Google’s Suncatcher space data center prototype is online and communicating with the ground after launching earlier this month alongside more than 100 other spacecraft on a SpaceX Falcon 9 rocket.
The satellite, which carries four Google Tensor Processing Units (TPUs) and generates up to 1 kW of electricity, is intended to pave the way for future networks of short-range flying satellites performing large inference workloads in orbit. However, the TPU’s power requirements mean it can only power one up for about 20 minutes at a time, complicating the already delicate job of space-to-ground communications.
After a commissioning process in which the satellite’s subsystems were tested for approximately three weeks, the spacecraft was ready for operation. When engineers on the ground sent their first prompt this week, they asked the model preloaded on the satellite, Google DeepMind’s Gemma 4, to write a haiku about being in space. And so:
Blue marble below
Data streams in silent grace
Starlight feeds my core. ✨
Google said it tested the model with these space-themed limericks (should they have sent a poet?) and, perhaps more importantly, the company asked the spacecraft to report on orbital conditions around the satellite.
A spokesperson told TechCrunch that the satellite was working nominally.
The ongoing test campaign will examine the performance of the chips under thermal stress and in the space radiation environment. Google has extensively tested its TPUs in particle accelerators on Earth, but now needs to see how the real-world effects of radiation on its hardware in space affect its processor. Radiation can cause “bit flips” – causing a value to be stored in the computer’s memory – and this can in turn corrupt data without triggering conventional monitoring tools.
Field tests suggested that the impact of radiation would be minimal for inference workloads and would pose more of a challenge for training a model, but the Google team will now be able to evaluate these dynamics in reality, not in a simulation.
If the TPUs on board this spacecraft prove capable, the next step will be to launch two more satellites equipped with advanced chips in 2027. These spacecraft will support higher power output and demonstrate inter-satellite optical links.
Google’s Project Suncatcher is largely alone among space data center visions planning high-speed links between satellites to support consistent distributed workloads; SpaceX’s data centers are also expected to use laser links, but also orbit too far apart to effectively share processing. Companies like Starcloud and Cowboy Space expect to perform inference payloads on individual satellites.
The visions of Google and SpaceX are closely linked. Alphabet, Google’s parent company, has a significant stake in SpaceX. And to launch their orbital data centers, both companies will need Starship, the still-in-development heavy-duty rocket at the center of Elon Musk’s space vision, to be fully and quickly 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-related cost reductions as with its previous vehicles.
Suncatcher is a classic Google moonshot: the company is focused on designing infrastructure that will be economical in the 2030s. However, getting there may depend on how the rest of the company responds to near-term AI pressures and opportunities. Alphabet, Google’s parent company, saw its stock fall this week after reporting that efforts to insert AI into its search business were hurting advertising revenue growth.
Fixed: The Suncatcher satellite prototype carries four TPUs, but can only operate one at a time.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
Gn tech
