News for hackers. Stuff that matters.
Re:Data Centers in Space by stephenhandley on Tuesday May 13, 2025 @07:04PM (#43978783)
I'm wondering why this is being proposed to be in space rather than at the lunar poles? Is it a latency question? I'd assume if this is being used for LLM training purposes and it is a relatively long training run, latency is less a pressing concern than bandwidth constraints before / after training?

Wouldn't the moon would provide better environment for debris shielding and regolith could be used to passively dump heat?

I'm way out of my depth here, so of course I asked ChatGPT 4o to come up with a three paragraph concept sketch for an LLM training facility based at one of the lunar poles:

<gpt>

A lunar pole is an ideal location for large-scale LLM training due to its unique advantages in thermal management and power availability. Near-continuous sunlight at the “peaks of eternal light” allows for stable solar power generation, while nearby permanently shadowed craters provide ultra-cold regions that can act as natural heat sinks — solving one of the biggest challenges of orbital compute: radiative cooling. Compute clusters could be housed in regolith-insulated modules, with heat piped into crater floors or dissipated through passive radiators embedded in the lunar terrain.

The architecture would be modular: prefabricated training units delivered by landers (like Starship) could be arranged into scalable compute farms. Power would be provided by solar arrays on crater rims, with supercapacitors or RTGs as backup. Data transfer wouldn’t require low latency, just high throughput, so high-gain antennas or laser links could periodically sync model weights back to Earth. With no atmosphere, lunar surface conditions are harsh but predictable, and maintenance could be handled robotically.

Compared to the estimated $480 trillion needed for space-based radiators to support orbital training, a lunar facility could achieve similar scale at orders of magnitude lower cost, leveraging the Moon’s surface as both a structural anchor and a thermal reservoir. As launch costs drop and lunar infrastructure matures, this becomes a compelling direction for off-planet AI training.

</gpt>

Re:Data Centers in Space by jadbox on Tuesday May 13, 2025 @07:11PM (#43978855)
I guess it would be heat pollution at the poles that may contribute to global climate change if we start building immense data centers there.
Re:Data Centers in Space by stephenhandley on Tuesday May 13, 2025 @07:28PM (#43979021)
I believe they'd mainly need to be strategic about where and how deep they were radiating into the surface since there wouldn't be atmosphere that would trap heat.
Re:Data Centers in Space by Etheryte on Tuesday May 13, 2025 @07:17PM (#43978916)
Please don't litter HN with LLM generated slop. Each and every one of us here is more than capable of doing that themselves if they please. The value of HN is the human discussion.
Re:Data Centers in Space by stephenhandley on Tuesday May 13, 2025 @07:21PM (#43978959)
Lol you mind giving me a quick heads up on what have you contributed other than this meta gripe? Any concrete response to the actual questions I'm asking or just a soapbox?
Re:Data Centers in Space by mplanchard on Tuesday May 13, 2025 @07:45PM (#43979138)
I thought your question was interesting. The ChatGPT quote was not. If I want to know what an LLM “thinks”, I can ask it. I have more respect for and am more interested in reading people’s thoughts than machine output.

I do appreciate your at least labeling at such, so I could skip over it.

Re:Data Centers in Space by stephenhandley on Tuesday May 13, 2025 @08:43PM (#43979514)
If you're under the impression that all the content on here is 100% human, that's probably slightly optimistic, and the reason why I labeled it explicitly rather than regurgitating / rewording. That said, I'd rather read GPT output than skim the meta-discussion that ruins so many HN threads. Totally stoked that we're now contributing to that rather than talking about the actual content.
Re:Data Centers in Space by aspenmayer on Tuesday May 13, 2025 @09:06PM (#43979641)
Labeling isn’t the problem. Generated comments are against HN guidelines.

https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

> That said, I'd rather read GPT output than skim the meta-discussion that ruins so many HN threads. Totally stoked that we're now contributing to that rather than talking about the actual content.

You brought this meta-discussion on yourself and upon the rest of the community, knowing the likely result beforehand. It’s hard to say you’re not part of the problem you rail against when you create the initial start conditions for the thread.

Re:Data Centers in Space by stephenhandley on Tuesday May 13, 2025 @09:43PM (#43979844)
Honestly was not aware of that guideline. Insinuating that I was someone trolling for this discussion is a stretch.
Re:Data Centers in Space by aspenmayer on Tuesday May 13, 2025 @09:59PM (#43979933)
> Honestly was not aware of that guideline. Insinuating that I was someone trolling for this discussion is a stretch.

I’m not insinuating anything, quite the opposite. You have experience on the site, and so are aware of how often threads can devolve into and revolve around meta-discussions, and one of those common meta-discussions on HN is in response to generated comments and AI slop. I’m not saying you’re trolling by doing it intentionally, and in fact I’ve appealed to dang to add the provision to the guidelines proper myself in email and on HN. But surely we both agree that to continue to do so now that you know would be a flagrant violation of the guidelines?

> That said, I'd rather read GPT output than skim the meta-discussion that ruins so many HN threads.

I try to make the kinds of posts I want to read. Perhaps you might find this guiding precept helpful also, as I do. As generated comments are against the guidelines, your own words will suffice.