TECH · FIRST LOOK
Google’s Satellite Sent Back Its First AI Answer. It Was a Haiku
Google says a Gemma 4 model on its Project Suncatcher test satellite answered a prompt from orbit with a haiku. Heat and radiation tests come next. Skeptics point to cost, heat and how hard repairs would be.

Transcript of the video edition
Google got an answer from a chip in space, and it was a poem. Is this where giant computers go next?
The test satellite went up October first. This week Google asked its Gemma four model for a space haiku. Google posted the reply: Blue marble below. Data streams in silent grace. Starlight feeds my core.
The big idea is sunlight. Data centers on Earth need huge power, and more neighbors oppose them.
Skeptics point to cost, heat and tough repairs. Reports say the chips run fifteen to twenty minutes at a time, limited by heat and power.
Next come heat and radiation tests, then twin satellites in twenty twenty-seven. So is a poem from space the future?
Sources
- Google, Oct 30, 2026: Google receives response from Project Suncatcher prototype satellite in space (launched Oct 1; Gemma 4 haiku, text as posted by Google; heat and radiation tests next; twin satellites in 2027)
- The Verge, Oct 30, 2026: Google’s satellite AI server transmits its first response to Earth (Gemma 4 haiku; test plan; twin satellites in 2027)
- NPR, Oct 1, 2026: Google launches Project Suncatcher, a step towards AI data centers in space (launch; chips to run 15 minutes at a time because of heat; cost, heat and repair doubts; many satellites needed to match a data center; growing opposition to data centers)
- Light Reading, Sept 28, 2026: Google sets test launch of ‘Suncatcher’ orbital data center (Google’s Travis Beals: chips can run about 15 minutes before shutting down to cool, as he told The New York Times)
- TechCrunch, Oct 30, 2026: Google’s first data center satellite is online and writing poetry (first prompt this week; power demands limit a chip to about 20 minutes at a time; tests next; two more satellites in 2027)
This video edition was scripted with automated tools from the sources listed, narrated by a synthetic voice and published by Daniel Varrone. Corrections: editor@monochromechronicles.com