some applications that make sense and here are some cooling ideas that might work”.
My read on the statements from the article is that the benefits and efficiencies of what Google’s PR people are claiming right now is BS, but there are possible solutions that might make orbital datacenters work in the future. But that it’s going to require a lot more investment, developing technology and engineering work before any such claims will be more than PR buzz.
Or, of course, that the claimed benefits aren’t the actual reasons why they’re pushing it forward.
I have… opinions, but those aren’t suited for FQ.
Ars Technica is running a three article series:
In May:
Today:
And the third sometime in the future.
Other than the points brought up, part of it is the Silicon Valley mindset, i.e. propose something grandiose, get some investment, and set the engineers on it.
The assumption is that if you throw enough people at it, there will be a Eureka moment or a series of them that produce your original claims, or a reasonable facsimile thereof. Or not. Not every startup actually comes through with the breakthrough innovations.
Of course, as the editorial page of the same issue of Spectrum notes, Musk has made many extreme claims about developing other technologies that have not remotely come to pass. The current claim of cost effective space based AI data centers in three years appears to be of that mold.
As these other think pieces work out, even with optimistic projections, more modest projections that are at least arguably cost effective in some respect if you squint, are closer to a decade out, not 3 years. But the big money investments in space based data centers now seem possible, which is the important part and would have been laughed out of a room a few years ago. That seems to be important bit, rather than the practical ability to make it happen
Hehe, yeah. I do love how the basic gist of the second is “it’s complicated”. Not quite fanciful, but a lot of things are going to have to break the right way to make it anywhere near practical.
I think this is true of Musk’s 2-3 year claims, but Google has been pretty upfront in their announcement and paper that this is ~8 year project. It is more of an omnibus research project to try to tackle 5 or so technical challenges for high-performance computing in space. Most of their “moonshots” don’t work out, but they are still useful incubators.
The article lists a couple of applications for compute in space. There could be a lot of value in crunching big data streams in obit and sending the final analysis down to Earth.
Nobody wants a data center in their town; putting a chunk of them in orbit could be good for everyone.
So these “data centres” are supposedly able to run without fossil fuels and without consuming water, because neither is available in orbit.
What if someone took one of those satellites and just planted it on the ground? Give it 5x the solar panels, to compensate for the atmosphere. Use a fiber optic connection instead of line-of-sight lasers. Let the radiator radiate the heat away. Add an access hatch to let humans, without spacesuits, change the processor for a “better” one in 3 years.
Wouldn’t that work just as well as an actual satellite, at less cost? And wouldn’t it be more acceptable, socially, than one of the current terrestrial data centres powered by natural gas turbines and using groundwater for cooling?
Heh, but then you have to deal with those pesky local laws, which are inconvenient and icky.
But really, your idea has a lot of merit. I’m still not really sure what problem the idea of orbital data centers is supposed to solve, other than data centers aren’t sexy and launched on rockets.
I agree that Google’s timeframe, while IMHO still optimistic, is far more reasonable and grounded than Elon. Which probably surprises no one in the thread.
I do wonder how much of this urgency is the need to keep blowing the AI bubble bigger, or at least, stave off any possible collapse for the various big Tech companies.