How would AI data centers in space even work? A former NASA robotics chief explains
Nobody is talking about where the next 10 exaflops of compute lives. Spoiler: not on Earth.
Why it matters
As AI infrastructure demands push against terrestrial power and cooling limits, space-based data centers are moving from sci-fi to serious engineering discussion—with real implications for where the next generation of AI training happens.
The key facts
9 to knowSpace-based power generation + heat rejection model eliminates Earth-bound thermal/electrical constraints
Only data transmission returns to Earth, reducing infrastructure footprint
Former NASA robotics expert lends credibility to feasibility discussion
Addresses fundamental AI capex bottleneck: power and cooling, not compute itself
Shifts long-term AI infrastructure strategy conversation beyond terrestrial data center buildout
Space data centers could eliminate heat rejection problem by radiating directly into space
Only data payload would return to Earth, reducing transmission constraints
Former NASA robotics chief consulted on feasibility
Addresses fundamental infrastructure bottleneck: power density and thermal management on terrestrial grid
Go to the source
ZDNet AIzdnet.com
Publisher excerpt: 'You collect electricity in space, and you eject heat in space. The only thing that comes to Earth is data.'