Forgot your password?
typodupeerror
AI Space Hardware

Nvidia Announces Vera Rubin Space-1 Chip System For Orbital AI Data Centers 147

Nvidia unveiled its Vera Rubin Space-1 system for powering AI workloads in orbital data centers. "Space computing, the final frontier, has arrived," said CEO Jensen Huang. "As we deploy satellite constellations and explore deeper into space, intelligence must live wherever data is generated." CNBC reports: In a press release, the company said that its Vera Rubin Space-1 Module, which includes the IGX Thor and Jetson Orin, will be used on space missions led by multiple companies. The chips are specifically "engineered for size-, weight- and power-constrained environments." Partners include Axiom Space, Starcloud and Planet.

Huang said Nvidia is working with partners on a new computer for orbital data centers, but there are still engineering hurdles to overcome. "In space, there's no convection, there's just radiation," Huang said during his GTC keynote, "and so we have to figure out how to cool these systems out in space, but we've got lots of great engineers working on it."
This discussion has been archived. No new comments can be posted.

Nvidia Announces Vera Rubin Space-1 Chip System For Orbital AI Data Centers

Comments Filter:
  • Methinks they are drinking their own Kool-Aid. They seem to think their chips are actually intelligent ?!!?
    Keep this guy away from anything scientific please...

  • It feels like they're making chips to fuel hype for a thing we all know can't work because physics.

    I guess there's probably a huge circular investment with SpaceX or something though?

    • Economics, perhaps, not physics.

      • by AvitarX ( 172628 ) <me@b[ ]dywinehundred.org ['ran' in gap]> on Wednesday March 18, 2026 @04:22AM (#66047212) Journal

        Even excluding launch costs do we have a feasible way to make a space data center work?

        I'm sure the physics can work with math, but I'm not convinced something could be built to maintain orbit, capture energy, and then shed the heat with unlimited budget that replaces a medium sized data center.

        • As I've said before, the heat the back of the solar panel can reject at reasonable temp (near 100C) is in the same region as the front can generate (not taking into account earth visibility though, too hard for back of envelope). So cooling seems the easier problem. Capturing energy is just a question of more light weight (flexible) panels and a polar orbit for non stop sunshine.

          LEO with never seen before surface area to weight ratios might be an issue.

          • by rossdee ( 243626 )

            Polar orbit costs more to get it there (deltaV) and it would have to be higher than LEO.

            Also what about maintenance and upgrades?

            • The answer to all these problems is simple. It follows Step 3: ??? to Step 4: Profit!!
            • by wiggles ( 30088 )

              The answer to the maintenance and upgrade question is simple - you don't maintain, you don't upgrade. You design the whole thing to be disposable. If an individual server breaks, you turn it off and pretend it doesn't exist anymore. Keep going until the thing is so broken you can't do anything more with it. When it stops working entirely, you abandon it as space junk or splash the whole orbital DC into the Indian Ocean where it can cause environmental havoc for centuries to come.

              • by Junta ( 36770 )

                Also remember, this won't be just a handful of potential deorbiting disposable nightmares, it will be tens or hundreds of thousands if they actually executed the way they want.

          • by Junta ( 36770 ) on Wednesday March 18, 2026 @01:43PM (#66048088)

            The math is that for each couple of GPU servers (we are talking GB300 sorts of servers), you'd need something about the size of the ISS' solar and cooling systems. Contrary to his "great engineers will solve it" sentiment, the cooling side of the equation is governed by very straightforward and incontrovertible physics, best possible solar isn't *much* better than what the ISS has already.

            Others in this "let's go orbital datacenters" crowd have already pegged that you'd need tens of thousands of these setups, meaning that we'd need to launch essentially tens of thousands of ISS', and probably plan to let them all burn up after 5-6 years.

            This is to compete with maybe 3 to 5 traditionally terrestial datacenters, the construction of which would probably be less than launching even a handful of those setups.

        • by nightflameauto ( 6607976 ) on Wednesday March 18, 2026 @09:14AM (#66047476)

          Even excluding launch costs do we have a feasible way to make a space data center work?

          Step 1: Place all the tech-bro AI broligarchs onto a SpaceX Starship with data terminals for each of them. Make sure live video feeds are available so we can watch them work.

          Step 2: Launch to a Lagrange point.

          Step 3: Allow the tech-bros access to their terminals, and flip on the live video feeds.

          Step 4: Enjoy watching them try to continue to hype their own farts from space while their supplies dwindle. See how long it takes until they start to realize they are totally boned.

          Step 5: Place bets on who eats who first and enjoy the show.

          Step 6: Once free of them, start to reclaim some small semblance of sanity back here on Earth. But, since we're humanity, this step is mostly optional.

      • by Junta ( 36770 )

        Yes, *technically* possible but physics is why the economics is completely stupid. Given enough materials and devoting an ISS-scale power and cooling solution per server we can work, but it's monumentally stupid because ultimately it's a massive waste of resources for a highly compromised theoretical end game versus just plopping them down in terrestrial datacenters. The physics involved in the data transmission, the power, the cooling, the launching is all very stupid, possible, but stupid given the huge

    • by HiThere ( 15173 )

      Sorry, but you don't "know it can't work because physics".
      There are good reasons to believe it's a difficult problem, and it's certainly questionable whether it's a reasonable goal. But I'd bet those chips have other uses, if "Space A!!!" doesn't take off. And I doubt that "Space AI!!" is impossible, I think it's is probably currently too expensive to be practical, however. But I'm not an engineer specializing is design of space projects, so my judgement here shouldn't be taken as "inside information".

      • It does not take a lot of physics. It takes people who have some experience with heating and cooling to know the difficulty. Cooling a chip on Earth is done by transferring the heat to air in most cases and water for some cases. Space is a vacuum which is terrible at transferring heat. In fact vacuum is a good insulator, and sandwiching a vacuum between two walls is known as a vacuum flask to keep liquids hot or cold for a long time.
        • I too, am puzzled by the claim that cooling things is "more efficient" in space. That's not my understanding at all.

          I suppose if you had radiators the size of Wyoming you could radiate enough heat away but I don't see how operating a data center in space would be a financially practical solution.

          It seems like a lot of work and expense for something that's going to be insanely expensive to service and maintain.

          Also, maybe I missed it but where, exactly, are the financial savings in this concept? How is goin

          • I too, am puzzled by the claim that cooling things is "more efficient" in space. That's not my understanding at all.

            Most people problem understand space is around 3K and computers on Earth get up to 373K. The heat differential is all they see. The method to transfer the heat on Earth is not practical in space.

            It seems like a lot of work and expense for something that's going to be insanely expensive to service and maintain.

            The proposal is trying to counter the fact that data centers on Earth are resource intensive. Moving it to space is the simple answer but it is an answer without any forethought that it is practical. It is a practical as sending my food into space to preserve it longer.

    • by tlhIngan ( 30335 )

      It feels like they're making chips to fuel hype for a thing we all know can't work because physics.

      I guess there's probably a huge circular investment with SpaceX or something though?

      Anything to make line go up. Because that's all that matters nowadays - line goes up.

      And the fact that line goes up doesn't require actually spending money or money moving around.

  • by GreyViking ( 3606993 ) on Wednesday March 18, 2026 @03:38AM (#66047182)
    >and so we have to figure out how to cool these systems out in space, but we've got lots of great engineers working on it." Why don't they just ask AI? They obviously don't have much faith in their product. ;)
  • "Space computing, the final frontier, has arrived" Too bad we don't get to see him jumping around a stage in a Star Trek uniform.

    I miss the Monkey Boy Dance from Steve Ballmer: https://www.youtube.com/watch?... [youtube.com].

    Elmo cannot hold a candle to him: https://www.youtube.com/watch?... [youtube.com].

    And we should give la Presidenta honorable....well, nothing about him is honorable....let's just leave it at mention: https://www.youtube.com/watch?... [youtube.com].

    These guys need to form a dance troupe behind Huang when he goes into orbit on hi

  • I have announced Oxygen+. Better than stale old oxygen, I've re-engineered it for the future of breathing. Better, faster, stronger, and since "old" communist deployed oxygen suddenly seems to be running low as of late you'd better take out a lifetime subscription now, as you could certainly be caught dead without Oxygen+!
  • They don't make satellites.

    Their problem is probably more making lightweight boiling refrigerant based cooling blocks which can keep the their system cool in a vacuum, including PCB, connectors, cables, etc. Dumping the heat from refrigerant to space is not their problem.

    • by boner ( 27505 )

      Have they even considered how much radiation hardening they need - even though they will still operate within the Van Allen belt, the increased radiation is a serious concern.

    • A vacuum is the perfect insulator. How do they cool in space? heat pipes are fine but you have to radiate heat somewhere and it's all in a vacuum.

      • Radiators.

        I don't know if it will be better to use heatpumps or just use more area. With heatpumps, you could probably get away using front for PV and back for radiators. Without heatpumps you'd probably have a bunch of the sun side area be mirrored, with more radiator area than PV.

        • You can ONLY radiate IR light - you can't "radiate" like you do on earth which is mostly done by convention not by giving off IR light and that takes AIR to do. you need to dump a ton of watts somewhere and light is all you've got to work with -- last I looked, my CPU/GPU isn't lighting up like a light bulb.

    • by HiThere ( 15173 )

      You DON'T want to boil off coolant in space, because resupplying it is quite difficult. Since you need radiation hardened chips anyway, what you need to do is use designs that work at a higher temperature, so cooling is a smaller problem. (Basically you're going to need to depend on radiation based cooling at the system level. The efficiency of that goes up, IIRC, as the fourth power of the difference between the emitter temperature and the background.)

      OTOH, these were probably designed to meet the specs

  • What I am not hearing is what problems this actually solves..

    Cooling - nope you can only radiate, so that is actually harder, maybe you can get the waste heat really low and turn most of it into power generation - but that is expensive.

    Power - Nobody wants to put that much radio active isotope in orbit for obvious reasons, solar works really well in space but that is a lot rather expense material and lift cost.

    Space - well there is a lot of space in space... but there is a lot of space on the ground too tha

    • What I am not hearing is what problems this actually solves..

      It keeps suckers from losing interest in propping up comically inflated stocks.

    • Datacentres in space are such an obviously bad idea that only an idiot or a grifter would even suggest them and Huang is no idiot.

    • Power - solar with no night downtime
      Launch cost - irrelevant xAI now part of SpaceX
      latency comparable or reduces compared to other AI datacenters - up to Starlink/Datacenter then down with no other connection needed
      Same setup works for moon and Mars
      • by Junta ( 36770 )

        Power: We have plenty of power now, and as was stated, the panel sizes are still huge for the power needed, even in space
        Launch cost: Doesn't matter who 'owns' it, the launches are still expensive, regardless of how you move the shell game of accounting
        Latency is still at a disadvantage compared to a boring old fiber, regardless of your fondness for Starlink
        There is *zero* reason to be concerned about a common solution for Moon and Mars at this point. Even if there were, landing on those rocks provides a m

    • It's interesting that since their GPUs are too expensive for the consumer market, they're now looking at making chip systems for space, where it will be harder to argue w/ the pricing
  • If I were Vera Rubin's relatives I would be pissed that her name is being used for something so utterly unintelligent.

    • by dargaud ( 518470 )
      Particularly since the name is already taken by a revolutionary telescope... that will now take bad images thanks to all those fucking satellites passing in front of it !
  • Is cosmic rays, they flip bits, cause resets.
      And high computation applications are especially sensitive.

    • I know that is a showstopper for "regular computing", but if AI is just predicting another word, or rendering a picture, I'm not certain a bit flip here or there matters in the scheme of things. I could be wrong, but that is my impression. I wouldn't do CRC or data checks in applications like this. a pixel in a picture may be red instead of blue, not sure most people would notice.
  • There are good reasons why satellites might need one or two powerful chips, but there are no good reasons to build orbiting data centers. It makes far more sense to build more small data centers on the planet if your goal is to avoid outages. A satellite is located where any major power can easily shoot at it, so all you've done is reduce the pool of potential bad actors to those who have the greatest motive while increasing the odds of Kessler syndrome.

    Data centers already have the problem that once you've

    • by Junta ( 36770 )

      Though think of the opportunity for nVidia, if they get everyone sold on the 'your GPU will burn up on re-entry in 5 years', then they never have to worry about a client delaying a refresh...

  • Are these stupidest fucking thing I have ever heard in my entire life. Patrick Boyle recently did a good video on YouTube explaining why the SpaceX IPO is a giant fraud (too long didn't watch, they have used up all the real launch customers except for their internet service and their internet service is running out of customers who don't already have high-speed internet and are making more than $32 a day)

    There is no air in space and air is how cooling works. So if you don't have air you have to use enorm
  • Why do people keep talking about putting something HOT in space. Outer space is NOT COLD. The main requirement for computing (after power) is cooling. Infrared radiation is the usual form of cooling, and that ain't fast. They have to know this. So either someone is Dunning Krugering all over the place, or it's another scam to get money from people who are ignorant of the reality.

  • designed for size, weight, power contraints, but no radiation hardening? Solar flares, cosmic rays, and other charged particles all unfiltered without atmospheric protection. How is an orbital data center going to cope?
  • The bubble gonna pop any second now. NVIDIA wasting resources on "space chips" when supply chains are currently so restricted is the kind of hype that happens when the bubble is hyped out.

It was kinda like stuffing the wrong card in a computer, when you're stickin' those artificial stimulants in your arm. -- Dion, noted computer scientist

Working...