Google's First Orbital Data Center Is Online — and It's Writing Poetry

Google's Suncatcher prototype space data center is now online and communicating with ground stations after launching earlier this month aboard a SpaceX Falcon 9 rocket alongside more than 100 other spacecraft.

A Single TPU in Orbit

The satellite carries a single Google Tensor Processing Unit (TPU) and generates up to 1 kW of electricity. It is intended to pave the way for future constellations of closely flying satellites performing large inference workloads in orbit. However, the TPU's power demands mean it can only operate for roughly twenty minutes at a time — complicating the already difficult business of space-to-ground communications.

Commissioning and First Contact

After a commissioning process that exercised the satellite's subsystems over about three weeks, the spacecraft was ready for work. When ground engineers sent their first prompt this week, they asked the model pre-loaded on the satellite — Google DeepMind's Gemma 4 — to write a haiku about being in space. The result:

Blue marble below

Data streams in silent grace

Starlight feeds my core. ✨

Google said it tested the model with space-themed limericks (they should have sent a poet?) and, perhaps more importantly, asked the spacecraft to report on the orbital conditions around the satellite. A spokesperson told TechCrunch that the satellite was operating nominally.

The Real Test: Radiation and Thermal Stress

The ongoing test campaign will examine how the chips perform under thermal stress and in the space radiation environment. Google has extensively tested its TPUs in particle accelerators on Earth, but it now needs to observe how radiation's real effects on hardware in orbit affect processor behavior. Radiation can cause "bit flips" — altering a stored value in computer memory — which can corrupt data without triggering typical monitoring tools.

Ground testing suggested that the impact of radiation would be minimal for inference workloads and more of a problem for training a model. Now the Google team will assess those dynamics in reality rather than in simulation.

What Comes Next

If the TPU onboard this spacecraft proves capable, the next step will be launching two more satellites equipped with advanced chips in 2027. Those spacecraft will support higher power generation and demonstrate inter-satellite optical links.

Google's Suncatcher project is largely alone among space-based data center visions in planning high-speed links between satellites to support coherent distributed workloads. SpaceX's data centers are also expected to use laser links, but those satellites will orbit too far apart to share processing effectively. Companies like Starcloud and Cowboy Space expect to perform inference loads on individual satellites instead.

Ties to SpaceX and Starship

Google and SpaceX's visions are closely linked. Alphabet, Google's parent company, holds a significant investment stake in SpaceX. To launch their orbital data centers, both companies will require Starship — the still-in-development heavy-lift rocket at the center of Elon Musk's space ambitions — to be fully and rapidly reusable.

In a peer-reviewed analysis published earlier this month, Suncatcher researchers estimated that SpaceX would need to fly Starship 1,800 times to achieve the same learning-driven cost reductions it saw with its earlier vehicles.

Suncatcher is a classic Google moonshot: the company's focus is on designing infrastructure that will be economical in the 2030s, betting that launch costs will fall far enough to make orbital inference a viable complement to terrestrial data centers.

via TechCrunch AI