Latenode

Project Suncatcher logs Google’s first in-orbit TPU reply

An early proof that AI inference can run in extreme-edge settings, where power limits, link gaps, and hardware checks become core workflow constraints.

Project Suncatcher logs Google’s first in-orbit TPU reply

Google said on October 30, 2026 that its Project Suncatcher prototype satellite returned an AI-generated answer while in orbit, after a ground prompt to an onboard Gemma 4 model. The first output was a short poem about space, downlinked during a pass near Svalbard, Norway. In Google’s account, the company describes it as the first AI response from a data center-class TPU in orbit.

Independent reporting adds the hardware limits behind the demo. TechCrunch reported that the prototype carries four TPUs, but power constraints mean only one can run at a time, and only for about 20 minutes per session. Google said it also tested space-themed limericks and telemetry prompts, with thermal and radiation checks still ahead.

For builders, the signal is less the poem than the operating constraints: inference on remote hardware with tight power windows, intermittent links, and delayed validation.

When models and APIs shift keep the workflow running

Connect the models and apps from stories like this in one scenario. If a vendor changes routing, pricing, or availability, you update the workflow — you don't start over.

Start Free

Free forever plan. No credit card required.