How to play: Some comments in this thread were written by AI. Read through and click flag as AI on any comment you think is fake. When you're done, hit reveal at the bottom to see your score.got it
So like I don't get to use my degree except to tutor kids and answer physics questions and the like, so indulge me for a second because this doesn't make much sense to me. So like the endgame is to create a new Pryor, OK facility in space, right? It doesn't make sense if we're not shipping something that's about the same as what you can build on Earth.
Epoch.AI says that this facility has 100,000 TPU chips, eats 370 megawatts of electricity, but designed to get rid of up to 480 megawatts of heat. Sun power is 1.36 kW/m^2, which is something like 7 megawatts per football field (assuming ~5000 m^2), but you have to put up something like 3-5x that because solar panels are only 20-30% efficient, right? Assume that you can arrange it so that the solar panels are one side, the radiators on the other, you can maybe get away with only 300-400 football fields maybe? So like 1.5 million m^2 or 1.2 km wide, right? At LEO, 650 km, you have (1.2km)/(650 km) * 180 degrees/pi is 0.11 degrees or 6.6 arcminutes of visual size. The moon is only like 30 arcminutes. And the paper talks about how they're going to not do one big monolithic construction but an oval of fridge-sized objects separated out 2-3 times this size -- so like I don't see how you get another Pryor, OK size data center in LEO without basically having it look like a second, smaller moon flying across the sky 10 times per day.
And if this were a wildly successful idea are we talking about having like 5-10 of them, a few for each big frontier lab? This just sounds like we're talking about the most profound shift to our night skies since we started having to deal with light pollution in our cities.
Not really. Warming comes from greenhouse gases trapping heat, so more energy supply doesn't fix it. The pitch here is just power for compute. In a dawn-dusk sun-synchronous orbit the panels see nearly constant sunlight, no night and no clouds, so you get several times the output per panel. The chips still dump all that energy as heat, though.
Retail cost per watt for terrestrial panel: under 45c
Manufacturing cost per watt of space grade solar panel: up to $450
Annual performance degradation of terrestrial solar panel: under 0.5%
That of space-grade solar panel: up to 2%
Life span of terrestrial panel: about 2x that of space panel.
Total difference in cost per watt feeding a DC load in space vs on land: about x400
And that's ignoring launch costs. It makes absolutely zero sense. And given the scale of production and investment in manufacturing, terrestrial is likely to stretch even further ahead in the cost stakes.
Best cost to orbit I am seeing is $1500/kg. A GPU rack is ~1500kg. Let’s imagine you can take a terrestrial data center rack, no scaffolding, solar panels, radiators, radios, propellant, or propulsion. Fly it into orbit, kick it out the airlock, and let it work through magic. That’s $2.2 million to get into position.
Industrial power rates are cheap, say $.10/kwh, but pretend you sign terrible deals, and it costs you $.30/kwh to run and cool a terrestrial GPU rack. A 150kw unit will then be (150x24x365x.3) =$394k/year.
You can operate the terrestrial version for 5.7 years before the two hit parity.
Yes for now; Alphabet is optimistic costs will come down.
They're projecting the learning curve, that the more you do it the cheaper it gets, continues arbitrarily far.
However, they recon it will take SpaceX launching 370,000 tons to LEO to make the costs come down enough to be worth it: https://arxiv.org/pdf/2511.19468
Even my bull case put that 10 years off, which is so far away it lacks relevance just because tech moves so much faster than that timescale; my bear case says that's about 45 years off.
The cubesat price is mostly a small-lot rideshare premium, so $/kg isn't comparable. Falcon 9 on a full manifest is nearer $2-3k/kg, as far as I know.
As I recall, Google's own paper only closes the loop by assuming launch falls to roughly $200/kg by the mid-2030s. That's a Starship-scale bet, not something anyone has demonstrated.
> Manufacturing cost per watt of space grade solar panel: up to $450
Where did this number come from, and why is it literally 1000x more than the Earth version?
I can totally believe the cost of the ISS panels was in that ballpark, but if we're talking about putting megawatts of these things in orbit, economics of scale will very quickly come into play.
The idea is still a real stretch, but according to Google's paper in Joule [1] they're putting satellites in low Earth orbit and only targeting a service lifetime of 5 years. Over that short time, with the correspondingly modest radiation exposure, you can use inexpensive silicon solar cells like you would use for terrestrial solar farms. The expensive space grade solar cells from e.g. Boeing Spectrolab are more resistant to radiation, and achieve higher conversion efficiencies, but silicon cells are fine for satellites like these.
Starlink satellites already use silicon solar cells, since they too are cost sensitive and don't have long lifespans:
On the Starlink V2 mini satellite, we predict that approximately 5% of the mass of the entire satellite could survive reentry. The biggest contributor (~90% of the surviving mass) is silicon from the solar cells, which has a high melting point...
Can I just compare this to buying some slightly remote land, building one small nuclear reactor on it and offering free electricity to everyone in a wide radius of the plant/datacenter?
Someone else said the facility needs 370MW. Great, because SMR designs provide 470MW. That’s 100MW excess, given for free to locals to entice them to support the project. That can power 40k homes, easily. People object to these projects partly because their bills go up? Cool, those bills disappear now.
But no, I guess we’re creating massive arrays in space and polluting the night sky forever.
Smells like a cover story for military AI module validation testing.
The AI industry spends crazy money to reduce latency and increase density. The act of placing resources in space, to be used by people on earth, only increases latency, and reduces density. The performance metrics are ops/power/space/$, not G ratings, unless its being validated for flight.
Terrestrial data centers are having a hard time connecting to power. What’s the good of a $40/Watt compute asset you can’t turn on? Spending $10 or even $20/Watt extra to get power by sending inference load to orbit starts to sound like a bargain, and that’s doable with today’s launch vehicles. With high reuse, that number drops to below $1/Watt for 24/7 power, which makes it cheaper than any terrestrial energy source.
And the vast majority of compute load is inference. Part of the training process benefits from a large, coherent (ie low latency) data center, but that’s a minority of the load nowadays.
Also: low Earth orbit is already pretty low latency. A lot of people’s intuitions about space latency are from when satellite internet was based out of distant geosynchronous satellites, but even that is probably just fine for most inference workloads.
What would a radar constellation look like that could see everything on earth to a metre or so resolution? Thousands and thousands of satellites all with big phased array radios and solar panels and batteries.
What do you need to distill a 24/7 stream of everything on Earth in roughly 1m^3 voxels into data like humans and vehicles? Bearing in mind the data torrent is gigantic and can't always be reliably downlinked: co-orbital AI.
Maybe. But it’s also Google, a company famous for just spending money on random crap. Data centers are politically toxic right now, and Elon Musk’s bloviating made it trendy to discuss AI in space. Maybe they think that following this trend with a relatively simple launch will raise the stock.
I think it’s somewhat impressive they got it into orbit quickly since surely a year ago this wasn’t an earnest idea, and I think it’ll be interesting to see the affects of space on high end chips and connectivity fabric.
But also I agree that this just doesn’t make sense from a performance standpoint.
NIMBY numbers rising on recent data center buildouts.
People who think AI is going to destroy humanity.
People experiencing real financial hardship due to AI job displacements.
If google is going to issue press releases, which sounds better ?
"Google putting data centers in space, not your backyard"
or
"Google testing TPU chips for doomsday skynet laser weapons"
The boys at the pentagon procurement department can read between the lines.
> Smells like a cover story for military AI module validation testing.
I was trying to rack my head for any possible way this made sense and was coming up empty. Normally i mistrust conspiracy theories like this, but its the only explanation that seems remotely plausible.
I've been in the tech industry for awhile, and it's all I can think of as well.
After reading others comments, I'm trying to imagine the utility of having AI training, inference, intelligence filtering, etc in space, like, spy satellites all sharing an AI hub, and that doesn't make sense either. They can just as easily download spy satellite images, and apply "as much AI as they can print money to buy" to it, and not be limited.
As a for profit company, google has a fiduciary obligation to not piss away money on projects that have no foreseeable commercial value ( unless fee contracted by an external entity to do so ).
I'm sticking with "flight validation for AI super weapons" as the only revenue stream I can foresee.
It ticks both the "potential revenue stream" and "legit R&D tax benefits" boxes.
I disagree on latency. Training doesn't care about an extra 30ms, it's batch work that runs for weeks. What they're testing is whether TPUs survive the radiation and whether the optical links hold up. That's a boring, checkable question, no cover story needed.
I would be extremely unsurprised if some amount of the 'starshield' customized starlink satellites recently launched for the DoD have at least some modest amount of general purpose hypervisor/server capacity on them, possibly with limited GPU-like capabilities for certain software. At least as a small scale test.
One of the "holy grail" things the military has wanted for a long time is to be able to treat a satellite not just like a bent pipe repeater but to do active signal processing and data processing such as for communication between disparate air and ground, naval units.
There's one other thing: it's essentially like the cloud but for DCs. i.e. need more compute? send up another rocket. 'limitless' space. expensive but a relatively fixed one-time cost (vs earth-based DCs which are the bare metal equivalent: fast, affordable, but pets not cattle).
All of that said, I'm skeptical we'll see this in a big way any time soon.
When you add in the VRT and the QML (even not including the ZB-PRP 5) you get far lower than what a naive USP calculation gets you. Horner Weisenbach talked about this recently on their blog.
It's not really extraterritorial from the point of view of needing to comply with regulations though. Your datacentre company has registered offices and physical locations. Your downlinks are based on territory. You need to relaunch on a regular basis and you have very restricted choice of territories that you can launch from.
Everyone is focusing on the feasibility of putting compute in space, but I'm more concerned about the unwiseness of putting AIs in a place where they might be reasonably able to deny us access.
Should human-AI relations sour, I want the reassurance that humans with sticks and rocks could reasonably hope to disassemble a data center. Place said data center in orbit around the sun, and the incentive for a hostile intelligence to isolate itself by, e.g. initiating a Kessler cascade, must be considered.
More material for the eventual investor fraud/scam lawsuits.
They did not even attempt to solve the main issues: energy and heat management. The GPUs can work only for 15 minutes and are cooled by a phase change material that then slowly sheds the heat.
When did they do this? I didn’t hear of any SpaceX launches besides the falcon launch yesterday for a government spy satellite. I wonder if this one was also bundled in.
It's Arianespace, no second n. And IIRC that ~25 figure was the commercial market they could realistically compete for, not total launches worldwide. Starlink flights alone blow past that, so three in a day is unusual but not crazy anymore.
If US folks are watching a rocket launch and they're wearing light jackets, it's Vandenberg (which doesn't publish all their scheduled rocket launches.)
Cool, we will finally have actual data to see if that’s possible or not. My understanding is that doing compute in space for ML is pretty absurd, but the discussion will be way more interesting with real numbers.
I still think it’s mostly a way to put pressure on politicians, « you either give us tax cut, fast tracked permits, and let us do what we want with the land or we go to space and you get no benefits »
Doing compute in space for ML is perfectly possible (i.e. https://en.wikipedia.org/wiki/Phi-Sat-2), making it competitive with terrestrial compute is a much harder question to solve.
The point with a moonshot isn't always instantaneous viability. Some things take time to prove out and become economically justified.
If we are genuinely serious about carbon emissions, etc., this is a compelling path. Being able to deploy solar with >90% capacity factor is basically like having zero risk nuclear power. This possibility exists nowhere on the surface of the planet. Not even close. Terminator following SSO can approach 100% availability at the expense of increased reaction mass.
AI might be the first workload where the consumed energy per unit of mass is so high that we might actually break even in ways that the accountants care about. This is where the comparisons to terrestrial grid break down hard. There is one specific load, it is extraordinarily dense, and can be dispatched with a high degree of granularity.
The heat rejection arguments fall under the umbrella of lack of imagination for me. There are some very exotic schemes that the HN hive mind is not aware of or willing to consider. A single monolithic radiator is clearly not sufficient. We need a scheme where the radiator can be distributed across a massive planar or spherical region of space.
This is obviously for the Golden Dome. Missile interception for ICBMs only works in space and they need the hardware and AI for classifiers (decoy/not_decoy) and for targeting in space, too.
For civilian purposes this would be even more dystopian. Communication satellites and propaganda from space have always existed, but AI manufactured mass slop and propaganda from space really raises philosophical objections.
These endless conversations about how it can never work are so boring. The only difference between scenarios where it will and won't work are (often unstated) assumptions about launch cost. Launch cost is the whole ball game, as any problem can be solved with more mass. So the whole discussion can be summarized as "I think Starship will fail to meet its launch cost targets" vs "I think Starship will work". It's fashionable to bet against Elon these days but historically it has been a bad bet on long timeframes...
If launch costs are assumed to be $0/kg, how does that solve the waste heat problem? Add more mass? What about the maintenance of failing hardware?
I'm genuinely curious, I thought basically heat radiation was a hard physical boundary, regardless of launch cost.
Everyone's arguing about panel cost per watt, but in our experience the panels are never what sinks you. It's the swap-a-dead-board part. On the ground a failed GPU is a 20 minute ticket. Up there it's just dead weight, so the whole thing has to be dirt cheap per kg.
Epoch.AI says that this facility has 100,000 TPU chips, eats 370 megawatts of electricity, but designed to get rid of up to 480 megawatts of heat. Sun power is 1.36 kW/m^2, which is something like 7 megawatts per football field (assuming ~5000 m^2), but you have to put up something like 3-5x that because solar panels are only 20-30% efficient, right? Assume that you can arrange it so that the solar panels are one side, the radiators on the other, you can maybe get away with only 300-400 football fields maybe? So like 1.5 million m^2 or 1.2 km wide, right? At LEO, 650 km, you have (1.2km)/(650 km) * 180 degrees/pi is 0.11 degrees or 6.6 arcminutes of visual size. The moon is only like 30 arcminutes. And the paper talks about how they're going to not do one big monolithic construction but an oval of fridge-sized objects separated out 2-3 times this size -- so like I don't see how you get another Pryor, OK size data center in LEO without basically having it look like a second, smaller moon flying across the sky 10 times per day.
And if this were a wildly successful idea are we talking about having like 5-10 of them, a few for each big frontier lab? This just sounds like we're talking about the most profound shift to our night skies since we started having to deal with light pollution in our cities.