zlacker

>>g-mork+(OP)
> it is possible to put 500 to 1000 TW/year of AI satellites into deep space, meaningfully ascend the Kardashev scale and harness a non-trivial percentage of the Sun’s power

We currently make around 1 TW of photovoltaic cells per year, globally. The proposal here is to launch that much to space every 9 hours, complete with attached computers, continuously, from the moon.

edit: Also, this would capture a very trivial percentage of the Sun's power. A few trillionths per year.

>>gok+h4
We also shouldn't overlook the fact that the proposal entirely glosses over the implication of the alternative benefits we might realize if humanity achieved the incredible engineering and technical capacity necessary to make this version of space AI happen.

Think about it. Elon conjures up a vision of the future where we've managed to increase our solar cell manufacturing capacity by two whole orders of magnitude and have the space launch capability for all of it along with tons and tons of other stuff and the best he comes up with is...GPUs in orbit?

This is essentially the superhero gadget technology problem, where comic books and movies gloss over the the civilization changing implications of some technology the hero invents to punch bad guys harder. Don't get me wrong, the idea of orbiting data centers is kind of cool if we can pull it off. But being able to pull if off implies an ability to do a lot more interesting things. The problem is that this is both wildly overambitious and somehow incredibly myopic at the same time.

>>rainsf+RA
A lot of great inventions we now take for granted initially came with little motivation other than being able to kill each other more effectively. GPS, radar, jet engines, drones, super glue, microwaves, canned food, computers, even the internet. Contrary to the narrative of the internet being about sharing science, ARPANET was pushed by the DoD as a means of maintaining comms during nuclear war. It was then adopted by universities and research labs and started along the trajectory most are more familiar with.

The tale of computers is even more absurd. The first programmable, electric, and general-purpose digital computer was ENIAC. [1] It was built to... calculate artillery firing tables. I expect in the future that the idea of putting a bunch of solar into space to run GPUs for LLMs will probably seem, at the minimum - quaint, but that doesn't mean the story ends there.

[1] - https://en.wikipedia.org/wiki/ENIAC

>>somena+Z91
That’s not the point of the person you are replying to. They are saying if we somehow come up with the tech that makes harnessing the sun a thing, the best we can still do is put a bunch of GPUs in space? It makes no sense.

>>WD-42+ny1
It kinda does make sense if you consider that solar panels in space have been used for a very long time (to power satellites). However, getting the electricity they generate down to Earth is very complicated, so you end up having to use it in space, and one of few things that would make sense for that is indeed data centers, because getting the data to Earth is easier (and Elon already handily has a solution for that).

However I'm curious how many solar panels you would need to power a typical data center. Are we talking something like a large satellite, or rather a huge satellite with ISS-size solar arrays bolted on? Getting rid of the copious amounts of heat that data centers generate might also be a challenge (https://en.wikipedia.org/wiki/Spacecraft_thermal_control)...

>>rob74+nA1
The plan seems to be for lots and lots of smaller satellites.

For inferencing it can work well. One satellite could contain a handful of CPUs and do batch inferencing of even very large models, perhaps in the beginning at low speeds. Currently most AI workloads are interactive but I can't see that staying true for long, as things improve and they can be trusted to work independently for longer it makes more sense to just queue stuff up and not worry about exactly how high your TTFT is.

For training I don't see it today. In future maybe. But then, most AI workloads in future should be inferencing not training anyway.

>>mike_h+hE1
Latency means this still makes no sense to me. Perhaps some batch background processing job such as research or something but that's stretching.

>>KoolKa+VX2
I think the most providers all give high latency batch APIs significant discounts. A lot of AI workloads feel batch-oriented to me, or could be once they move beyond the prototype and testing phases. Chat will end up being a small fraction of load in the long term.

>>mike_h+5v5
That would imply there's still capacity here on earth for this type of traffic.