Comment by nixonpjoshua
2 days ago
I am reminded of how space exploration has come largely before deep ocean exploration, seems like a human bias.
Putting data centers under water makes way way more sense than into space.
2 days ago
I am reminded of how space exploration has come largely before deep ocean exploration, seems like a human bias.
Putting data centers under water makes way way more sense than into space.
https://en.wikipedia.org/wiki/Project_Natick shut down in 2024 (though apparently it hadn't been in the water since 2020.) It seems like it basically worked, but it wasn't clear that the cooling advantage was all that big relative to the hassle of having them in a difficult-to-maintain environment.
> Putting data centers under water makes way way more sense than into space
You need permits underwater. You don’t in space.
The FCC regulates satellites launched from or communicating with the US, including stuff which extends beyond spectrum licensing like mandatory 5 year deorbiting capability for newly launched LEO satellites. Europe, China and India are not regulation-free utopias either.
You've actually got more option to jurisdiction-shop with underwater data, but I'm not convinced that's the major issue with building datacentres anyway.
Ultimately there are latency and minimise data-transfer arguments for doing certain types of data processing on local machines in space, but the generalised compute and model-training argument only works if the unit economics stack up as sufficiently good to cover the risk and R&D, and they're not obviously favourable compared with cold place on earth with clear skies and access to cold water even assuming launch costs become minimal. (It's slightly amusing to see how much some advocates of that other controversial futurist vision of spaced-based solar power - whose chances of success equally depend on low launch costs - viscerally hate the latest wave of datacentres-in-space hype...)
> FCC regulates satellites launched from or communicating with the US
FCC is easier to deal with than multiple layers of environmental, planning, power, and water concerns at the local, state and federal levels.
> they're not obviously favourable compared with cold place on earth with clear skies and access to cold water
There are fewer of those places that can be developed than there is space. The bottleneck to space is launch. The bottleneck on the ground is power.
I don’t think anyone thinks the math works right now. But as OP showed, it’s surprisingly proximate in a way SBSP is not.
1 reply →
Yeah try tell average eco joe you are planning to warm up oceans by 0.00000001% of what sun does already.
(I agree right now it probably makes sense, but decades and centuries away we probably don't want to warm up earth anymore. If anything space datacenters could provide shade for earth lol.)