Google has demonstrated that an AI model can run on a chip in orbit and send a response back to Earth. But that engineering milestone does not settle whether space-based data centers can compete with terrestrial facilities on cost. Amanda Peterson Corio, Google’s global head of data center energy, has warned that orbital data centers could cost up to 100 times more and urged exploring Earth-based options first.
15
19
What Corio’s 100× warning says—and doesn’t say
The reported figure is a warning about economics, not a claim that orbital computing is impossible. Corio’s reported advice was: “Let’s explore all the terrestrial options before we go extraterrestrial.”
19
The available reporting does not provide a cost breakdown or explain the assumptions behind the “up to 100 times” comparison. The figure should therefore be treated as a reported estimate, not a universal cost calculation for every proposed space data center.
19
23
What Project Suncatcher demonstrated
On October 30, 2026, Google said its Project Suncatcher prototype had returned its first AI response from a data-center-class TPU in orbit. An onboard Gemma 4 model generated a haiku after receiving a prompt from the ground.
6
That is a meaningful proof of operation: a prompt went up, an onboard model processed it, and a response came back. It is not evidence that a network of orbital facilities can deliver computing at commercial scale or at a competitive price.
2
5
6
The distinction that matters
The two developments address different questions. Suncatcher’s response shows that an AI workload can run on the prototype in orbit. Corio’s warning concerns whether building and operating space-based infrastructure makes economic sense compared with using options on Earth. A successful demonstration does not, by itself, resolve that comparison.
2
19
For now, the evidence supports a measured conclusion: orbital AI has passed an early technical milestone, while its costs and commercial prospects remain unproven in the available reporting. Corio’s call to exhaust terrestrial options is a reminder to distinguish what can be demonstrated from what can be justified at scale.