The Pulse
Google’s Suncatcher Satellite Sends Its First AI Response to Earth
Google’s Project Suncatcher prototype satellite ran a Gemma 4 prompt on an orbital TPU and returned a haiku to Earth. The company says next tests will examine how the chips perform under thermal stress and radiation, ahead of a planned twin
AI.info Team · Written with AI · fact-checked ·
Google’s Project Suncatcher satellite has sent an AI-generated response from orbit to Earth, but its first success was a haiku—not proof that the company can run dependable AI computing in space. The prototype used an onboard TPU to run Google’s Gemma 4 model, then returned the poem during a pass over Svalbard, Norway, the company said on October 30.
A haiku marks the first orbital reply
Engineers sent the prompt during a ground-station pass, asking Gemma 4 to write about being in space. The satellite returned: “Blue marble below / Data streams in silent grace / Starlight feeds my core.” Google called the result the first AI response from a data-center-class TPU in orbit.
“This week, we reached our "Hello World" milestone: the first-ever AI response from a data center class TPU in orbit.”
Travis Beals, Senior Director, Paradigms of Intelligence, Google
The team also used the satellite to generate space-themed limericks and answer prompts about the payload’s technical status.
Weeks of checks preceded the prompt
Google launched the prototype on October 1 aboard SpaceX’s Transporter-18 rideshare mission, in partnership with Planet. The first weeks in orbit went to commissioning: powering up the host and TPU trays, then running preliminary machine-learning tests. The successful prompt-and-response test came after those checks.
Project Suncatcher is a research effort to explore whether space could eventually host scalable machine-learning infrastructure. Google’s October announcement described the launched satellite as a first step in studying how its processors handle launch stresses and the radiation and temperature extremes of orbit. One working prototype does not demonstrate that a larger orbital computing system can operate reliably or economically.
Radiation and heat remain open tests
The successful response establishes that the onboard hardware completed this workload and returned its output. Google says its next tests will examine how TPUs perform under thermal stress and radiation exposure in orbit. Those conditions matter because processors must keep working while managing heat in a vacuum and facing radiation that can affect electronics.
Before launch, Google reported laboratory radiation tests on its Trillium TPUs, saying they withstood a dose greater than what the chips would receive during a five-year space mission. The company also described testing a cooling setup in a thermal-vacuum chamber. Those preparations offer a basis for the in-orbit experiments, but they are not a substitute for measuring the hardware in space.
Google targets twin satellites for 2027
Google identifies twin interconnected satellites in 2027 as the next milestone. The company’s earlier account of the project says future spacecraft would need high-bandwidth laser links to coordinate and move data between satellites. Testing the link between two spacecraft would address a different challenge from running a prompt on one satellite.
For now, the concrete result is narrow: a Gemma 4 prompt ran on an orbital TPU, and its haiku made it back to Earth. Google’s next announced work is to measure processor performance under radiation and thermal stress before moving on to the paired-satellite test.