Project Suncatcher is a long-term research effort to investigate whether AI computing can be powered and operated in space. Its prototype satellite launched on October 1, and after completing initial commissioning - including powering up its host systems and TPU trays - the team began testing machine-learning workloads in orbit. The first major milestone was a successful “Hello World” exchange: during a ground-station pass, researchers sent a prompt to the onboard Gemma 4 open model and received an AI-generated haiku back as the satellite passed over Svalbard, Norway. The response demonstrated that a data-center-class TPU had run an AI model and returned its output from orbit.
Researchers followed the haiku test with space-themed limericks and prompts asking the model to report payload telemetry. These early results show that the hardware can execute basic AI tasks in space, but they do not yet establish how reliably it will perform over time or under demanding orbital conditions. The next phase will test TPU performance amid thermal stress and radiation exposure. If the chips prove dependable, that would support the project’s next planned milestone: launching two interconnected satellites in 2027. The successful response is an initial technical demonstration, not a completed space-based computing system, and the broader research goal remains a work in progress.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.