xAI joins SpaceX
spacex.com660 points by g-mork 10 hours ago
660 points by g-mork 10 hours ago
> 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.
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.
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.
I think the Colossus[1] predated the ENIAC but is still in line with your general theme of doing stuff for the military. In this case it was used for cipher breaking, not firing calculations.
You could argue that it doesn't really count though because it was only turing complete in theory: "A Colossus computer was thus not a fully Turing complete machine. However, University of San Francisco professor Benjamin Wells has shown that if all ten Colossus machines made were rearranged in a specific cluster, then the entire set of computers could have simulated a universal Turing machine, and thus be Turing complete."
> You could argue that it doesn't really count though because it was only turing complete in theory
Then you have to also count the Z3 which predates the Colossus by 2 years.
Yes, but as Ron Perlman famously said in the beginning of Fallout, "War never changes".
I would be more shocked that we eliminated war than if we achieved this version of Elon's future.
It makes sense to think that we will continue to make scientific progress through war and self defense.
Reason being, nothing is more motivating than wanting to survive
I'm starting to wonder if a person like Elon with his... morals... is who we want to be creating a vision for the future.
[flagged]
Bollocks, by your standards we can't discuss the most vile people because 'nobody's perfect' but there is a huge gap between the likes of Musk and ordinary people.
Indeed, at least a $700 billion gap. One is reminded of a great Mark Twain quote, "Whereas principle is a great and noble protection against showy and degrading vanities and vices, poverty is worth six of it."
The problem is that the Venn diagram of 'vile people' and 'billionaires' has a lot of overlap so these people are doing a disproportionate amount of damage.
Sorry guy, but I don't go asking the king of the pedos for a special invite to the best pedo parties. I'm pretty certain I could do all kinds of extremely shady things and still be an order of magnitude more morally sound than Musk. Some of the allegations coming out in the last few days are pretty severe, and that's without looking into the rest of his recent history.
The fact that you're here being snide about it is pretty revealing in itself.
Not to go heads I win, tails you lose, but even if we go down this path - it's the same story because militaries are investing heavily in LLM stuff, both overtly and covertly. Outside of its obvious uses in modeling, data management, and other such things - there also seems to be a fairly widespread belief, among the powers that be, that if you just say the magic words to somebody, that you can make them believe anything. So hyper-scaling LLM potential has direct military application, same as Starlink and Starship.
The digital internet began with the telegraphy network in the early 1800s.
Many, many network protocols were developed and used.
> with the telegraphy network in the early 1800s.
Late 1700 actually, and war was indeed a key motivation for the deployment of the Télégraphe Chappe.
See "The Victorian Internet: The Remarkable Story of the Telegraph and the Nineteenth Century's On-line Pioneers"
https://www.amazon.com/dp/162040592X
Télégraphe Chappe was a semaphore system using flags. It was not an electrical telegraph, nor was it binary.
Really? That is so interesting - which ones? Any ancestors of commonly used ones today?
Off the top of my head BIX, Prodigy, Compuserve, MCIMail, BBS, Ethernet, Token Ring, $25 Network, AOL, Timeshare, Kermit, Fax
Anyone with 2+ computers immediately thought about connecting them.
The data centers in space is 100% about Golden Dome,
https://wikipedia.org/wiki/Golden_Dome_(missile_defense_syst...
Nope, it's 100% about building the stock valuation of SpaceX for an IPO in the face of significant risk from a cold war its CEO started on X with the U.S. federal government and increasing competition from Blue Origin, Quinfan and Guowang. DOD will play Bedrock vs Grok until there is feature parity and then make a decision not based on the features.
Disclaimer: Not an Elon hater, but far from a sycophant, similar to how I felt about Steve Jobs for 40+ years.
Yeah it does not make a whole lot of sense as the useful lifespan of the gpus in 4-6 years. Sooo what happens when you need to upgrade or repair?
This is a question that analysts don't even ask on earnings calls for companies with lowly earthbound datacenters full of the same GPUs.
The stock moves based on the same promise that's already unchecked without this new "in space" suffix:
We'll build datacenters using money we don't have yet, fill them with GPUs we haven't secured or even sourced, power them with infrastructure that can't be built in the promised time, and profit on their inference time over an ever-increasing (on paper) lifespan.
> This is a question that analysts don't even ask
On the contrary, data centers continue to pop up deploying thousands of GPUs specifically because the numbers work out.
The H100 launched at $30k GPU and rented for $2.50/hr. It's been 3 years since launch, the rent price is still around $2.50.
During these 3 years, it has brought in $65k in revenue.
They can run these things at 100% utilization for 3 years straight? And not burn them out? That's impressive.
Not really. GPUs are stateless so your bounded lifetime regardless of how much you use them is the lifetime of the shitties capacitor on there (essentially). Modulo a design defect or manufacturing defect, I’d expect a usable lifetime of at least 10 years, well beyond the manufacturer’s desire to support the drivers for it (ie the sw should “fail” first).
Same that happens with Starlink satellites that are obsolete or exhausted their fuel - they burn up in the atmosphere.
> the useful lifespan of the gpus in 4-6 years. Sooo what happens when you need to upgrade or repair?
Average life of starlink satellite is around 4-5 years
damn. at this point its not even about a pretense for progress, just a fetish for a very dirty space
A "fully and rapidly reusable" Starship would bring the cost of launch down orders of magnitude, perhaps to a level where it makes sense to send up satellites to repair/refuel other satellites.
With zero energy cost it will run until it stops working or runs out of fuel, which I'm guessing is between 5-7 years.
5 to 7 months given they want 100kw Per ton and magical mystery sauce shielding is going to do shit all.
> Sooo what happens when you need to upgrade or repair?
The satellite deorbits and you launch the next one.
not to mention that radiation hardening of chips has a big impact on cost and performance
You could immersion cool them and get radiation resistance as a bonus.
Yes, because launching then immersed in something that will greatly increase the launch weight will help...
So what are the other things? You said he glossed over them and didn't mention a single one.
Reliably and efficiently transport energy generated in space back to earth, for starters
Or let me guess, its going to be profitable to mine crypto in space (thereby solving the problem of transporting the "work" back to earth)
It's always better to generate electricity on the ground than attempt to beam it to the ground from space. The efficiency loss of beamed power is huge.
The efficiency loss of nighttime is approximately 100% if we’re talking about solar energy. At least at a most basic level, it’s not totally absurd to stick some kind of power beaming contraption in space where it is mostly not shadowed by the Earth and beam power to a ground station.
I concur it’s not necessarily totally absurd — but when you consider that such contraptions require large — very large! — receiving arrays to be built on the ground, it’s hard to avoid concluding that building gigantic photovoltaic arrays in, say Arizona (for the US) along with batteries for overnight buffering and transmission lines would still be massively more efficient.
Is that more or less absurd than making deals with our neighbours to share their electricity? Build some solar farms around the planet and then distribute it over wire.
I honestly don't know the answer. I know there's some efficiency loss running over long wires too but I don't know what's more realistic.
There is absolutely nothing realistic about power transmission from space to earth, wired or wireless.
We have these things called batteries, you charge them during the day, and drain them at night.
A solar+battery setup is already cheaper than a new gas plant. Beaming power from space is absolutely asinine, quite frankly. The losses are absurd, the sun already does it 24/7, and we know how to make wires and batteries to shuffle the sun's power around however we need to. Why on earth would we involve satellites?
Why would you transfer the energy to earth? The energy powers ai compute = $
Dead on, You can transmit data to and from space and have the compute completed at potentially fractions of the cost.
Tell me about your cooling medium in space
A large piece of aluminum with ammonia pumped through it?
Right up to the radiation limit and then you'll either have to throttle your precious GPUs or you'll be melting your satellite or at least the guts of it. You're looking at an absolutely massive radiator here, many times larger than the solar panels that collect the energy to begin with.
Nothing about this is sounding economically competitive with ground based solutions