
A successful in-orbit test advances Satlyt’s satellite computing software.
Google’s Suncatcher prototype is communicating with the ground, running a Google DeepMind Gemma 4 model on TPUs in space, and beginning tests that could shape future orbital AI infrastructure.

Google’s Suncatcher prototype space data center is online and communicating with the ground after launching earlier this month on a SpaceX Falcon 9 rocket. The satellite carries four Google Tensor Processing Units and is designed to help test whether future satellite networks could handle large AI inference workloads in orbit.
The first visible demo was lighthearted: engineers prompted Google DeepMind’s Gemma 4 model to write a space-themed haiku. The more important milestone is that the spacecraft is operating nominally and beginning real-world orbital tests.
The satellite generates up to 1 kW of electricity, but the TPU power demand means it can operate only one TPU for about 20 minutes at a time. That limitation adds complexity to the already difficult task of space-to-ground communications.
For readers tracking AI infrastructure, the takeaway is clear: putting compute in orbit is not just about launching chips. It also requires power management, reliable links, and carefully timed operations.
Google will use the test campaign to study how its chips perform under thermal stress and in the space radiation environment. The company has tested TPUs in particle accelerators on Earth, but the satellite provides a chance to measure real orbital effects.
Radiation can cause “bit flips,” which may corrupt data without triggering normal monitoring tools. Ground testing suggested radiation should be less disruptive for inference workloads than for model training, but Suncatcher will now test that assumption in space.
If the onboard TPUs prove capable, Google’s next step is to launch two more satellites with advanced chips in 2027. Those spacecraft are expected to support higher power generation and demonstrate inter-satellite optical links.
That architecture matters because Google’s Suncatcher vision emphasizes high-speed connections between close-flying satellites for coherent distributed workloads. TechCrunch notes that this approach differs from other space data center visions focused on individual satellites or more widely spaced systems.
Google and SpaceX’s space data center ambitions are closely linked through launch economics. Alphabet holds a significant investment stake in SpaceX, and both companies’ orbital data center visions depend on Starship becoming fully and rapidly reusable.
Suncatcher is framed as a Google moonshot aimed at infrastructure that could become economical in the 2030s. In the near term, its progress will be watched alongside the broader pressures around AI spending, infrastructure demand, and Google’s core business.

A successful in-orbit test advances Satlyt’s satellite computing software.
![Illustration of on vacation. [Dom Guzman]](https://news.crunchbase.com/wp-content/uploads/Vacation-AI.jpg)
AI has not reversed the funding pullback in travel tech startups.

Blackstone reportedly expects a high junk rating after Anthropic’s IPO.
Claude’s next growth story is drug discovery and biology research.