1 hr ago
Google Prepares First Orbital Test of AI Chips
Google wants to see whether powerful computer chips can work in space.
Its first test satellite is expected to launch on a SpaceX rocket on or around October 1.
The small satellite will carry four Google chips called Tensor Processing Units.
These chips will try to run simple artificial-intelligence tasks.
Scientists will check whether radiation, heat, cold and rocket shaking damage them.
They will also test how to cool the chips because ordinary fans do not work in space.
The satellite is an experiment, not a full space data center.
Google says it may launch more satellites in 2027.
Experts say building useful data centers in space could still take many years because launches and repairs are expensive and difficult.
Google plans to launch the MVP prototype satellite on a SpaceX rideshare mission, with reports identifying October 1 as the launch date.
The satellite, built with Planet Labs, will carry four Google Tensor Processing Units to run AI workloads in low Earth orbit.
Project Suncatcher will test the chips’ resistance to radiation, extreme temperatures, launch forces and cooling constraints.
Google says the mission is an early data-gathering test, not an operational orbital data center, and expects it to operate for about a year.
Google plans additional satellite tests in 2027, while experts say launch costs, maintenance, cooling and production challenges could delay commercial viability.
- Who
- Google, working with Planet Labs and SpaceX, is conducting the test.
- What
- A prototype satellite carrying four Google Tensor Processing Units will test AI computing in orbit.
- Where
- The satellite is expected to launch from Vandenberg Space Force Base in California and operate in low Earth orbit.
- When
- The launch is planned for next week; one report specifies October 1, and further satellite tests are planned for 2027.
- Why
- Google is investigating whether solar-powered, space-based computing could help support future AI workloads while reducing some terrestrial electricity and cooling constraints.
Potential benefits
Engineering concerns
Why put AI computing in space?
Potential benefits
Google says satellites could use near-continuous sunlight and potentially reduce dependence on some terrestrial electricity and cooling infrastructure.
Engineering concerns
Experts say high launch costs, difficult maintenance and satellite-production bottlenecks may prevent space-based computing from becoming commercially competitive soon.
Readiness of the technology
Potential benefits
The mission will provide real-world data on whether Google’s AI chips can run reliably in orbit and could support later satellite demonstrations.
Engineering concerns
Google describes the mission as an early experiment and says it does not expect anything usefully operational in the next few years.
Technical feasibility
Potential benefits
The satellite has passed vibration testing, and Google is testing heat pipes and radiators to cool chips in the vacuum of space.
Engineering concerns
Radiation can cause data errors, conventional airflow cooling is unavailable, and the chips may reportedly need to shut down after about 15 minutes to cool.
Key facts
- Project
- Project Suncatcher
- Prototype satellite
- MVP
- Payload
- Four Google Tensor Processing Units
- Launch vehicle
- SpaceX Falcon 9 on a rideshare mission
- Launch site
- Vandenberg Space Force Base, California
- Expected mission duration
- About one year
- Future testing
- Two additional satellites are planned for 2027









