SpaceX and Nvidia adapt Vera Rubin AI platform for orbital deployment
The companies are redesigning the rack-scale AI system for space use, but radiation and thermal challenges remain.
What to know
- SpaceX and Nvidia are redesigning the Vera Rubin AI platform to operate in orbit, a major engineering challenge requiring solutions for radiation and thermal management.
- The project faces skepticism about its economic viability and environmental impact, with critics noting LEO satellites have 2–3 year lifespans versus land data centers' 20+ year potential.
- Some commenters view the effort as regulatory avoidance rather than genuine innovation, citing Earth-based alternatives' superior cost-effectiveness and longevity.
The dispute Whether orbital AI data centers serve genuine technical advantages or exist primarily as a workaround to evade terrestrial environmental and regulatory constraints. · positions read across 6 posts and comments
Orbital data centers are economically unfeasible and driven by regulatory avoidance rather than genuine technical or business needs.
-
“At this point all the reason I can think of for them is to avoid regulations and negative publicity for data centers on land. I can't find any economic reasons for a space based data centers.”
ctiger12 · Reddit
Orbital deployment contradicts claims that AI data centers will create jobs, since space-based infrastructure minimizes labor needs once built.
-
“I'm a big fan of the mutually incompatible lies of "we're going to put AI datacenters in space!" and "AI datacenters will create a ton of jobs!"”
ParliamenttFnkadelic · Reddit
Orbital deployment will increase e-waste and environmental harm compared to terrestrial data centers.
-
“instead of all that e waste being processed in Taiwan or China... it will be just chucked into the ocean!”
wartornhero2 · Reddit
“SpaceX & Nvidia say they are adapting the Vera Rubin NVL72 rack-scale AI platform for use in orbit. Good luck. First, they must deal with heating and radiation.”
sjvn, Technology writer on Mastodon · Mastodon · Aug 30
SpaceX Space launch and infrastructure companyNvidia AI hardware and platform company
Elon Musk SpaceX decision-maker
The record 4 articles and posts · last 30 days
What people are saying 6 voices from 1 site · verbatim
- What are the actual technical or business advantages of orbital deployment versus land-based data centers?
- How will the industry manage e-waste and orbital debris from short-lived LEO satellites?
- Sep 2
-
Here's an economic calculator. The main costs are the satellite hardware cost $/W, and launch cost $/kg. Get those 2 variables low enough and you can beat ground based datacenters. At 7$/W satellite hardware cost and 100$/kg launch cost, it beats ground based datacenters. Getting these 2 cheap enough is the main challenge…
-
Manufacturing and launching a million satellites won't create jobs? Do they assemble themselves through magic spells?
- Aug 31
-
The people making these comments have only heard about data centers online and couldn’t tell you a single thing about them otherwise.
-
At this point all the reason I can think of for them is to avoid regulations and negative publicity for data centers on land. I can’t find any economic reasons for a space based data centers. LEO sats have a life of 2 to 3 years, and land based data centers can be at least 5 years before the hardware outdated but can work another 20 years just as…
-
They need people on the ground to build rockets and satellites and replace them every 2 years either because of orbital decay because if you want ping times it is going to need to be in LEO. Oh an technology advancements where they will need to upgrade them ever couple of years. But seriously.. AI as a jobs creator is just stupid... yeah you need…
-
I'm a big fan of the mutually incompatible lies of "we're going to put AI datacenters in space!" and "AI datacenters will create a ton of jobs!" That daily commute to LEO is going to be a bear.